fs/ext4/extents.c

   1 /*
   2  * Copyright (c) 2003-2006, Cluster File Systems, Inc, info@clusterfs.com
   3  * Written by Alex Tomas <alex@clusterfs.com>
   4  *
   5  * Architecture independence:
   6  *   Copyright (c) 2005, Bull S.A.
   7  *   Written by Pierre Peiffer <pierre.peiffer@bull.net>
   8  *
   9  * This program is free software; you can redistribute it and/or modify
  10  * it under the terms of the GNU General Public License version 2 as
  11  * published by the Free Software Foundation.
  12  *
  13  * This program is distributed in the hope that it will be useful,
  14  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  15  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  16  * GNU General Public License for more details.
  17  *
  18  * You should have received a copy of the GNU General Public Licens
  19  * along with this program; if not, write to the Free Software
  20  * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-
  21  */
  22
  23 /*
  24  * Extents support for EXT4
  25  *
  26  * TODO:
  27  *   - ext4*_error() should be used in some situations
  28  *   - analyze all BUG()/BUG_ON(), use -EIO where appropriate
  29  *   - smart tree reduction
  30  */
  31
  32 #include <linux/module.h>
  33 #include <linux/fs.h>
  34 #include <linux/time.h>
  35 #include <linux/jbd2.h>
  36 #include <linux/highuid.h>
  37 #include <linux/pagemap.h>
  38 #include <linux/quotaops.h>
  39 #include <linux/string.h>
  40 #include <linux/slab.h>
  41 #include <linux/falloc.h>
  42 #include <asm/uaccess.h>
  43 #include <linux/fiemap.h>
  44 #include "ext4_jbd2.h"
  45 #include "ext4_extents.h"
  46
  47
  48 /*
  49  * ext_pblock:
  50  * combine low and high parts of physical block number into ext4_fsblk_t
  51  */
  52 ext4_fsblk_t ext_pblock(struct ext4_extent *ex)
  53 {
  54         ext4_fsblk_t block;
  55
  56         block = le32_to_cpu(ex->ee_start_lo);
  57         block |= ((ext4_fsblk_t) le16_to_cpu(ex->ee_start_hi) << 31) << 1;
  58         return block;
  59 }
  60
  61 /*
  62  * idx_pblock:
  63  * combine low and high parts of a leaf physical block number into ext4_fsblk_t
  64  */
  65 ext4_fsblk_t idx_pblock(struct ext4_extent_idx *ix)
  66 {
  67         ext4_fsblk_t block;
  68
  69         block = le32_to_cpu(ix->ei_leaf_lo);
  70         block |= ((ext4_fsblk_t) le16_to_cpu(ix->ei_leaf_hi) << 31) << 1;
  71         return block;
  72 }
  73
  74 /*
  75  * ext4_ext_store_pblock:
  76  * stores a large physical block number into an extent struct,
  77  * breaking it into parts
  78  */
  79 void ext4_ext_store_pblock(struct ext4_extent *ex, ext4_fsblk_t pb)
  80 {
  81         ex->ee_start_lo = cpu_to_le32((unsigned long) (pb & 0xffffffff));
  82         ex->ee_start_hi = cpu_to_le16((unsigned long) ((pb >> 31) >> 1) & 0xffff);
  83 }
  84
  85 /*
  86  * ext4_idx_store_pblock:
  87  * stores a large physical block number into an index struct,
  88  * breaking it into parts
  89  */
  90 static void ext4_idx_store_pblock(struct ext4_extent_idx *ix, ext4_fsblk_t pb)
  91 {
  92         ix->ei_leaf_lo = cpu_to_le32((unsigned long) (pb & 0xffffffff));
  93         ix->ei_leaf_hi = cpu_to_le16((unsigned long) ((pb >> 31) >> 1) & 0xffff);
  94 }
  95
  96 static int ext4_ext_truncate_extend_restart(handle_t *handle,
  97                                             struct inode *inode,
  98                                             int needed)
  99 {
 100         int err;
 101
 102         if (!ext4_handle_valid(handle))
 103                 return 0;
 104         if (handle->h_buffer_credits > needed)
 105                 return 0;
 106         err = ext4_journal_extend(handle, needed);
 107         if (err <= 0)
 108                 return err;
 109         err = ext4_truncate_restart_trans(handle, inode, needed);
 110         if (err == 0)
 111                 err = -EAGAIN;
 112
 113         return err;
 114 }
 115
 116 /*
 117  * could return:
 118  *  - EROFS
 119  *  - ENOMEM
 120  */
 121 static int ext4_ext_get_access(handle_t *handle, struct inode *inode,
 122                                 struct ext4_ext_path *path)
 123 {
 124         if (path->p_bh) {
 125                 /* path points to block */
 126                 return ext4_journal_get_write_access(handle, path->p_bh);
 127         }
 128         /* path points to leaf/index in inode body */
 129         /* we use in-core data, no need to protect them */
 130         return 0;
 131 }
 132
 133 /*
 134  * could return:
 135  *  - EROFS
 136  *  - ENOMEM
 137  *  - EIO
 138  */
 139 static int ext4_ext_dirty(handle_t *handle, struct inode *inode,
 140                                 struct ext4_ext_path *path)
 141 {
 142         int err;
 143         if (path->p_bh) {
 144                 /* path points to block */
 145                 err = ext4_handle_dirty_metadata(handle, inode, path->p_bh);
 146         } else {
 147                 /* path points to leaf/index in inode body */
 148                 err = ext4_mark_inode_dirty(handle, inode);
 149         }
 150         return err;
 151 }
 152
 153 static ext4_fsblk_t ext4_ext_find_goal(struct inode *inode,
 154                               struct ext4_ext_path *path,
 155                               ext4_lblk_t block)
 156 {
 157         struct ext4_inode_info *ei = EXT4_I(inode);
 158         ext4_fsblk_t bg_start;
 159         ext4_fsblk_t last_block;
 160         ext4_grpblk_t colour;
 161         ext4_group_t block_group;
 162         int flex_size = ext4_flex_bg_size(EXT4_SB(inode->i_sb));
 163         int depth;
 164
 165         if (path) {
 166                 struct ext4_extent *ex;
 167                 depth = path->p_depth;
 168
 169                 /* try to predict block placement */
 170                 ex = path[depth].p_ext;
 171                 if (ex)
 172                         return ext_pblock(ex)+(block-le32_to_cpu(ex->ee_block));
 173
 174                 /* it looks like index is empty;
 175                  * try to find starting block from index itself */
 176                 if (path[depth].p_bh)
 177                         return path[depth].p_bh->b_blocknr;
 178         }
 179
 180         /* OK. use inode's group */
 181         block_group = ei->i_block_group;
 182         if (flex_size >= EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME) {
 183                 /*
 184                  * If there are at least EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME
 185                  * block groups per flexgroup, reserve the first block
 186                  * group for directories and special files.  Regular
 187                  * files will start at the second block group.  This
 188                  * tends to speed up directory access and improves
 189                  * fsck times.
 190                  */
 191                 block_group &= ~(flex_size-1);
 192                 if (S_ISREG(inode->i_mode))
 193                         block_group++;
 194         }
 195         bg_start = ext4_group_first_block_no(inode->i_sb, block_group);
 196         last_block = ext4_blocks_count(EXT4_SB(inode->i_sb)->s_es) - 1;
 197
 198         /*
 199          * If we are doing delayed allocation, we don't need take
 200          * colour into account.
 201          */
 202         if (test_opt(inode->i_sb, DELALLOC))
 203                 return bg_start;
 204
 205         if (bg_start + EXT4_BLOCKS_PER_GROUP(inode->i_sb) <= last_block)
 206                 colour = (current->pid % 16) *
 207                         (EXT4_BLOCKS_PER_GROUP(inode->i_sb) / 16);
 208         else
 209                 colour = (current->pid % 16) * ((last_block - bg_start) / 16);
 210         return bg_start + colour + block;
 211 }
 212
 213 /*
 214  * Allocation for a meta data block
 215  */
 216 static ext4_fsblk_t
 217 ext4_ext_new_meta_block(handle_t *handle, struct inode *inode,
 218                         struct ext4_ext_path *path,
 219                         struct ext4_extent *ex, int *err)
 220 {
 221         ext4_fsblk_t goal, newblock;
 222
 223         goal = ext4_ext_find_goal(inode, path, le32_to_cpu(ex->ee_block));
 224         newblock = ext4_new_meta_blocks(handle, inode, goal, NULL, err);
 225         return newblock;
 226 }
 227
 228 static inline int ext4_ext_space_block(struct inode *inode, int check)
 229 {
 230         int size;
 231
 232         size = (inode->i_sb->s_blocksize - sizeof(struct ext4_extent_header))
 233                         / sizeof(struct ext4_extent);
 234         if (!check) {
 235 #ifdef AGGRESSIVE_TEST
 236                 if (size > 6)
 237                         size = 6;
 238 #endif
 239         }
 240         return size;
 241 }
 242
 243 static inline int ext4_ext_space_block_idx(struct inode *inode, int check)
 244 {
 245         int size;
 246
 247         size = (inode->i_sb->s_blocksize - sizeof(struct ext4_extent_header))
 248                         / sizeof(struct ext4_extent_idx);
 249         if (!check) {
 250 #ifdef AGGRESSIVE_TEST
 251                 if (size > 5)
 252                         size = 5;
 253 #endif
 254         }
 255         return size;
 256 }
 257
 258 static inline int ext4_ext_space_root(struct inode *inode, int check)
 259 {
 260         int size;
 261
 262         size = sizeof(EXT4_I(inode)->i_data);
 263         size -= sizeof(struct ext4_extent_header);
 264         size /= sizeof(struct ext4_extent);
 265         if (!check) {
 266 #ifdef AGGRESSIVE_TEST
 267                 if (size > 3)
 268                         size = 3;
 269 #endif
 270         }
 271         return size;
 272 }
 273
 274 static inline int ext4_ext_space_root_idx(struct inode *inode, int check)
 275 {
 276         int size;
 277
 278         size = sizeof(EXT4_I(inode)->i_data);
 279         size -= sizeof(struct ext4_extent_header);
 280         size /= sizeof(struct ext4_extent_idx);
 281         if (!check) {
 282 #ifdef AGGRESSIVE_TEST
 283                 if (size > 4)
 284                         size = 4;
 285 #endif
 286         }
 287         return size;
 288 }
 289
 290 /*
 291  * Calculate the number of metadata blocks needed
 292  * to allocate @blocks
 293  * Worse case is one block per extent
 294  */
 295 int ext4_ext_calc_metadata_amount(struct inode *inode, sector_t lblock)
 296 {
 297         struct ext4_inode_info *ei = EXT4_I(inode);
 298         int idxs, num = 0;
 299
 300         idxs = ((inode->i_sb->s_blocksize - sizeof(struct ext4_extent_header))
 301                 / sizeof(struct ext4_extent_idx));
 302
 303         /*
 304          * If the new delayed allocation block is contiguous with the
 305          * previous da block, it can share index blocks with the
 306          * previous block, so we only need to allocate a new index
 307          * block every idxs leaf blocks.  At ldxs**2 blocks, we need
 308          * an additional index block, and at ldxs**3 blocks, yet
 309          * another index blocks.
 310          */
 311         if (ei->i_da_metadata_calc_len &&
 312             ei->i_da_metadata_calc_last_lblock+1 == lblock) {
 313                 if ((ei->i_da_metadata_calc_len % idxs) == 0)
 314                         num++;
 315                 if ((ei->i_da_metadata_calc_len % (idxs*idxs)) == 0)
 316                         num++;
 317                 if ((ei->i_da_metadata_calc_len % (idxs*idxs*idxs)) == 0) {
 318                         num++;
 319                         ei->i_da_metadata_calc_len = 0;
 320                 } else
 321                         ei->i_da_metadata_calc_len++;
 322                 ei->i_da_metadata_calc_last_lblock++;
 323                 return num;
 324         }
 325
 326         /*
 327          * In the worst case we need a new set of index blocks at
 328          * every level of the inode's extent tree.
 329          */
 330         ei->i_da_metadata_calc_len = 1;
 331         ei->i_da_metadata_calc_last_lblock = lblock;
 332         return ext_depth(inode) + 1;
 333 }
 334
 335 static int
 336 ext4_ext_max_entries(struct inode *inode, int depth)
 337 {
 338         int max;
 339
 340         if (depth == ext_depth(inode)) {
 341                 if (depth == 0)
 342                         max = ext4_ext_space_root(inode, 1);
 343                 else
 344                         max = ext4_ext_space_root_idx(inode, 1);
 345         } else {
 346                 if (depth == 0)
 347                         max = ext4_ext_space_block(inode, 1);
 348                 else
 349                         max = ext4_ext_space_block_idx(inode, 1);
 350         }
 351
 352         return max;
 353 }
 354
 355 static int ext4_valid_extent(struct inode *inode, struct ext4_extent *ext)
 356 {
 357         ext4_fsblk_t block = ext_pblock(ext);
 358         int len = ext4_ext_get_actual_len(ext);
 359
 360         return ext4_data_block_valid(EXT4_SB(inode->i_sb), block, len);
 361 }
 362
 363 static int ext4_valid_extent_idx(struct inode *inode,
 364                                 struct ext4_extent_idx *ext_idx)
 365 {
 366         ext4_fsblk_t block = idx_pblock(ext_idx);
 367
 368         return ext4_data_block_valid(EXT4_SB(inode->i_sb), block, 1);
 369 }
 370
 371 static int ext4_valid_extent_entries(struct inode *inode,
 372                                 struct ext4_extent_header *eh,
 373                                 int depth)
 374 {
 375         struct ext4_extent *ext;
 376         struct ext4_extent_idx *ext_idx;
 377         unsigned short entries;
 378         if (eh->eh_entries == 0)
 379                 return 1;
 380
 381         entries = le16_to_cpu(eh->eh_entries);
 382
 383         if (depth == 0) {
 384                 /* leaf entries */
 385                 ext = EXT_FIRST_EXTENT(eh);
 386                 while (entries) {
 387                         if (!ext4_valid_extent(inode, ext))
 388                                 return 0;
 389                         ext++;
 390                         entries--;
 391                 }
 392         } else {
 393                 ext_idx = EXT_FIRST_INDEX(eh);
 394                 while (entries) {
 395                         if (!ext4_valid_extent_idx(inode, ext_idx))
 396                                 return 0;
 397                         ext_idx++;
 398                         entries--;
 399                 }
 400         }
 401         return 1;
 402 }
 403
 404 static int __ext4_ext_check(const char *function, unsigned int line,
 405                             struct inode *inode, struct ext4_extent_header *eh,
 406                             int depth)
 407 {
 408         const char *error_msg;
 409         int max = 0;
 410
 411         if (unlikely(eh->eh_magic != EXT4_EXT_MAGIC)) {
 412                 error_msg = "invalid magic";
 413                 goto corrupted;
 414         }
 415         if (unlikely(le16_to_cpu(eh->eh_depth) != depth)) {
 416                 error_msg = "unexpected eh_depth";
 417                 goto corrupted;
 418         }
 419         if (unlikely(eh->eh_max == 0)) {
 420                 error_msg = "invalid eh_max";
 421                 goto corrupted;
 422         }
 423         max = ext4_ext_max_entries(inode, depth);
 424         if (unlikely(le16_to_cpu(eh->eh_max) > max)) {
 425                 error_msg = "too large eh_max";
 426                 goto corrupted;
 427         }
 428         if (unlikely(le16_to_cpu(eh->eh_entries) > le16_to_cpu(eh->eh_max))) {
 429                 error_msg = "invalid eh_entries";
 430                 goto corrupted;
 431         }
 432         if (!ext4_valid_extent_entries(inode, eh, depth)) {
 433                 error_msg = "invalid extent entries";
 434                 goto corrupted;
 435         }
 436         return 0;
 437
 438 corrupted:
 439         ext4_error_inode(inode, function, line, 0,
 440                         "bad header/extent: %s - magic %x, "
 441                         "entries %u, max %u(%u), depth %u(%u)",
 442                         error_msg, le16_to_cpu(eh->eh_magic),
 443                         le16_to_cpu(eh->eh_entries), le16_to_cpu(eh->eh_max),
 444                         max, le16_to_cpu(eh->eh_depth), depth);
 445
 446         return -EIO;
 447 }
 448
 449 #define ext4_ext_check(inode, eh, depth)        \
 450         __ext4_ext_check(__func__, __LINE__, inode, eh, depth)
 451
 452 int ext4_ext_check_inode(struct inode *inode)
 453 {
 454         return ext4_ext_check(inode, ext_inode_hdr(inode), ext_depth(inode));
 455 }
 456
 457 #ifdef EXT_DEBUG
 458 static void ext4_ext_show_path(struct inode *inode, struct ext4_ext_path *path)
 459 {
 460         int k, l = path->p_depth;
 461
 462         ext_debug("path:");
 463         for (k = 0; k <= l; k++, path++) {
 464                 if (path->p_idx) {
 465                   ext_debug("  %d->%llu", le32_to_cpu(path->p_idx->ei_block),
 466                             idx_pblock(path->p_idx));
 467                 } else if (path->p_ext) {
 468                         ext_debug("  %d:[%d]%d:%llu ",
 469                                   le32_to_cpu(path->p_ext->ee_block),
 470                                   ext4_ext_is_uninitialized(path->p_ext),
 471                                   ext4_ext_get_actual_len(path->p_ext),
 472                                   ext_pblock(path->p_ext));
 473                 } else
 474                         ext_debug("  []");
 475         }
 476         ext_debug("\n");
 477 }
 478
 479 static void ext4_ext_show_leaf(struct inode *inode, struct ext4_ext_path *path)
 480 {
 481         int depth = ext_depth(inode);
 482         struct ext4_extent_header *eh;
 483         struct ext4_extent *ex;
 484         int i;
 485
 486         if (!path)
 487                 return;
 488
 489         eh = path[depth].p_hdr;
 490         ex = EXT_FIRST_EXTENT(eh);
 491
 492         ext_debug("Displaying leaf extents for inode %lu\n", inode->i_ino);
 493
 494         for (i = 0; i < le16_to_cpu(eh->eh_entries); i++, ex++) {
 495                 ext_debug("%d:[%d]%d:%llu ", le32_to_cpu(ex->ee_block),
 496                           ext4_ext_is_uninitialized(ex),
 497                           ext4_ext_get_actual_len(ex), ext_pblock(ex));
 498         }
 499         ext_debug("\n");
 500 }
 501 #else
 502 #define ext4_ext_show_path(inode, path)
 503 #define ext4_ext_show_leaf(inode, path)
 504 #endif
 505
 506 void ext4_ext_drop_refs(struct ext4_ext_path *path)
 507 {
 508         int depth = path->p_depth;
 509         int i;
 510
 511         for (i = 0; i <= depth; i++, path++)
 512                 if (path->p_bh) {
 513                         brelse(path->p_bh);
 514                         path->p_bh = NULL;
 515                 }
 516 }
 517
 518 /*
 519  * ext4_ext_binsearch_idx:
 520  * binary search for the closest index of the given block
 521  * the header must be checked before calling this
 522  */
 523 static void
 524 ext4_ext_binsearch_idx(struct inode *inode,
 525                         struct ext4_ext_path *path, ext4_lblk_t block)
 526 {
 527         struct ext4_extent_header *eh = path->p_hdr;
 528         struct ext4_extent_idx *r, *l, *m;
 529
 530
 531         ext_debug("binsearch for %u(idx):  ", block);
 532
 533         l = EXT_FIRST_INDEX(eh) + 1;
 534         r = EXT_LAST_INDEX(eh);
 535         while (l <= r) {
 536                 m = l + (r - l) / 2;
 537                 if (block < le32_to_cpu(m->ei_block))
 538                         r = m - 1;
 539                 else
 540                         l = m + 1;
 541                 ext_debug("%p(%u):%p(%u):%p(%u) ", l, le32_to_cpu(l->ei_block),
 542                                 m, le32_to_cpu(m->ei_block),
 543                                 r, le32_to_cpu(r->ei_block));
 544         }
 545
 546         path->p_idx = l - 1;
 547         ext_debug("  -> %d->%lld ", le32_to_cpu(path->p_idx->ei_block),
 548                   idx_pblock(path->p_idx));
 549
 550 #ifdef CHECK_BINSEARCH
 551         {
 552                 struct ext4_extent_idx *chix, *ix;
 553                 int k;
 554
 555                 chix = ix = EXT_FIRST_INDEX(eh);
 556                 for (k = 0; k < le16_to_cpu(eh->eh_entries); k++, ix++) {
 557                   if (k != 0 &&
 558                       le32_to_cpu(ix->ei_block) <= le32_to_cpu(ix[-1].ei_block)) {
 559                                 printk(KERN_DEBUG "k=%d, ix=0x%p, "
 560                                        "first=0x%p\n", k,
 561                                        ix, EXT_FIRST_INDEX(eh));
 562                                 printk(KERN_DEBUG "%u <= %u\n",
 563                                        le32_to_cpu(ix->ei_block),
 564                                        le32_to_cpu(ix[-1].ei_block));
 565                         }
 566                         BUG_ON(k && le32_to_cpu(ix->ei_block)
 567                                            <= le32_to_cpu(ix[-1].ei_block));
 568                         if (block < le32_to_cpu(ix->ei_block))
 569                                 break;
 570                         chix = ix;
 571                 }
 572                 BUG_ON(chix != path->p_idx);
 573         }
 574 #endif
 575
 576 }
 577
 578 /*
 579  * ext4_ext_binsearch:
 580  * binary search for closest extent of the given block
 581  * the header must be checked before calling this
 582  */
 583 static void
 584 ext4_ext_binsearch(struct inode *inode,
 585                 struct ext4_ext_path *path, ext4_lblk_t block)
 586 {
 587         struct ext4_extent_header *eh = path->p_hdr;
 588         struct ext4_extent *r, *l, *m;
 589
 590         if (eh->eh_entries == 0) {
 591                 /*
 592                  * this leaf is empty:
 593                  * we get such a leaf in split/add case
 594                  */
 595                 return;
 596         }
 597
 598         ext_debug("binsearch for %u:  ", block);
 599
 600         l = EXT_FIRST_EXTENT(eh) + 1;
 601         r = EXT_LAST_EXTENT(eh);
 602
 603         while (l <= r) {
 604                 m = l + (r - l) / 2;
 605                 if (block < le32_to_cpu(m->ee_block))
 606                         r = m - 1;
 607                 else
 608                         l = m + 1;
 609                 ext_debug("%p(%u):%p(%u):%p(%u) ", l, le32_to_cpu(l->ee_block),
 610                                 m, le32_to_cpu(m->ee_block),
 611                                 r, le32_to_cpu(r->ee_block));
 612         }
 613
 614         path->p_ext = l - 1;
 615         ext_debug("  -> %d:%llu:[%d]%d ",
 616                         le32_to_cpu(path->p_ext->ee_block),
 617                         ext_pblock(path->p_ext),
 618                         ext4_ext_is_uninitialized(path->p_ext),
 619                         ext4_ext_get_actual_len(path->p_ext));
 620
 621 #ifdef CHECK_BINSEARCH
 622         {
 623                 struct ext4_extent *chex, *ex;
 624                 int k;
 625
 626                 chex = ex = EXT_FIRST_EXTENT(eh);
 627                 for (k = 0; k < le16_to_cpu(eh->eh_entries); k++, ex++) {
 628                         BUG_ON(k && le32_to_cpu(ex->ee_block)
 629                                           <= le32_to_cpu(ex[-1].ee_block));
 630                         if (block < le32_to_cpu(ex->ee_block))
 631                                 break;
 632                         chex = ex;
 633                 }
 634                 BUG_ON(chex != path->p_ext);
 635         }
 636 #endif
 637
 638 }
 639
 640 int ext4_ext_tree_init(handle_t *handle, struct inode *inode)
 641 {
 642         struct ext4_extent_header *eh;
 643
 644         eh = ext_inode_hdr(inode);
 645         eh->eh_depth = 0;
 646         eh->eh_entries = 0;
 647         eh->eh_magic = EXT4_EXT_MAGIC;
 648         eh->eh_max = cpu_to_le16(ext4_ext_space_root(inode, 0));
 649         ext4_mark_inode_dirty(handle, inode);
 650         ext4_ext_invalidate_cache(inode);
 651         return 0;
 652 }
 653
 654 struct ext4_ext_path *
 655 ext4_ext_find_extent(struct inode *inode, ext4_lblk_t block,
 656                                         struct ext4_ext_path *path)
 657 {
 658         struct ext4_extent_header *eh;
 659         struct buffer_head *bh;
 660         short int depth, i, ppos = 0, alloc = 0;
 661
 662         eh = ext_inode_hdr(inode);
 663         depth = ext_depth(inode);
 664
 665         /* account possible depth increase */
 666         if (!path) {
 667                 path = kzalloc(sizeof(struct ext4_ext_path) * (depth + 2),
 668                                 GFP_NOFS);
 669                 if (!path)
 670                         return ERR_PTR(-ENOMEM);
 671                 alloc = 1;
 672         }
 673         path[0].p_hdr = eh;
 674         path[0].p_bh = NULL;
 675
 676         i = depth;
 677         /* walk through the tree */
 678         while (i) {
 679                 int need_to_validate = 0;
 680
 681                 ext_debug("depth %d: num %d, max %d\n",
 682                           ppos, le16_to_cpu(eh->eh_entries), le16_to_cpu(eh->eh_max));
 683
 684                 ext4_ext_binsearch_idx(inode, path + ppos, block);
 685                 path[ppos].p_block = idx_pblock(path[ppos].p_idx);
 686                 path[ppos].p_depth = i;
 687                 path[ppos].p_ext = NULL;
 688
 689                 bh = sb_getblk(inode->i_sb, path[ppos].p_block);
 690                 if (unlikely(!bh))
 691                         goto err;
 692                 if (!bh_uptodate_or_lock(bh)) {
 693                         if (bh_submit_read(bh) < 0) {
 694                                 put_bh(bh);
 695                                 goto err;
 696                         }
 697                         /* validate the extent entries */
 698                         need_to_validate = 1;
 699                 }
 700                 eh = ext_block_hdr(bh);
 701                 ppos++;
 702                 if (unlikely(ppos > depth)) {
 703                         put_bh(bh);
 704                         EXT4_ERROR_INODE(inode,
 705                                          "ppos %d > depth %d", ppos, depth);
 706                         goto err;
 707                 }
 708                 path[ppos].p_bh = bh;
 709                 path[ppos].p_hdr = eh;
 710                 i--;
 711
 712                 if (need_to_validate && ext4_ext_check(inode, eh, i))
 713                         goto err;
 714         }
 715
 716         path[ppos].p_depth = i;
 717         path[ppos].p_ext = NULL;
 718         path[ppos].p_idx = NULL;
 719
 720         /* find extent */
 721         ext4_ext_binsearch(inode, path + ppos, block);
 722         /* if not an empty leaf */
 723         if (path[ppos].p_ext)
 724                 path[ppos].p_block = ext_pblock(path[ppos].p_ext);
 725
 726         ext4_ext_show_path(inode, path);
 727
 728         return path;
 729
 730 err:
 731         ext4_ext_drop_refs(path);
 732         if (alloc)
 733                 kfree(path);
 734         return ERR_PTR(-EIO);
 735 }
 736
 737 /*
 738  * ext4_ext_insert_index:
 739  * insert new index [@logical;@ptr] into the block at @curp;
 740  * check where to insert: before @curp or after @curp
 741  */
 742 int ext4_ext_insert_index(handle_t *handle, struct inode *inode,
 743                                 struct ext4_ext_path *curp,
 744                                 int logical, ext4_fsblk_t ptr)
 745 {
 746         struct ext4_extent_idx *ix;
 747         int len, err;
 748
 749         err = ext4_ext_get_access(handle, inode, curp);
 750         if (err)
 751                 return err;
 752
 753         if (unlikely(logical == le32_to_cpu(curp->p_idx->ei_block))) {
 754                 EXT4_ERROR_INODE(inode,
 755                                  "logical %d == ei_block %d!",
 756                                  logical, le32_to_cpu(curp->p_idx->ei_block));
 757                 return -EIO;
 758         }
 759         len = EXT_MAX_INDEX(curp->p_hdr) - curp->p_idx;
 760         if (logical > le32_to_cpu(curp->p_idx->ei_block)) {
 761                 /* insert after */
 762                 if (curp->p_idx != EXT_LAST_INDEX(curp->p_hdr)) {
 763                         len = (len - 1) * sizeof(struct ext4_extent_idx);
 764                         len = len < 0 ? 0 : len;
 765                         ext_debug("insert new index %d after: %llu. "
 766                                         "move %d from 0x%p to 0x%p\n",
 767                                         logical, ptr, len,
 768                                         (curp->p_idx + 1), (curp->p_idx + 2));
 769                         memmove(curp->p_idx + 2, curp->p_idx + 1, len);
 770                 }
 771                 ix = curp->p_idx + 1;
 772         } else {
 773                 /* insert before */
 774                 len = len * sizeof(struct ext4_extent_idx);
 775                 len = len < 0 ? 0 : len;
 776                 ext_debug("insert new index %d before: %llu. "
 777                                 "move %d from 0x%p to 0x%p\n",
 778                                 logical, ptr, len,
 779                                 curp->p_idx, (curp->p_idx + 1));
 780                 memmove(curp->p_idx + 1, curp->p_idx, len);
 781                 ix = curp->p_idx;
 782         }
 783
 784         ix->ei_block = cpu_to_le32(logical);
 785         ext4_idx_store_pblock(ix, ptr);
 786         le16_add_cpu(&curp->p_hdr->eh_entries, 1);
 787
 788         if (unlikely(le16_to_cpu(curp->p_hdr->eh_entries)
 789                              > le16_to_cpu(curp->p_hdr->eh_max))) {
 790                 EXT4_ERROR_INODE(inode,
 791                                  "logical %d == ei_block %d!",
 792                                  logical, le32_to_cpu(curp->p_idx->ei_block));
 793                 return -EIO;
 794         }
 795         if (unlikely(ix > EXT_LAST_INDEX(curp->p_hdr))) {
 796                 EXT4_ERROR_INODE(inode, "ix > EXT_LAST_INDEX!");
 797                 return -EIO;
 798         }
 799
 800         err = ext4_ext_dirty(handle, inode, curp);
 801         ext4_std_error(inode->i_sb, err);
 802
 803         return err;
 804 }
 805
 806 /*
 807  * ext4_ext_split:
 808  * inserts new subtree into the path, using free index entry
 809  * at depth @at:
 810  * - allocates all needed blocks (new leaf and all intermediate index blocks)
 811  * - makes decision where to split
 812  * - moves remaining extents and index entries (right to the split point)
 813  *   into the newly allocated blocks
 814  * - initializes subtree
 815  */
 816 static int ext4_ext_split(handle_t *handle, struct inode *inode,
 817                                 struct ext4_ext_path *path,
 818                                 struct ext4_extent *newext, int at)
 819 {
 820         struct buffer_head *bh = NULL;
 821         int depth = ext_depth(inode);
 822         struct ext4_extent_header *neh;
 823         struct ext4_extent_idx *fidx;
 824         struct ext4_extent *ex;
 825         int i = at, k, m, a;
 826         ext4_fsblk_t newblock, oldblock;
 827         __le32 border;
 828         ext4_fsblk_t *ablocks = NULL; /* array of allocated blocks */
 829         int err = 0;
 830
 831         /* make decision: where to split? */
 832         /* FIXME: now decision is simplest: at current extent */
 833
 834         /* if current leaf will be split, then we should use
 835          * border from split point */
 836         if (unlikely(path[depth].p_ext > EXT_MAX_EXTENT(path[depth].p_hdr))) {
 837                 EXT4_ERROR_INODE(inode, "p_ext > EXT_MAX_EXTENT!");
 838                 return -EIO;
 839         }
 840         if (path[depth].p_ext != EXT_MAX_EXTENT(path[depth].p_hdr)) {
 841                 border = path[depth].p_ext[1].ee_block;
 842                 ext_debug("leaf will be split."
 843                                 " next leaf starts at %d\n",
 844                                   le32_to_cpu(border));
 845         } else {
 846                 border = newext->ee_block;
 847                 ext_debug("leaf will be added."
 848                                 " next leaf starts at %d\n",
 849                                 le32_to_cpu(border));
 850         }
 851
 852         /*
 853          * If error occurs, then we break processing
 854          * and mark filesystem read-only. index won't
 855          * be inserted and tree will be in consistent
 856          * state. Next mount will repair buffers too.
 857          */
 858
 859         /*
 860          * Get array to track all allocated blocks.
 861          * We need this to handle errors and free blocks
 862          * upon them.
 863          */
 864         ablocks = kzalloc(sizeof(ext4_fsblk_t) * depth, GFP_NOFS);
 865         if (!ablocks)
 866                 return -ENOMEM;
 867
 868         /* allocate all needed blocks */
 869         ext_debug("allocate %d blocks for indexes/leaf\n", depth - at);
 870         for (a = 0; a < depth - at; a++) {
 871                 newblock = ext4_ext_new_meta_block(handle, inode, path,
 872                                                    newext, &err);
 873                 if (newblock == 0)
 874                         goto cleanup;
 875                 ablocks[a] = newblock;
 876         }
 877
 878         /* initialize new leaf */
 879         newblock = ablocks[--a];
 880         if (unlikely(newblock == 0)) {
 881                 EXT4_ERROR_INODE(inode, "newblock == 0!");
 882                 err = -EIO;
 883                 goto cleanup;
 884         }
 885         bh = sb_getblk(inode->i_sb, newblock);
 886         if (!bh) {
 887                 err = -EIO;
 888                 goto cleanup;
 889         }
 890         lock_buffer(bh);
 891
 892         err = ext4_journal_get_create_access(handle, bh);
 893         if (err)
 894                 goto cleanup;
 895
 896         neh = ext_block_hdr(bh);
 897         neh->eh_entries = 0;
 898         neh->eh_max = cpu_to_le16(ext4_ext_space_block(inode, 0));
 899         neh->eh_magic = EXT4_EXT_MAGIC;
 900         neh->eh_depth = 0;
 901         ex = EXT_FIRST_EXTENT(neh);
 902
 903         /* move remainder of path[depth] to the new leaf */
 904         if (unlikely(path[depth].p_hdr->eh_entries !=
 905                      path[depth].p_hdr->eh_max)) {
 906                 EXT4_ERROR_INODE(inode, "eh_entries %d != eh_max %d!",
 907                                  path[depth].p_hdr->eh_entries,
 908                                  path[depth].p_hdr->eh_max);
 909                 err = -EIO;
 910                 goto cleanup;
 911         }
 912         /* start copy from next extent */
 913         /* TODO: we could do it by single memmove */
 914         m = 0;
 915         path[depth].p_ext++;
 916         while (path[depth].p_ext <=
 917                         EXT_MAX_EXTENT(path[depth].p_hdr)) {
 918                 ext_debug("move %d:%llu:[%d]%d in new leaf %llu\n",
 919                                 le32_to_cpu(path[depth].p_ext->ee_block),
 920                                 ext_pblock(path[depth].p_ext),
 921                                 ext4_ext_is_uninitialized(path[depth].p_ext),
 922                                 ext4_ext_get_actual_len(path[depth].p_ext),
 923                                 newblock);
 924                 /*memmove(ex++, path[depth].p_ext++,
 925                                 sizeof(struct ext4_extent));
 926                 neh->eh_entries++;*/
 927                 path[depth].p_ext++;
 928                 m++;
 929         }
 930         if (m) {
 931                 memmove(ex, path[depth].p_ext-m, sizeof(struct ext4_extent)*m);
 932                 le16_add_cpu(&neh->eh_entries, m);
 933         }
 934
 935         set_buffer_uptodate(bh);
 936         unlock_buffer(bh);
 937
 938         err = ext4_handle_dirty_metadata(handle, inode, bh);
 939         if (err)
 940                 goto cleanup;
 941         brelse(bh);
 942         bh = NULL;
 943
 944         /* correct old leaf */
 945         if (m) {
 946                 err = ext4_ext_get_access(handle, inode, path + depth);
 947                 if (err)
 948                         goto cleanup;
 949                 le16_add_cpu(&path[depth].p_hdr->eh_entries, -m);
 950                 err = ext4_ext_dirty(handle, inode, path + depth);
 951                 if (err)
 952                         goto cleanup;
 953
 954         }
 955
 956         /* create intermediate indexes */
 957         k = depth - at - 1;
 958         if (unlikely(k < 0)) {
 959                 EXT4_ERROR_INODE(inode, "k %d < 0!", k);
 960                 err = -EIO;
 961                 goto cleanup;
 962         }
 963         if (k)
 964                 ext_debug("create %d intermediate indices\n", k);
 965         /* insert new index into current index block */
 966         /* current depth stored in i var */
 967         i = depth - 1;
 968         while (k--) {
 969                 oldblock = newblock;
 970                 newblock = ablocks[--a];
 971                 bh = sb_getblk(inode->i_sb, newblock);
 972                 if (!bh) {
 973                         err = -EIO;
 974                         goto cleanup;
 975                 }
 976                 lock_buffer(bh);
 977
 978                 err = ext4_journal_get_create_access(handle, bh);
 979                 if (err)
 980                         goto cleanup;
 981
 982                 neh = ext_block_hdr(bh);
 983                 neh->eh_entries = cpu_to_le16(1);
 984                 neh->eh_magic = EXT4_EXT_MAGIC;
 985                 neh->eh_max = cpu_to_le16(ext4_ext_space_block_idx(inode, 0));
 986                 neh->eh_depth = cpu_to_le16(depth - i);
 987                 fidx = EXT_FIRST_INDEX(neh);
 988                 fidx->ei_block = border;
 989                 ext4_idx_store_pblock(fidx, oldblock);
 990
 991                 ext_debug("int.index at %d (block %llu): %u -> %llu\n",
 992                                 i, newblock, le32_to_cpu(border), oldblock);
 993                 /* copy indexes */
 994                 m = 0;
 995                 path[i].p_idx++;
 996
 997                 ext_debug("cur 0x%p, last 0x%p\n", path[i].p_idx,
 998                                 EXT_MAX_INDEX(path[i].p_hdr));
 999                 if (unlikely(EXT_MAX_INDEX(path[i].p_hdr) !=
1000                                         EXT_LAST_INDEX(path[i].p_hdr))) {
1001                         EXT4_ERROR_INODE(inode,
1002                                          "EXT_MAX_INDEX != EXT_LAST_INDEX ee_block %d!",
1003                                          le32_to_cpu(path[i].p_ext->ee_block));
1004                         err = -EIO;
1005                         goto cleanup;
1006                 }
1007                 while (path[i].p_idx <= EXT_MAX_INDEX(path[i].p_hdr)) {
1008                         ext_debug("%d: move %d:%llu in new index %llu\n", i,
1009                                         le32_to_cpu(path[i].p_idx->ei_block),
1010                                         idx_pblock(path[i].p_idx),
1011                                         newblock);
1012                         /*memmove(++fidx, path[i].p_idx++,
1013                                         sizeof(struct ext4_extent_idx));
1014                         neh->eh_entries++;
1015                         BUG_ON(neh->eh_entries > neh->eh_max);*/
1016                         path[i].p_idx++;
1017                         m++;
1018                 }
1019                 if (m) {
1020                         memmove(++fidx, path[i].p_idx - m,
1021                                 sizeof(struct ext4_extent_idx) * m);
1022                         le16_add_cpu(&neh->eh_entries, m);
1023                 }
1024                 set_buffer_uptodate(bh);
1025                 unlock_buffer(bh);
1026
1027                 err = ext4_handle_dirty_metadata(handle, inode, bh);
1028                 if (err)
1029                         goto cleanup;
1030                 brelse(bh);
1031                 bh = NULL;
1032
1033                 /* correct old index */
1034                 if (m) {
1035                         err = ext4_ext_get_access(handle, inode, path + i);
1036                         if (err)
1037                                 goto cleanup;
1038                         le16_add_cpu(&path[i].p_hdr->eh_entries, -m);
1039                         err = ext4_ext_dirty(handle, inode, path + i);
1040                         if (err)
1041                                 goto cleanup;
1042                 }
1043
1044                 i--;
1045         }
1046
1047         /* insert new index */
1048         err = ext4_ext_insert_index(handle, inode, path + at,
1049                                     le32_to_cpu(border), newblock);
1050
1051 cleanup:
1052         if (bh) {
1053                 if (buffer_locked(bh))
1054                         unlock_buffer(bh);
1055                 brelse(bh);
1056         }
1057
1058         if (err) {
1059                 /* free all allocated blocks in error case */
1060                 for (i = 0; i < depth; i++) {
1061                         if (!ablocks[i])
1062                                 continue;
1063                         ext4_free_blocks(handle, inode, 0, ablocks[i], 1,
1064                                          EXT4_FREE_BLOCKS_METADATA);
1065                 }
1066         }
1067         kfree(ablocks);
1068
1069         return err;
1070 }
1071
1072 /*
1073  * ext4_ext_grow_indepth:
1074  * implements tree growing procedure:
1075  * - allocates new block
1076  * - moves top-level data (index block or leaf) into the new block
1077  * - initializes new top-level, creating index that points to the
1078  *   just created block
1079  */
1080 static int ext4_ext_grow_indepth(handle_t *handle, struct inode *inode,
1081                                         struct ext4_ext_path *path,
1082                                         struct ext4_extent *newext)
1083 {
1084         struct ext4_ext_path *curp = path;
1085         struct ext4_extent_header *neh;
1086         struct buffer_head *bh;
1087         ext4_fsblk_t newblock;
1088         int err = 0;
1089
1090         newblock = ext4_ext_new_meta_block(handle, inode, path, newext, &err);
1091         if (newblock == 0)
1092                 return err;
1093
1094         bh = sb_getblk(inode->i_sb, newblock);
1095         if (!bh) {
1096                 err = -EIO;
1097                 ext4_std_error(inode->i_sb, err);
1098                 return err;
1099         }
1100         lock_buffer(bh);
1101
1102         err = ext4_journal_get_create_access(handle, bh);
1103         if (err) {
1104                 unlock_buffer(bh);
1105                 goto out;
1106         }
1107
1108         /* move top-level index/leaf into new block */
1109         memmove(bh->b_data, curp->p_hdr, sizeof(EXT4_I(inode)->i_data));
1110
1111         /* set size of new block */
1112         neh = ext_block_hdr(bh);
1113         /* old root could have indexes or leaves
1114          * so calculate e_max right way */
1115         if (ext_depth(inode))
1116                 neh->eh_max = cpu_to_le16(ext4_ext_space_block_idx(inode, 0));
1117         else
1118                 neh->eh_max = cpu_to_le16(ext4_ext_space_block(inode, 0));
1119         neh->eh_magic = EXT4_EXT_MAGIC;
1120         set_buffer_uptodate(bh);
1121         unlock_buffer(bh);
1122
1123         err = ext4_handle_dirty_metadata(handle, inode, bh);
1124         if (err)
1125                 goto out;
1126
1127         /* create index in new top-level index: num,max,pointer */
1128         err = ext4_ext_get_access(handle, inode, curp);
1129         if (err)
1130                 goto out;
1131
1132         curp->p_hdr->eh_magic = EXT4_EXT_MAGIC;
1133         curp->p_hdr->eh_max = cpu_to_le16(ext4_ext_space_root_idx(inode, 0));
1134         curp->p_hdr->eh_entries = cpu_to_le16(1);
1135         curp->p_idx = EXT_FIRST_INDEX(curp->p_hdr);
1136
1137         if (path[0].p_hdr->eh_depth)
1138                 curp->p_idx->ei_block =
1139                         EXT_FIRST_INDEX(path[0].p_hdr)->ei_block;
1140         else
1141                 curp->p_idx->ei_block =
1142                         EXT_FIRST_EXTENT(path[0].p_hdr)->ee_block;
1143         ext4_idx_store_pblock(curp->p_idx, newblock);
1144
1145         neh = ext_inode_hdr(inode);
1146         ext_debug("new root: num %d(%d), lblock %d, ptr %llu\n",
1147                   le16_to_cpu(neh->eh_entries), le16_to_cpu(neh->eh_max),
1148                   le32_to_cpu(EXT_FIRST_INDEX(neh)->ei_block),
1149                   idx_pblock(EXT_FIRST_INDEX(neh)));
1150
1151         neh->eh_depth = cpu_to_le16(path->p_depth + 1);
1152         err = ext4_ext_dirty(handle, inode, curp);
1153 out:
1154         brelse(bh);
1155
1156         return err;
1157 }
1158
1159 /*
1160  * ext4_ext_create_new_leaf:
1161  * finds empty index and adds new leaf.
1162  * if no free index is found, then it requests in-depth growing.
1163  */
1164 static int ext4_ext_create_new_leaf(handle_t *handle, struct inode *inode,
1165                                         struct ext4_ext_path *path,
1166                                         struct ext4_extent *newext)
1167 {
1168         struct ext4_ext_path *curp;
1169         int depth, i, err = 0;
1170
1171 repeat:
1172         i = depth = ext_depth(inode);
1173
1174         /* walk up to the tree and look for free index entry */
1175         curp = path + depth;
1176         while (i > 0 && !EXT_HAS_FREE_INDEX(curp)) {
1177                 i--;
1178                 curp--;
1179         }
1180
1181         /* we use already allocated block for index block,
1182          * so subsequent data blocks should be contiguous */
1183         if (EXT_HAS_FREE_INDEX(curp)) {
1184                 /* if we found index with free entry, then use that
1185                  * entry: create all needed subtree and add new leaf */
1186                 err = ext4_ext_split(handle, inode, path, newext, i);
1187                 if (err)
1188                         goto out;
1189
1190                 /* refill path */
1191                 ext4_ext_drop_refs(path);
1192                 path = ext4_ext_find_extent(inode,
1193                                     (ext4_lblk_t)le32_to_cpu(newext->ee_block),
1194                                     path);
1195                 if (IS_ERR(path))
1196                         err = PTR_ERR(path);
1197         } else {
1198                 /* tree is full, time to grow in depth */
1199                 err = ext4_ext_grow_indepth(handle, inode, path, newext);
1200                 if (err)
1201                         goto out;
1202
1203                 /* refill path */
1204                 ext4_ext_drop_refs(path);
1205                 path = ext4_ext_find_extent(inode,
1206                                    (ext4_lblk_t)le32_to_cpu(newext->ee_block),
1207                                     path);
1208                 if (IS_ERR(path)) {
1209                         err = PTR_ERR(path);
1210                         goto out;
1211                 }
1212
1213                 /*
1214                  * only first (depth 0 -> 1) produces free space;
1215                  * in all other cases we have to split the grown tree
1216                  */
1217                 depth = ext_depth(inode);
1218                 if (path[depth].p_hdr->eh_entries == path[depth].p_hdr->eh_max) {
1219                         /* now we need to split */
1220                         goto repeat;
1221                 }
1222         }
1223
1224 out:
1225         return err;
1226 }
1227
1228 /*
1229  * search the closest allocated block to the left for *logical
1230  * and returns it at @logical + it's physical address at @phys
1231  * if *logical is the smallest allocated block, the function
1232  * returns 0 at @phys
1233  * return value contains 0 (success) or error code
1234  */
1235 int
1236 ext4_ext_search_left(struct inode *inode, struct ext4_ext_path *path,
1237                         ext4_lblk_t *logical, ext4_fsblk_t *phys)
1238 {
1239         struct ext4_extent_idx *ix;
1240         struct ext4_extent *ex;
1241         int depth, ee_len;
1242
1243         if (unlikely(path == NULL)) {
1244                 EXT4_ERROR_INODE(inode, "path == NULL *logical %d!", *logical);
1245                 return -EIO;
1246         }
1247         depth = path->p_depth;
1248         *phys = 0;
1249
1250         if (depth == 0 && path->p_ext == NULL)
1251                 return 0;
1252
1253         /* usually extent in the path covers blocks smaller
1254          * then *logical, but it can be that extent is the
1255          * first one in the file */
1256
1257         ex = path[depth].p_ext;
1258         ee_len = ext4_ext_get_actual_len(ex);
1259         if (*logical < le32_to_cpu(ex->ee_block)) {
1260                 if (unlikely(EXT_FIRST_EXTENT(path[depth].p_hdr) != ex)) {
1261                         EXT4_ERROR_INODE(inode,
1262                                          "EXT_FIRST_EXTENT != ex *logical %d ee_block %d!",
1263                                          *logical, le32_to_cpu(ex->ee_block));
1264                         return -EIO;
1265                 }
1266                 while (--depth >= 0) {
1267                         ix = path[depth].p_idx;
1268                         if (unlikely(ix != EXT_FIRST_INDEX(path[depth].p_hdr))) {
1269                                 EXT4_ERROR_INODE(inode,
1270                                   "ix (%d) != EXT_FIRST_INDEX (%d) (depth %d)!",
1271                                   ix != NULL ? ix->ei_block : 0,
1272                                   EXT_FIRST_INDEX(path[depth].p_hdr) != NULL ?
1273                                     EXT_FIRST_INDEX(path[depth].p_hdr)->ei_block : 0,
1274                                   depth);
1275                                 return -EIO;
1276                         }
1277                 }
1278                 return 0;
1279         }
1280
1281         if (unlikely(*logical < (le32_to_cpu(ex->ee_block) + ee_len))) {
1282                 EXT4_ERROR_INODE(inode,
1283                                  "logical %d < ee_block %d + ee_len %d!",
1284                                  *logical, le32_to_cpu(ex->ee_block), ee_len);
1285                 return -EIO;
1286         }
1287
1288         *logical = le32_to_cpu(ex->ee_block) + ee_len - 1;
1289         *phys = ext_pblock(ex) + ee_len - 1;
1290         return 0;
1291 }
1292
1293 /*
1294  * search the closest allocated block to the right for *logical
1295  * and returns it at @logical + it's physical address at @phys
1296  * if *logical is the smallest allocated block, the function
1297  * returns 0 at @phys
1298  * return value contains 0 (success) or error code
1299  */
1300 int
1301 ext4_ext_search_right(struct inode *inode, struct ext4_ext_path *path,
1302                         ext4_lblk_t *logical, ext4_fsblk_t *phys)
1303 {
1304         struct buffer_head *bh = NULL;
1305         struct ext4_extent_header *eh;
1306         struct ext4_extent_idx *ix;
1307         struct ext4_extent *ex;
1308         ext4_fsblk_t block;
1309         int depth;      /* Note, NOT eh_depth; depth from top of tree */
1310         int ee_len;
1311
1312         if (unlikely(path == NULL)) {
1313                 EXT4_ERROR_INODE(inode, "path == NULL *logical %d!", *logical);
1314                 return -EIO;
1315         }
1316         depth = path->p_depth;
1317         *phys = 0;
1318
1319         if (depth == 0 && path->p_ext == NULL)
1320                 return 0;
1321
1322         /* usually extent in the path covers blocks smaller
1323          * then *logical, but it can be that extent is the
1324          * first one in the file */
1325
1326         ex = path[depth].p_ext;
1327         ee_len = ext4_ext_get_actual_len(ex);
1328         if (*logical < le32_to_cpu(ex->ee_block)) {
1329                 if (unlikely(EXT_FIRST_EXTENT(path[depth].p_hdr) != ex)) {
1330                         EXT4_ERROR_INODE(inode,
1331                                          "first_extent(path[%d].p_hdr) != ex",
1332                                          depth);
1333                         return -EIO;
1334                 }
1335                 while (--depth >= 0) {
1336                         ix = path[depth].p_idx;
1337                         if (unlikely(ix != EXT_FIRST_INDEX(path[depth].p_hdr))) {
1338                                 EXT4_ERROR_INODE(inode,
1339                                                  "ix != EXT_FIRST_INDEX *logical %d!",
1340                                                  *logical);
1341                                 return -EIO;
1342                         }
1343                 }
1344                 *logical = le32_to_cpu(ex->ee_block);
1345                 *phys = ext_pblock(ex);
1346                 return 0;
1347         }
1348
1349         if (unlikely(*logical < (le32_to_cpu(ex->ee_block) + ee_len))) {
1350                 EXT4_ERROR_INODE(inode,
1351                                  "logical %d < ee_block %d + ee_len %d!",
1352                                  *logical, le32_to_cpu(ex->ee_block), ee_len);
1353                 return -EIO;
1354         }
1355
1356         if (ex != EXT_LAST_EXTENT(path[depth].p_hdr)) {
1357                 /* next allocated block in this leaf */
1358                 ex++;
1359                 *logical = le32_to_cpu(ex->ee_block);
1360                 *phys = ext_pblock(ex);
1361                 return 0;
1362         }
1363
1364         /* go up and search for index to the right */
1365         while (--depth >= 0) {
1366                 ix = path[depth].p_idx;
1367                 if (ix != EXT_LAST_INDEX(path[depth].p_hdr))
1368                         goto got_index;
1369         }
1370
1371         /* we've gone up to the root and found no index to the right */
1372         return 0;
1373
1374 got_index:
1375         /* we've found index to the right, let's
1376          * follow it and find the closest allocated
1377          * block to the right */
1378         ix++;
1379         block = idx_pblock(ix);
1380         while (++depth < path->p_depth) {
1381                 bh = sb_bread(inode->i_sb, block);
1382                 if (bh == NULL)
1383                         return -EIO;
1384                 eh = ext_block_hdr(bh);
1385                 /* subtract from p_depth to get proper eh_depth */
1386                 if (ext4_ext_check(inode, eh, path->p_depth - depth)) {
1387                         put_bh(bh);
1388                         return -EIO;
1389                 }
1390                 ix = EXT_FIRST_INDEX(eh);
1391                 block = idx_pblock(ix);
1392                 put_bh(bh);
1393         }
1394
1395         bh = sb_bread(inode->i_sb, block);
1396         if (bh == NULL)
1397                 return -EIO;
1398         eh = ext_block_hdr(bh);
1399         if (ext4_ext_check(inode, eh, path->p_depth - depth)) {
1400                 put_bh(bh);
1401                 return -EIO;
1402         }
1403         ex = EXT_FIRST_EXTENT(eh);
1404         *logical = le32_to_cpu(ex->ee_block);
1405         *phys = ext_pblock(ex);
1406         put_bh(bh);
1407         return 0;
1408 }
1409
1410 /*
1411  * ext4_ext_next_allocated_block:
1412  * returns allocated block in subsequent extent or EXT_MAX_BLOCK.
1413  * NOTE: it considers block number from index entry as
1414  * allocated block. Thus, index entries have to be consistent
1415  * with leaves.
1416  */
1417 static ext4_lblk_t
1418 ext4_ext_next_allocated_block(struct ext4_ext_path *path)
1419 {
1420         int depth;
1421
1422         BUG_ON(path == NULL);
1423         depth = path->p_depth;
1424
1425         if (depth == 0 && path->p_ext == NULL)
1426                 return EXT_MAX_BLOCK;
1427
1428         while (depth >= 0) {
1429                 if (depth == path->p_depth) {
1430                         /* leaf */
1431                         if (path[depth].p_ext !=
1432                                         EXT_LAST_EXTENT(path[depth].p_hdr))
1433                           return le32_to_cpu(path[depth].p_ext[1].ee_block);
1434                 } else {
1435                         /* index */
1436                         if (path[depth].p_idx !=
1437                                         EXT_LAST_INDEX(path[depth].p_hdr))
1438                           return le32_to_cpu(path[depth].p_idx[1].ei_block);
1439                 }
1440                 depth--;
1441         }
1442
1443         return EXT_MAX_BLOCK;
1444 }
1445
1446 /*
1447  * ext4_ext_next_leaf_block:
1448  * returns first allocated block from next leaf or EXT_MAX_BLOCK
1449  */
1450 static ext4_lblk_t ext4_ext_next_leaf_block(struct inode *inode,
1451                                         struct ext4_ext_path *path)
1452 {
1453         int depth;
1454
1455         BUG_ON(path == NULL);
1456         depth = path->p_depth;
1457
1458         /* zero-tree has no leaf blocks at all */
1459         if (depth == 0)
1460                 return EXT_MAX_BLOCK;
1461
1462         /* go to index block */
1463         depth--;
1464
1465         while (depth >= 0) {
1466                 if (path[depth].p_idx !=
1467                                 EXT_LAST_INDEX(path[depth].p_hdr))
1468                         return (ext4_lblk_t)
1469                                 le32_to_cpu(path[depth].p_idx[1].ei_block);
1470                 depth--;
1471         }
1472
1473         return EXT_MAX_BLOCK;
1474 }
1475
1476 /*
1477  * ext4_ext_correct_indexes:
1478  * if leaf gets modified and modified extent is first in the leaf,
1479  * then we have to correct all indexes above.
1480  * TODO: do we need to correct tree in all cases?
1481  */
1482 static int ext4_ext_correct_indexes(handle_t *handle, struct inode *inode,
1483                                 struct ext4_ext_path *path)
1484 {
1485         struct ext4_extent_header *eh;
1486         int depth = ext_depth(inode);
1487         struct ext4_extent *ex;
1488         __le32 border;
1489         int k, err = 0;
1490
1491         eh = path[depth].p_hdr;
1492         ex = path[depth].p_ext;
1493
1494         if (unlikely(ex == NULL || eh == NULL)) {
1495                 EXT4_ERROR_INODE(inode,
1496                                  "ex %p == NULL or eh %p == NULL", ex, eh);
1497                 return -EIO;
1498         }
1499
1500         if (depth == 0) {
1501                 /* there is no tree at all */
1502                 return 0;
1503         }
1504
1505         if (ex != EXT_FIRST_EXTENT(eh)) {
1506                 /* we correct tree if first leaf got modified only */
1507                 return 0;
1508         }
1509
1510         /*
1511          * TODO: we need correction if border is smaller than current one
1512          */
1513         k = depth - 1;
1514         border = path[depth].p_ext->ee_block;
1515         err = ext4_ext_get_access(handle, inode, path + k);
1516         if (err)
1517                 return err;
1518         path[k].p_idx->ei_block = border;
1519         err = ext4_ext_dirty(handle, inode, path + k);
1520         if (err)
1521                 return err;
1522
1523         while (k--) {
1524                 /* change all left-side indexes */
1525                 if (path[k+1].p_idx != EXT_FIRST_INDEX(path[k+1].p_hdr))
1526                         break;
1527                 err = ext4_ext_get_access(handle, inode, path + k);
1528                 if (err)
1529                         break;
1530                 path[k].p_idx->ei_block = border;
1531                 err = ext4_ext_dirty(handle, inode, path + k);
1532                 if (err)
1533                         break;
1534         }
1535
1536         return err;
1537 }
1538
1539 int
1540 ext4_can_extents_be_merged(struct inode *inode, struct ext4_extent *ex1,
1541                                 struct ext4_extent *ex2)
1542 {
1543         unsigned short ext1_ee_len, ext2_ee_len, max_len;
1544
1545         /*
1546          * Make sure that either both extents are uninitialized, or
1547          * both are _not_.
1548          */
1549         if (ext4_ext_is_uninitialized(ex1) ^ ext4_ext_is_uninitialized(ex2))
1550                 return 0;
1551
1552         if (ext4_ext_is_uninitialized(ex1))
1553                 max_len = EXT_UNINIT_MAX_LEN;
1554         else
1555                 max_len = EXT_INIT_MAX_LEN;
1556
1557         ext1_ee_len = ext4_ext_get_actual_len(ex1);
1558         ext2_ee_len = ext4_ext_get_actual_len(ex2);
1559
1560         if (le32_to_cpu(ex1->ee_block) + ext1_ee_len !=
1561                         le32_to_cpu(ex2->ee_block))
1562                 return 0;
1563
1564         /*
1565          * To allow future support for preallocated extents to be added
1566          * as an RO_COMPAT feature, refuse to merge to extents if
1567          * this can result in the top bit of ee_len being set.
1568          */
1569         if (ext1_ee_len + ext2_ee_len > max_len)
1570                 return 0;
1571 #ifdef AGGRESSIVE_TEST
1572         if (ext1_ee_len >= 4)
1573                 return 0;
1574 #endif
1575
1576         if (ext_pblock(ex1) + ext1_ee_len == ext_pblock(ex2))
1577                 return 1;
1578         return 0;
1579 }
1580
1581 /*
1582  * This function tries to merge the "ex" extent to the next extent in the tree.
1583  * It always tries to merge towards right. If you want to merge towards
1584  * left, pass "ex - 1" as argument instead of "ex".
1585  * Returns 0 if the extents (ex and ex+1) were _not_ merged and returns
1586  * 1 if they got merged.
1587  */
1588 int ext4_ext_try_to_merge(struct inode *inode,
1589                           struct ext4_ext_path *path,
1590                           struct ext4_extent *ex)
1591 {
1592         struct ext4_extent_header *eh;
1593         unsigned int depth, len;
1594         int merge_done = 0;
1595         int uninitialized = 0;
1596
1597         depth = ext_depth(inode);
1598         BUG_ON(path[depth].p_hdr == NULL);
1599         eh = path[depth].p_hdr;
1600
1601         while (ex < EXT_LAST_EXTENT(eh)) {
1602                 if (!ext4_can_extents_be_merged(inode, ex, ex + 1))
1603                         break;
1604                 /* merge with next extent! */
1605                 if (ext4_ext_is_uninitialized(ex))
1606                         uninitialized = 1;
1607                 ex->ee_len = cpu_to_le16(ext4_ext_get_actual_len(ex)
1608                                 + ext4_ext_get_actual_len(ex + 1));
1609                 if (uninitialized)
1610                         ext4_ext_mark_uninitialized(ex);
1611
1612                 if (ex + 1 < EXT_LAST_EXTENT(eh)) {
1613                         len = (EXT_LAST_EXTENT(eh) - ex - 1)
1614                                 * sizeof(struct ext4_extent);
1615                         memmove(ex + 1, ex + 2, len);
1616                 }
1617                 le16_add_cpu(&eh->eh_entries, -1);
1618                 merge_done = 1;
1619                 WARN_ON(eh->eh_entries == 0);
1620                 if (!eh->eh_entries)
1621                         EXT4_ERROR_INODE(inode, "eh->eh_entries = 0!");
1622         }
1623
1624         return merge_done;
1625 }
1626
1627 /*
1628  * check if a portion of the "newext" extent overlaps with an
1629  * existing extent.
1630  *
1631  * If there is an overlap discovered, it updates the length of the newext
1632  * such that there will be no overlap, and then returns 1.
1633  * If there is no overlap found, it returns 0.
1634  */
1635 unsigned int ext4_ext_check_overlap(struct inode *inode,
1636                                     struct ext4_extent *newext,
1637                                     struct ext4_ext_path *path)
1638 {
1639         ext4_lblk_t b1, b2;
1640         unsigned int depth, len1;
1641         unsigned int ret = 0;
1642
1643         b1 = le32_to_cpu(newext->ee_block);
1644         len1 = ext4_ext_get_actual_len(newext);
1645         depth = ext_depth(inode);
1646         if (!path[depth].p_ext)
1647                 goto out;
1648         b2 = le32_to_cpu(path[depth].p_ext->ee_block);
1649
1650         /*
1651          * get the next allocated block if the extent in the path
1652          * is before the requested block(s)
1653          */
1654         if (b2 < b1) {
1655                 b2 = ext4_ext_next_allocated_block(path);
1656                 if (b2 == EXT_MAX_BLOCK)
1657                         goto out;
1658         }
1659
1660         /* check for wrap through zero on extent logical start block*/
1661         if (b1 + len1 < b1) {
1662                 len1 = EXT_MAX_BLOCK - b1;
1663                 newext->ee_len = cpu_to_le16(len1);
1664                 ret = 1;
1665         }
1666
1667         /* check for overlap */
1668         if (b1 + len1 > b2) {
1669                 newext->ee_len = cpu_to_le16(b2 - b1);
1670                 ret = 1;
1671         }
1672 out:
1673         return ret;
1674 }
1675
1676 /*
1677  * ext4_ext_insert_extent:
1678  * tries to merge requsted extent into the existing extent or
1679  * inserts requested extent as new one into the tree,
1680  * creating new leaf in the no-space case.
1681  */
1682 int ext4_ext_insert_extent(handle_t *handle, struct inode *inode,
1683                                 struct ext4_ext_path *path,
1684                                 struct ext4_extent *newext, int flag)
1685 {
1686         struct ext4_extent_header *eh;
1687         struct ext4_extent *ex, *fex;
1688         struct ext4_extent *nearex; /* nearest extent */
1689         struct ext4_ext_path *npath = NULL;
1690         int depth, len, err;
1691         ext4_lblk_t next;
1692         unsigned uninitialized = 0;
1693
1694         if (unlikely(ext4_ext_get_actual_len(newext) == 0)) {
1695                 EXT4_ERROR_INODE(inode, "ext4_ext_get_actual_len(newext) == 0");
1696                 return -EIO;
1697         }
1698         depth = ext_depth(inode);
1699         ex = path[depth].p_ext;
1700         if (unlikely(path[depth].p_hdr == NULL)) {
1701                 EXT4_ERROR_INODE(inode, "path[%d].p_hdr == NULL", depth);
1702                 return -EIO;
1703         }
1704
1705         /* try to insert block into found extent and return */
1706         if (ex && !(flag & EXT4_GET_BLOCKS_PRE_IO)
1707                 && ext4_can_extents_be_merged(inode, ex, newext)) {
1708                 ext_debug("append [%d]%d block to %d:[%d]%d (from %llu)\n",
1709                                 ext4_ext_is_uninitialized(newext),
1710                                 ext4_ext_get_actual_len(newext),
1711                                 le32_to_cpu(ex->ee_block),
1712                                 ext4_ext_is_uninitialized(ex),
1713                                 ext4_ext_get_actual_len(ex), ext_pblock(ex));
1714                 err = ext4_ext_get_access(handle, inode, path + depth);
1715                 if (err)
1716                         return err;
1717
1718                 /*
1719                  * ext4_can_extents_be_merged should have checked that either
1720                  * both extents are uninitialized, or both aren't. Thus we
1721                  * need to check only one of them here.
1722                  */
1723                 if (ext4_ext_is_uninitialized(ex))
1724                         uninitialized = 1;
1725                 ex->ee_len = cpu_to_le16(ext4_ext_get_actual_len(ex)
1726                                         + ext4_ext_get_actual_len(newext));
1727                 if (uninitialized)
1728                         ext4_ext_mark_uninitialized(ex);
1729                 eh = path[depth].p_hdr;
1730                 nearex = ex;
1731                 goto merge;
1732         }
1733
1734 repeat:
1735         depth = ext_depth(inode);
1736         eh = path[depth].p_hdr;
1737         if (le16_to_cpu(eh->eh_entries) < le16_to_cpu(eh->eh_max))
1738                 goto has_space;
1739
1740         /* probably next leaf has space for us? */
1741         fex = EXT_LAST_EXTENT(eh);
1742         next = ext4_ext_next_leaf_block(inode, path);
1743         if (le32_to_cpu(newext->ee_block) > le32_to_cpu(fex->ee_block)
1744             && next != EXT_MAX_BLOCK) {
1745                 ext_debug("next leaf block - %d\n", next);
1746                 BUG_ON(npath != NULL);
1747                 npath = ext4_ext_find_extent(inode, next, NULL);
1748                 if (IS_ERR(npath))
1749                         return PTR_ERR(npath);
1750                 BUG_ON(npath->p_depth != path->p_depth);
1751                 eh = npath[depth].p_hdr;
1752                 if (le16_to_cpu(eh->eh_entries) < le16_to_cpu(eh->eh_max)) {
1753                         ext_debug("next leaf isnt full(%d)\n",
1754                                   le16_to_cpu(eh->eh_entries));
1755                         path = npath;
1756                         goto repeat;
1757                 }
1758                 ext_debug("next leaf has no free space(%d,%d)\n",
1759                           le16_to_cpu(eh->eh_entries), le16_to_cpu(eh->eh_max));
1760         }
1761
1762         /*
1763          * There is no free space in the found leaf.
1764          * We're gonna add a new leaf in the tree.
1765          */
1766         err = ext4_ext_create_new_leaf(handle, inode, path, newext);
1767         if (err)
1768                 goto cleanup;
1769         depth = ext_depth(inode);
1770         eh = path[depth].p_hdr;
1771
1772 has_space:
1773         nearex = path[depth].p_ext;
1774
1775         err = ext4_ext_get_access(handle, inode, path + depth);
1776         if (err)
1777                 goto cleanup;
1778
1779         if (!nearex) {
1780                 /* there is no extent in this leaf, create first one */
1781                 ext_debug("first extent in the leaf: %d:%llu:[%d]%d\n",
1782                                 le32_to_cpu(newext->ee_block),
1783                                 ext_pblock(newext),
1784                                 ext4_ext_is_uninitialized(newext),
1785                                 ext4_ext_get_actual_len(newext));
1786                 path[depth].p_ext = EXT_FIRST_EXTENT(eh);
1787         } else if (le32_to_cpu(newext->ee_block)
1788                            > le32_to_cpu(nearex->ee_block)) {
1789 /*              BUG_ON(newext->ee_block == nearex->ee_block); */
1790                 if (nearex != EXT_LAST_EXTENT(eh)) {
1791                         len = EXT_MAX_EXTENT(eh) - nearex;
1792                         len = (len - 1) * sizeof(struct ext4_extent);
1793                         len = len < 0 ? 0 : len;
1794                         ext_debug("insert %d:%llu:[%d]%d after: nearest 0x%p, "
1795                                         "move %d from 0x%p to 0x%p\n",
1796                                         le32_to_cpu(newext->ee_block),
1797                                         ext_pblock(newext),
1798                                         ext4_ext_is_uninitialized(newext),
1799                                         ext4_ext_get_actual_len(newext),
1800                                         nearex, len, nearex + 1, nearex + 2);
1801                         memmove(nearex + 2, nearex + 1, len);
1802                 }
1803                 path[depth].p_ext = nearex + 1;
1804         } else {
1805                 BUG_ON(newext->ee_block == nearex->ee_block);
1806                 len = (EXT_MAX_EXTENT(eh) - nearex) * sizeof(struct ext4_extent);
1807                 len = len < 0 ? 0 : len;
1808                 ext_debug("insert %d:%llu:[%d]%d before: nearest 0x%p, "
1809                                 "move %d from 0x%p to 0x%p\n",
1810                                 le32_to_cpu(newext->ee_block),
1811                                 ext_pblock(newext),
1812                                 ext4_ext_is_uninitialized(newext),
1813                                 ext4_ext_get_actual_len(newext),
1814                                 nearex, len, nearex + 1, nearex + 2);
1815                 memmove(nearex + 1, nearex, len);
1816                 path[depth].p_ext = nearex;
1817         }
1818
1819         le16_add_cpu(&eh->eh_entries, 1);
1820         nearex = path[depth].p_ext;
1821         nearex->ee_block = newext->ee_block;
1822         ext4_ext_store_pblock(nearex, ext_pblock(newext));
1823         nearex->ee_len = newext->ee_len;
1824
1825 merge:
1826         /* try to merge extents to the right */
1827         if (!(flag & EXT4_GET_BLOCKS_PRE_IO))
1828                 ext4_ext_try_to_merge(inode, path, nearex);
1829
1830         /* try to merge extents to the left */
1831
1832         /* time to correct all indexes above */
1833         err = ext4_ext_correct_indexes(handle, inode, path);
1834         if (err)
1835                 goto cleanup;
1836
1837         err = ext4_ext_dirty(handle, inode, path + depth);
1838
1839 cleanup:
1840         if (npath) {
1841                 ext4_ext_drop_refs(npath);
1842                 kfree(npath);
1843         }
1844         ext4_ext_invalidate_cache(inode);
1845         return err;
1846 }
1847
1848 int ext4_ext_walk_space(struct inode *inode, ext4_lblk_t block,
1849                         ext4_lblk_t num, ext_prepare_callback func,
1850                         void *cbdata)
1851 {
1852         struct ext4_ext_path *path = NULL;
1853         struct ext4_ext_cache cbex;
1854         struct ext4_extent *ex;
1855         ext4_lblk_t next, start = 0, end = 0;
1856         ext4_lblk_t last = block + num;
1857         int depth, exists, err = 0;
1858
1859         BUG_ON(func == NULL);
1860         BUG_ON(inode == NULL);
1861
1862         while (block < last && block != EXT_MAX_BLOCK) {
1863                 num = last - block;
1864                 /* find extent for this block */
1865                 down_read(&EXT4_I(inode)->i_data_sem);
1866                 path = ext4_ext_find_extent(inode, block, path);
1867                 up_read(&EXT4_I(inode)->i_data_sem);
1868                 if (IS_ERR(path)) {
1869                         err = PTR_ERR(path);
1870                         path = NULL;
1871                         break;
1872                 }
1873
1874                 depth = ext_depth(inode);
1875                 if (unlikely(path[depth].p_hdr == NULL)) {
1876                         EXT4_ERROR_INODE(inode, "path[%d].p_hdr == NULL", depth);
1877                         err = -EIO;
1878                         break;
1879                 }
1880                 ex = path[depth].p_ext;
1881                 next = ext4_ext_next_allocated_block(path);
1882
1883                 exists = 0;
1884                 if (!ex) {
1885                         /* there is no extent yet, so try to allocate
1886                          * all requested space */
1887                         start = block;
1888                         end = block + num;
1889                 } else if (le32_to_cpu(ex->ee_block) > block) {
1890                         /* need to allocate space before found extent */
1891                         start = block;
1892                         end = le32_to_cpu(ex->ee_block);
1893                         if (block + num < end)
1894                                 end = block + num;
1895                 } else if (block >= le32_to_cpu(ex->ee_block)
1896                                         + ext4_ext_get_actual_len(ex)) {
1897                         /* need to allocate space after found extent */
1898                         start = block;
1899                         end = block + num;
1900                         if (end >= next)
1901                                 end = next;
1902                 } else if (block >= le32_to_cpu(ex->ee_block)) {
1903                         /*
1904                          * some part of requested space is covered
1905                          * by found extent
1906                          */
1907                         start = block;
1908                         end = le32_to_cpu(ex->ee_block)
1909                                 + ext4_ext_get_actual_len(ex);
1910                         if (block + num < end)
1911                                 end = block + num;
1912                         exists = 1;
1913                 } else {
1914                         BUG();
1915                 }
1916                 BUG_ON(end <= start);
1917
1918                 if (!exists) {
1919                         cbex.ec_block = start;
1920                         cbex.ec_len = end - start;
1921                         cbex.ec_start = 0;
1922                         cbex.ec_type = EXT4_EXT_CACHE_GAP;
1923                 } else {
1924                         cbex.ec_block = le32_to_cpu(ex->ee_block);
1925                         cbex.ec_len = ext4_ext_get_actual_len(ex);
1926                         cbex.ec_start = ext_pblock(ex);
1927                         cbex.ec_type = EXT4_EXT_CACHE_EXTENT;
1928                 }
1929
1930                 if (unlikely(cbex.ec_len == 0)) {
1931                         EXT4_ERROR_INODE(inode, "cbex.ec_len == 0");
1932                         err = -EIO;
1933                         break;
1934                 }
1935                 err = func(inode, path, &cbex, ex, cbdata);
1936                 ext4_ext_drop_refs(path);
1937
1938                 if (err < 0)
1939                         break;
1940
1941                 if (err == EXT_REPEAT)
1942                         continue;
1943                 else if (err == EXT_BREAK) {
1944                         err = 0;
1945                         break;
1946                 }
1947
1948                 if (ext_depth(inode) != depth) {
1949                         /* depth was changed. we have to realloc path */
1950                         kfree(path);
1951                         path = NULL;
1952                 }
1953
1954                 block = cbex.ec_block + cbex.ec_len;
1955         }
1956
1957         if (path) {
1958                 ext4_ext_drop_refs(path);
1959                 kfree(path);
1960         }
1961
1962         return err;
1963 }
1964
1965 static void
1966 ext4_ext_put_in_cache(struct inode *inode, ext4_lblk_t block,
1967                         __u32 len, ext4_fsblk_t start, int type)
1968 {
1969         struct ext4_ext_cache *cex;
1970         BUG_ON(len == 0);
1971         spin_lock(&EXT4_I(inode)->i_block_reservation_lock);
1972         cex = &EXT4_I(inode)->i_cached_extent;
1973         cex->ec_type = type;
1974         cex->ec_block = block;
1975         cex->ec_len = len;
1976         cex->ec_start = start;
1977         spin_unlock(&EXT4_I(inode)->i_block_reservation_lock);
1978 }
1979
1980 /*
1981  * ext4_ext_put_gap_in_cache:
1982  * calculate boundaries of the gap that the requested block fits into
1983  * and cache this gap
1984  */
1985 static void
1986 ext4_ext_put_gap_in_cache(struct inode *inode, struct ext4_ext_path *path,
1987                                 ext4_lblk_t block)
1988 {
1989         int depth = ext_depth(inode);
1990         unsigned long len;
1991         ext4_lblk_t lblock;
1992         struct ext4_extent *ex;
1993
1994         ex = path[depth].p_ext;
1995         if (ex == NULL) {
1996                 /* there is no extent yet, so gap is [0;-] */
1997                 lblock = 0;
1998                 len = EXT_MAX_BLOCK;
1999                 ext_debug("cache gap(whole file):");
2000         } else if (block < le32_to_cpu(ex->ee_block)) {
2001                 lblock = block;
2002                 len = le32_to_cpu(ex->ee_block) - block;
2003                 ext_debug("cache gap(before): %u [%u:%u]",
2004                                 block,
2005                                 le32_to_cpu(ex->ee_block),
2006                                  ext4_ext_get_actual_len(ex));
2007         } else if (block >= le32_to_cpu(ex->ee_block)
2008                         + ext4_ext_get_actual_len(ex)) {
2009                 ext4_lblk_t next;
2010                 lblock = le32_to_cpu(ex->ee_block)
2011                         + ext4_ext_get_actual_len(ex);
2012
2013                 next = ext4_ext_next_allocated_block(path);
2014                 ext_debug("cache gap(after): [%u:%u] %u",
2015                                 le32_to_cpu(ex->ee_block),
2016                                 ext4_ext_get_actual_len(ex),
2017                                 block);
2018                 BUG_ON(next == lblock);
2019                 len = next - lblock;
2020         } else {
2021                 lblock = len = 0;
2022                 BUG();
2023         }
2024
2025         ext_debug(" -> %u:%lu\n", lblock, len);
2026         ext4_ext_put_in_cache(inode, lblock, len, 0, EXT4_EXT_CACHE_GAP);
2027 }
2028
2029 static int
2030 ext4_ext_in_cache(struct inode *inode, ext4_lblk_t block,
2031                         struct ext4_extent *ex)
2032 {
2033         struct ext4_ext_cache *cex;
2034         int ret = EXT4_EXT_CACHE_NO;
2035
2036         /*
2037          * We borrow i_block_reservation_lock to protect i_cached_extent
2038          */
2039         spin_lock(&EXT4_I(inode)->i_block_reservation_lock);
2040         cex = &EXT4_I(inode)->i_cached_extent;
2041
2042         /* has cache valid data? */
2043         if (cex->ec_type == EXT4_EXT_CACHE_NO)
2044                 goto errout;
2045
2046         BUG_ON(cex->ec_type != EXT4_EXT_CACHE_GAP &&
2047                         cex->ec_type != EXT4_EXT_CACHE_EXTENT);
2048         if (in_range(block, cex->ec_block, cex->ec_len)) {
2049                 ex->ee_block = cpu_to_le32(cex->ec_block);
2050                 ext4_ext_store_pblock(ex, cex->ec_start);
2051                 ex->ee_len = cpu_to_le16(cex->ec_len);
2052                 ext_debug("%u cached by %u:%u:%llu\n",
2053                                 block,
2054                                 cex->ec_block, cex->ec_len, cex->ec_start);
2055                 ret = cex->ec_type;
2056         }
2057 errout:
2058         spin_unlock(&EXT4_I(inode)->i_block_reservation_lock);
2059         return ret;
2060 }
2061
2062 /*
2063  * ext4_ext_rm_idx:
2064  * removes index from the index block.
2065  * It's used in truncate case only, thus all requests are for
2066  * last index in the block only.
2067  */
2068 static int ext4_ext_rm_idx(handle_t *handle, struct inode *inode,
2069                         struct ext4_ext_path *path)
2070 {
2071         int err;
2072         ext4_fsblk_t leaf;
2073
2074         /* free index block */
2075         path--;
2076         leaf = idx_pblock(path->p_idx);
2077         if (unlikely(path->p_hdr->eh_entries == 0)) {
2078                 EXT4_ERROR_INODE(inode, "path->p_hdr->eh_entries == 0");
2079                 return -EIO;
2080         }
2081         err = ext4_ext_get_access(handle, inode, path);
2082         if (err)
2083                 return err;
2084         le16_add_cpu(&path->p_hdr->eh_entries, -1);
2085         err = ext4_ext_dirty(handle, inode, path);
2086         if (err)
2087                 return err;
2088         ext_debug("index is empty, remove it, free block %llu\n", leaf);
2089         ext4_free_blocks(handle, inode, 0, leaf, 1,
2090                          EXT4_FREE_BLOCKS_METADATA | EXT4_FREE_BLOCKS_FORGET);
2091         return err;
2092 }
2093
2094 /*
2095  * ext4_ext_calc_credits_for_single_extent:
2096  * This routine returns max. credits that needed to insert an extent
2097  * to the extent tree.
2098  * When pass the actual path, the caller should calculate credits
2099  * under i_data_sem.
2100  */
2101 int ext4_ext_calc_credits_for_single_extent(struct inode *inode, int nrblocks,
2102                                                 struct ext4_ext_path *path)
2103 {
2104         if (path) {
2105                 int depth = ext_depth(inode);
2106                 int ret = 0;
2107
2108                 /* probably there is space in leaf? */
2109                 if (le16_to_cpu(path[depth].p_hdr->eh_entries)
2110                                 < le16_to_cpu(path[depth].p_hdr->eh_max)) {
2111
2112                         /*
2113                          *  There are some space in the leaf tree, no
2114                          *  need to account for leaf block credit
2115                          *
2116                          *  bitmaps and block group descriptor blocks
2117                          *  and other metadat blocks still need to be
2118                          *  accounted.
2119                          */
2120                         /* 1 bitmap, 1 block group descriptor */
2121                         ret = 2 + EXT4_META_TRANS_BLOCKS(inode->i_sb);
2122                         return ret;
2123                 }
2124         }
2125
2126         return ext4_chunk_trans_blocks(inode, nrblocks);
2127 }
2128
2129 /*
2130  * How many index/leaf blocks need to change/allocate to modify nrblocks?
2131  *
2132  * if nrblocks are fit in a single extent (chunk flag is 1), then
2133  * in the worse case, each tree level index/leaf need to be changed
2134  * if the tree split due to insert a new extent, then the old tree
2135  * index/leaf need to be updated too
2136  *
2137  * If the nrblocks are discontiguous, they could cause
2138  * the whole tree split more than once, but this is really rare.
2139  */
2140 int ext4_ext_index_trans_blocks(struct inode *inode, int nrblocks, int chunk)
2141 {
2142         int index;
2143         int depth = ext_depth(inode);
2144
2145         if (chunk)
2146                 index = depth * 2;
2147         else
2148                 index = depth * 3;
2149
2150         return index;
2151 }
2152
2153 static int ext4_remove_blocks(handle_t *handle, struct inode *inode,
2154                                 struct ext4_extent *ex,
2155                                 ext4_lblk_t from, ext4_lblk_t to)
2156 {
2157         unsigned short ee_len =  ext4_ext_get_actual_len(ex);
2158         int flags = EXT4_FREE_BLOCKS_FORGET;
2159
2160         if (S_ISDIR(inode->i_mode) || S_ISLNK(inode->i_mode))
2161                 flags |= EXT4_FREE_BLOCKS_METADATA;
2162 #ifdef EXTENTS_STATS
2163         {
2164                 struct ext4_sb_info *sbi = EXT4_SB(inode->i_sb);
2165                 spin_lock(&sbi->s_ext_stats_lock);
2166                 sbi->s_ext_blocks += ee_len;
2167                 sbi->s_ext_extents++;
2168                 if (ee_len < sbi->s_ext_min)
2169                         sbi->s_ext_min = ee_len;
2170                 if (ee_len > sbi->s_ext_max)
2171                         sbi->s_ext_max = ee_len;
2172                 if (ext_depth(inode) > sbi->s_depth_max)
2173                         sbi->s_depth_max = ext_depth(inode);
2174                 spin_unlock(&sbi->s_ext_stats_lock);
2175         }
2176 #endif
2177         if (from >= le32_to_cpu(ex->ee_block)
2178             && to == le32_to_cpu(ex->ee_block) + ee_len - 1) {
2179                 /* tail removal */
2180                 ext4_lblk_t num;
2181                 ext4_fsblk_t start;
2182
2183                 num = le32_to_cpu(ex->ee_block) + ee_len - from;
2184                 start = ext_pblock(ex) + ee_len - num;
2185                 ext_debug("free last %u blocks starting %llu\n", num, start);
2186                 ext4_free_blocks(handle, inode, 0, start, num, flags);
2187         } else if (from == le32_to_cpu(ex->ee_block)
2188                    && to <= le32_to_cpu(ex->ee_block) + ee_len - 1) {
2189                 printk(KERN_INFO "strange request: removal %u-%u from %u:%u\n",
2190                         from, to, le32_to_cpu(ex->ee_block), ee_len);
2191         } else {
2192                 printk(KERN_INFO "strange request: removal(2) "
2193                                 "%u-%u from %u:%u\n",
2194                                 from, to, le32_to_cpu(ex->ee_block), ee_len);
2195         }
2196         return 0;
2197 }
2198
2199 static int
2200 ext4_ext_rm_leaf(handle_t *handle, struct inode *inode,
2201                 struct ext4_ext_path *path, ext4_lblk_t start)
2202 {
2203         int err = 0, correct_index = 0;
2204         int depth = ext_depth(inode), credits;
2205         struct ext4_extent_header *eh;
2206         ext4_lblk_t a, b, block;
2207         unsigned num;
2208         ext4_lblk_t ex_ee_block;
2209         unsigned short ex_ee_len;
2210         unsigned uninitialized = 0;
2211         struct ext4_extent *ex;
2212
2213         /* the header must be checked already in ext4_ext_remove_space() */
2214         ext_debug("truncate since %u in leaf\n", start);
2215         if (!path[depth].p_hdr)
2216                 path[depth].p_hdr = ext_block_hdr(path[depth].p_bh);
2217         eh = path[depth].p_hdr;
2218         if (unlikely(path[depth].p_hdr == NULL)) {
2219                 EXT4_ERROR_INODE(inode, "path[%d].p_hdr == NULL", depth);
2220                 return -EIO;
2221         }
2222         /* find where to start removing */
2223         ex = EXT_LAST_EXTENT(eh);
2224
2225         ex_ee_block = le32_to_cpu(ex->ee_block);
2226         ex_ee_len = ext4_ext_get_actual_len(ex);
2227
2228         while (ex >= EXT_FIRST_EXTENT(eh) &&
2229                         ex_ee_block + ex_ee_len > start) {
2230
2231                 if (ext4_ext_is_uninitialized(ex))
2232                         uninitialized = 1;
2233                 else
2234                         uninitialized = 0;
2235
2236                 ext_debug("remove ext %u:[%d]%d\n", ex_ee_block,
2237                          uninitialized, ex_ee_len);
2238                 path[depth].p_ext = ex;
2239
2240                 a = ex_ee_block > start ? ex_ee_block : start;
2241                 b = ex_ee_block + ex_ee_len - 1 < EXT_MAX_BLOCK ?
2242                         ex_ee_block + ex_ee_len - 1 : EXT_MAX_BLOCK;
2243
2244                 ext_debug("  border %u:%u\n", a, b);
2245
2246                 if (a != ex_ee_block && b != ex_ee_block + ex_ee_len - 1) {
2247                         block = 0;
2248                         num = 0;
2249                         BUG();
2250                 } else if (a != ex_ee_block) {
2251                         /* remove tail of the extent */
2252                         block = ex_ee_block;
2253                         num = a - block;
2254                 } else if (b != ex_ee_block + ex_ee_len - 1) {
2255                         /* remove head of the extent */
2256                         block = a;
2257                         num = b - a;
2258                         /* there is no "make a hole" API yet */
2259                         BUG();
2260                 } else {
2261                         /* remove whole extent: excellent! */
2262                         block = ex_ee_block;
2263                         num = 0;
2264                         BUG_ON(a != ex_ee_block);
2265                         BUG_ON(b != ex_ee_block + ex_ee_len - 1);
2266                 }
2267
2268                 /*
2269                  * 3 for leaf, sb, and inode plus 2 (bmap and group
2270                  * descriptor) for each block group; assume two block
2271                  * groups plus ex_ee_len/blocks_per_block_group for
2272                  * the worst case
2273                  */
2274                 credits = 7 + 2*(ex_ee_len/EXT4_BLOCKS_PER_GROUP(inode->i_sb));
2275                 if (ex == EXT_FIRST_EXTENT(eh)) {
2276                         correct_index = 1;
2277                         credits += (ext_depth(inode)) + 1;
2278                 }
2279                 credits += EXT4_MAXQUOTAS_TRANS_BLOCKS(inode->i_sb);
2280
2281                 err = ext4_ext_truncate_extend_restart(handle, inode, credits);
2282                 if (err)
2283                         goto out;
2284
2285                 err = ext4_ext_get_access(handle, inode, path + depth);
2286                 if (err)
2287                         goto out;
2288
2289                 err = ext4_remove_blocks(handle, inode, ex, a, b);
2290                 if (err)
2291                         goto out;
2292
2293                 if (num == 0) {
2294                         /* this extent is removed; mark slot entirely unused */
2295                         ext4_ext_store_pblock(ex, 0);
2296                         le16_add_cpu(&eh->eh_entries, -1);
2297                 }
2298
2299                 ex->ee_block = cpu_to_le32(block);
2300                 ex->ee_len = cpu_to_le16(num);
2301                 /*
2302                  * Do not mark uninitialized if all the blocks in the
2303                  * extent have been removed.
2304                  */
2305                 if (uninitialized && num)
2306                         ext4_ext_mark_uninitialized(ex);
2307
2308                 err = ext4_ext_dirty(handle, inode, path + depth);
2309                 if (err)
2310                         goto out;
2311
2312                 ext_debug("new extent: %u:%u:%llu\n", block, num,
2313                                 ext_pblock(ex));
2314                 ex--;
2315                 ex_ee_block = le32_to_cpu(ex->ee_block);
2316                 ex_ee_len = ext4_ext_get_actual_len(ex);
2317         }
2318
2319         if (correct_index && eh->eh_entries)
2320                 err = ext4_ext_correct_indexes(handle, inode, path);
2321
2322         /* if this leaf is free, then we should
2323          * remove it from index block above */
2324         if (err == 0 && eh->eh_entries == 0 && path[depth].p_bh != NULL)
2325                 err = ext4_ext_rm_idx(handle, inode, path + depth);
2326
2327 out:
2328         return err;
2329 }
2330
2331 /*
2332  * ext4_ext_more_to_rm:
2333  * returns 1 if current index has to be freed (even partial)
2334  */
2335 static int
2336 ext4_ext_more_to_rm(struct ext4_ext_path *path)
2337 {
2338         BUG_ON(path->p_idx == NULL);
2339
2340         if (path->p_idx < EXT_FIRST_INDEX(path->p_hdr))
2341                 return 0;
2342
2343         /*
2344          * if truncate on deeper level happened, it wasn't partial,
2345          * so we have to consider current index for truncation
2346          */
2347         if (le16_to_cpu(path->p_hdr->eh_entries) == path->p_block)
2348                 return 0;
2349         return 1;
2350 }
2351
2352 static int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start)
2353 {
2354         struct super_block *sb = inode->i_sb;
2355         int depth = ext_depth(inode);
2356         struct ext4_ext_path *path;
2357         handle_t *handle;
2358         int i, err;
2359
2360         ext_debug("truncate since %u\n", start);
2361
2362         /* probably first extent we're gonna free will be last in block */
2363         handle = ext4_journal_start(inode, depth + 1);
2364         if (IS_ERR(handle))
2365                 return PTR_ERR(handle);
2366
2367 again:
2368         ext4_ext_invalidate_cache(inode);
2369
2370         /*
2371          * We start scanning from right side, freeing all the blocks
2372          * after i_size and walking into the tree depth-wise.
2373          */
2374         depth = ext_depth(inode);
2375         path = kzalloc(sizeof(struct ext4_ext_path) * (depth + 1), GFP_NOFS);
2376         if (path == NULL) {
2377                 ext4_journal_stop(handle);
2378                 return -ENOMEM;
2379         }
2380         path[0].p_depth = depth;
2381         path[0].p_hdr = ext_inode_hdr(inode);
2382         if (ext4_ext_check(inode, path[0].p_hdr, depth)) {
2383                 err = -EIO;
2384                 goto out;
2385         }
2386         i = err = 0;
2387
2388         while (i >= 0 && err == 0) {
2389                 if (i == depth) {
2390                         /* this is leaf block */
2391                         err = ext4_ext_rm_leaf(handle, inode, path, start);
2392                         /* root level has p_bh == NULL, brelse() eats this */
2393                         brelse(path[i].p_bh);
2394                         path[i].p_bh = NULL;
2395                         i--;
2396                         continue;
2397                 }
2398
2399                 /* this is index block */
2400                 if (!path[i].p_hdr) {
2401                         ext_debug("initialize header\n");
2402                         path[i].p_hdr = ext_block_hdr(path[i].p_bh);
2403                 }
2404
2405                 if (!path[i].p_idx) {
2406                         /* this level hasn't been touched yet */
2407                         path[i].p_idx = EXT_LAST_INDEX(path[i].p_hdr);
2408                         path[i].p_block = le16_to_cpu(path[i].p_hdr->eh_entries)+1;
2409                         ext_debug("init index ptr: hdr 0x%p, num %d\n",
2410                                   path[i].p_hdr,
2411                                   le16_to_cpu(path[i].p_hdr->eh_entries));
2412                 } else {
2413                         /* we were already here, see at next index */
2414                         path[i].p_idx--;
2415                 }
2416
2417                 ext_debug("level %d - index, first 0x%p, cur 0x%p\n",
2418                                 i, EXT_FIRST_INDEX(path[i].p_hdr),
2419                                 path[i].p_idx);
2420                 if (ext4_ext_more_to_rm(path + i)) {
2421                         struct buffer_head *bh;
2422                         /* go to the next level */
2423                         ext_debug("move to level %d (block %llu)\n",
2424                                   i + 1, idx_pblock(path[i].p_idx));
2425                         memset(path + i + 1, 0, sizeof(*path));
2426                         bh = sb_bread(sb, idx_pblock(path[i].p_idx));
2427                         if (!bh) {
2428                                 /* should we reset i_size? */
2429                                 err = -EIO;
2430                                 break;
2431                         }
2432                         if (WARN_ON(i + 1 > depth)) {
2433                                 err = -EIO;
2434                                 break;
2435                         }
2436                         if (ext4_ext_check(inode, ext_block_hdr(bh),
2437                                                         depth - i - 1)) {
2438                                 err = -EIO;
2439                                 break;
2440                         }
2441                         path[i + 1].p_bh = bh;
2442
2443                         /* save actual number of indexes since this
2444                          * number is changed at the next iteration */
2445                         path[i].p_block = le16_to_cpu(path[i].p_hdr->eh_entries);
2446                         i++;
2447                 } else {
2448                         /* we finished processing this index, go up */
2449                         if (path[i].p_hdr->eh_entries == 0 && i > 0) {
2450                                 /* index is empty, remove it;
2451                                  * handle must be already prepared by the
2452                                  * truncatei_leaf() */
2453                                 err = ext4_ext_rm_idx(handle, inode, path + i);
2454                         }
2455                         /* root level has p_bh == NULL, brelse() eats this */
2456                         brelse(path[i].p_bh);
2457                         path[i].p_bh = NULL;
2458                         i--;
2459                         ext_debug("return to level %d\n", i);
2460                 }
2461         }
2462
2463         /* TODO: flexible tree reduction should be here */
2464         if (path->p_hdr->eh_entries == 0) {
2465                 /*
2466                  * truncate to zero freed all the tree,
2467                  * so we need to correct eh_depth
2468                  */
2469                 err = ext4_ext_get_access(handle, inode, path);
2470                 if (err == 0) {
2471                         ext_inode_hdr(inode)->eh_depth = 0;
2472                         ext_inode_hdr(inode)->eh_max =
2473                                 cpu_to_le16(ext4_ext_space_root(inode, 0));
2474                         err = ext4_ext_dirty(handle, inode, path);
2475                 }
2476         }
2477 out:
2478         ext4_ext_drop_refs(path);
2479         kfree(path);
2480         if (err == -EAGAIN)
2481                 goto again;
2482         ext4_journal_stop(handle);
2483
2484         return err;
2485 }
2486
2487 /*
2488  * called at mount time
2489  */
2490 void ext4_ext_init(struct super_block *sb)
2491 {
2492         /*
2493          * possible initialization would be here
2494          */
2495
2496         if (EXT4_HAS_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_EXTENTS)) {
2497 #if defined(AGGRESSIVE_TEST) || defined(CHECK_BINSEARCH) || defined(EXTENTS_STATS)
2498                 printk(KERN_INFO "EXT4-fs: file extents enabled");
2499 #ifdef AGGRESSIVE_TEST
2500                 printk(", aggressive tests");
2501 #endif
2502 #ifdef CHECK_BINSEARCH
2503                 printk(", check binsearch");
2504 #endif
2505 #ifdef EXTENTS_STATS
2506                 printk(", stats");
2507 #endif
2508                 printk("\n");
2509 #endif
2510 #ifdef EXTENTS_STATS
2511                 spin_lock_init(&EXT4_SB(sb)->s_ext_stats_lock);
2512                 EXT4_SB(sb)->s_ext_min = 1 << 30;
2513                 EXT4_SB(sb)->s_ext_max = 0;
2514 #endif
2515         }
2516 }
2517
2518 /*
2519  * called at umount time
2520  */
2521 void ext4_ext_release(struct super_block *sb)
2522 {
2523         if (!EXT4_HAS_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_EXTENTS))
2524                 return;
2525
2526 #ifdef EXTENTS_STATS
2527         if (EXT4_SB(sb)->s_ext_blocks && EXT4_SB(sb)->s_ext_extents) {
2528                 struct ext4_sb_info *sbi = EXT4_SB(sb);
2529                 printk(KERN_ERR "EXT4-fs: %lu blocks in %lu extents (%lu ave)\n",
2530                         sbi->s_ext_blocks, sbi->s_ext_extents,
2531                         sbi->s_ext_blocks / sbi->s_ext_extents);
2532                 printk(KERN_ERR "EXT4-fs: extents: %lu min, %lu max, max depth %lu\n",
2533                         sbi->s_ext_min, sbi->s_ext_max, sbi->s_depth_max);
2534         }
2535 #endif
2536 }
2537
2538 static void bi_complete(struct bio *bio, int error)
2539 {
2540         complete((struct completion *)bio->bi_private);
2541 }
2542
2543 /* FIXME!! we need to try to merge to left or right after zero-out  */
2544 static int ext4_ext_zeroout(struct inode *inode, struct ext4_extent *ex)
2545 {
2546         int ret;
2547         struct bio *bio;
2548         int blkbits, blocksize;
2549         sector_t ee_pblock;
2550         struct completion event;
2551         unsigned int ee_len, len, done, offset;
2552
2553
2554         blkbits   = inode->i_blkbits;
2555         blocksize = inode->i_sb->s_blocksize;
2556         ee_len    = ext4_ext_get_actual_len(ex);
2557         ee_pblock = ext_pblock(ex);
2558
2559         /* convert ee_pblock to 512 byte sectors */
2560         ee_pblock = ee_pblock << (blkbits - 9);
2561
2562         while (ee_len > 0) {
2563
2564                 if (ee_len > BIO_MAX_PAGES)
2565                         len = BIO_MAX_PAGES;
2566                 else
2567                         len = ee_len;
2568
2569                 bio = bio_alloc(GFP_NOIO, len);
2570                 if (!bio)
2571                         return -ENOMEM;
2572
2573                 bio->bi_sector = ee_pblock;
2574                 bio->bi_bdev   = inode->i_sb->s_bdev;
2575
2576                 done = 0;
2577                 offset = 0;
2578                 while (done < len) {
2579                         ret = bio_add_page(bio, ZERO_PAGE(0),
2580                                                         blocksize, offset);
2581                         if (ret != blocksize) {
2582                                 /*
2583                                  * We can't add any more pages because of
2584                                  * hardware limitations.  Start a new bio.
2585                                  */
2586                                 break;
2587                         }
2588                         done++;
2589                         offset += blocksize;
2590                         if (offset >= PAGE_CACHE_SIZE)
2591                                 offset = 0;
2592                 }
2593
2594                 init_completion(&event);
2595                 bio->bi_private = &event;
2596                 bio->bi_end_io = bi_complete;
2597                 submit_bio(WRITE, bio);
2598                 wait_for_completion(&event);
2599
2600                 if (!test_bit(BIO_UPTODATE, &bio->bi_flags)) {
2601                         bio_put(bio);
2602                         return -EIO;
2603                 }
2604                 bio_put(bio);
2605                 ee_len    -= done;
2606                 ee_pblock += done  << (blkbits - 9);
2607         }
2608         return 0;
2609 }
2610
2611 #define EXT4_EXT_ZERO_LEN 7
2612 /*
2613  * This function is called by ext4_ext_map_blocks() if someone tries to write
2614  * to an uninitialized extent. It may result in splitting the uninitialized
2615  * extent into multiple extents (upto three - one initialized and two
2616  * uninitialized).
2617  * There are three possibilities:
2618  *   a> There is no split required: Entire extent should be initialized
2619  *   b> Splits in two extents: Write is happening at either end of the extent
2620  *   c> Splits in three extents: Somone is writing in middle of the extent
2621  */
2622 static int ext4_ext_convert_to_initialized(handle_t *handle,
2623                                            struct inode *inode,
2624                                            struct ext4_map_blocks *map,
2625                                            struct ext4_ext_path *path)
2626 {
2627         struct ext4_extent *ex, newex, orig_ex;
2628         struct ext4_extent *ex1 = NULL;
2629         struct ext4_extent *ex2 = NULL;
2630         struct ext4_extent *ex3 = NULL;
2631         struct ext4_extent_header *eh;
2632         ext4_lblk_t ee_block, eof_block;
2633         unsigned int allocated, ee_len, depth;
2634         ext4_fsblk_t newblock;
2635         int err = 0;
2636         int ret = 0;
2637         int may_zeroout;
2638
2639         ext_debug("ext4_ext_convert_to_initialized: inode %lu, logical"
2640                 "block %llu, max_blocks %u\n", inode->i_ino,
2641                 (unsigned long long)map->m_lblk, map->m_len);
2642
2643         eof_block = (inode->i_size + inode->i_sb->s_blocksize - 1) >>
2644                 inode->i_sb->s_blocksize_bits;
2645         if (eof_block < map->m_lblk + map->m_len)
2646                 eof_block = map->m_lblk + map->m_len;
2647
2648         depth = ext_depth(inode);
2649         eh = path[depth].p_hdr;
2650         ex = path[depth].p_ext;
2651         ee_block = le32_to_cpu(ex->ee_block);
2652         ee_len = ext4_ext_get_actual_len(ex);
2653         allocated = ee_len - (map->m_lblk - ee_block);
2654         newblock = map->m_lblk - ee_block + ext_pblock(ex);
2655
2656         ex2 = ex;
2657         orig_ex.ee_block = ex->ee_block;
2658         orig_ex.ee_len   = cpu_to_le16(ee_len);
2659         ext4_ext_store_pblock(&orig_ex, ext_pblock(ex));
2660
2661         /*
2662          * It is safe to convert extent to initialized via explicit
2663          * zeroout only if extent is fully insde i_size or new_size.
2664          */
2665         may_zeroout = ee_block + ee_len <= eof_block;
2666
2667         err = ext4_ext_get_access(handle, inode, path + depth);
2668         if (err)
2669                 goto out;
2670         /* If extent has less than 2*EXT4_EXT_ZERO_LEN zerout directly */
2671         if (ee_len <= 2*EXT4_EXT_ZERO_LEN && may_zeroout) {
2672                 err =  ext4_ext_zeroout(inode, &orig_ex);
2673                 if (err)
2674                         goto fix_extent_len;
2675                 /* update the extent length and mark as initialized */
2676                 ex->ee_block = orig_ex.ee_block;
2677                 ex->ee_len   = orig_ex.ee_len;
2678                 ext4_ext_store_pblock(ex, ext_pblock(&orig_ex));
2679                 ext4_ext_dirty(handle, inode, path + depth);
2680                 /* zeroed the full extent */
2681                 return allocated;
2682         }
2683
2684         /* ex1: ee_block to map->m_lblk - 1 : uninitialized */
2685         if (map->m_lblk > ee_block) {
2686                 ex1 = ex;
2687                 ex1->ee_len = cpu_to_le16(map->m_lblk - ee_block);
2688                 ext4_ext_mark_uninitialized(ex1);
2689                 ex2 = &newex;
2690         }
2691         /*
2692          * for sanity, update the length of the ex2 extent before
2693          * we insert ex3, if ex1 is NULL. This is to avoid temporary
2694          * overlap of blocks.
2695          */
2696         if (!ex1 && allocated > map->m_len)
2697                 ex2->ee_len = cpu_to_le16(map->m_len);
2698         /* ex3: to ee_block + ee_len : uninitialised */
2699         if (allocated > map->m_len) {
2700                 unsigned int newdepth;
2701                 /* If extent has less than EXT4_EXT_ZERO_LEN zerout directly */
2702                 if (allocated <= EXT4_EXT_ZERO_LEN && may_zeroout) {
2703                         /*
2704                          * map->m_lblk == ee_block is handled by the zerouout
2705                          * at the beginning.
2706                          * Mark first half uninitialized.
2707                          * Mark second half initialized and zero out the
2708                          * initialized extent
2709                          */
2710                         ex->ee_block = orig_ex.ee_block;
2711                         ex->ee_len   = cpu_to_le16(ee_len - allocated);
2712                         ext4_ext_mark_uninitialized(ex);
2713                         ext4_ext_store_pblock(ex, ext_pblock(&orig_ex));
2714                         ext4_ext_dirty(handle, inode, path + depth);
2715
2716                         ex3 = &newex;
2717                         ex3->ee_block = cpu_to_le32(map->m_lblk);
2718                         ext4_ext_store_pblock(ex3, newblock);
2719                         ex3->ee_len = cpu_to_le16(allocated);
2720                         err = ext4_ext_insert_extent(handle, inode, path,
2721                                                         ex3, 0);
2722                         if (err == -ENOSPC) {
2723                                 err =  ext4_ext_zeroout(inode, &orig_ex);
2724                                 if (err)
2725                                         goto fix_extent_len;
2726                                 ex->ee_block = orig_ex.ee_block;
2727                                 ex->ee_len   = orig_ex.ee_len;
2728                                 ext4_ext_store_pblock(ex, ext_pblock(&orig_ex));
2729                                 ext4_ext_dirty(handle, inode, path + depth);
2730                                 /* blocks available from map->m_lblk */
2731                                 return allocated;
2732
2733                         } else if (err)
2734                                 goto fix_extent_len;
2735
2736                         /*
2737                          * We need to zero out the second half because
2738                          * an fallocate request can update file size and
2739                          * converting the second half to initialized extent
2740                          * implies that we can leak some junk data to user
2741                          * space.
2742                          */
2743                         err =  ext4_ext_zeroout(inode, ex3);
2744                         if (err) {
2745                                 /*
2746                                  * We should actually mark the
2747                                  * second half as uninit and return error
2748                                  * Insert would have changed the extent
2749                                  */
2750                                 depth = ext_depth(inode);
2751                                 ext4_ext_drop_refs(path);
2752                                 path = ext4_ext_find_extent(inode, map->m_lblk,
2753                                                             path);
2754                                 if (IS_ERR(path)) {
2755                                         err = PTR_ERR(path);
2756                                         return err;
2757                                 }
2758                                 /* get the second half extent details */
2759                                 ex = path[depth].p_ext;
2760                                 err = ext4_ext_get_access(handle, inode,
2761                                                                 path + depth);
2762                                 if (err)
2763                                         return err;
2764                                 ext4_ext_mark_uninitialized(ex);
2765                                 ext4_ext_dirty(handle, inode, path + depth);
2766                                 return err;
2767                         }
2768
2769                         /* zeroed the second half */
2770                         return allocated;
2771                 }
2772                 ex3 = &newex;
2773                 ex3->ee_block = cpu_to_le32(map->m_lblk + map->m_len);
2774                 ext4_ext_store_pblock(ex3, newblock + map->m_len);
2775                 ex3->ee_len = cpu_to_le16(allocated - map->m_len);
2776                 ext4_ext_mark_uninitialized(ex3);
2777                 err = ext4_ext_insert_extent(handle, inode, path, ex3, 0);
2778                 if (err == -ENOSPC && may_zeroout) {
2779                         err =  ext4_ext_zeroout(inode, &orig_ex);
2780                         if (err)
2781                                 goto fix_extent_len;
2782                         /* update the extent length and mark as initialized */
2783                         ex->ee_block = orig_ex.ee_block;
2784                         ex->ee_len   = orig_ex.ee_len;
2785                         ext4_ext_store_pblock(ex, ext_pblock(&orig_ex));
2786                         ext4_ext_dirty(handle, inode, path + depth);
2787                         /* zeroed the full extent */
2788                         /* blocks available from map->m_lblk */
2789                         return allocated;
2790
2791                 } else if (err)
2792                         goto fix_extent_len;
2793                 /*
2794                  * The depth, and hence eh & ex might change
2795                  * as part of the insert above.
2796                  */
2797                 newdepth = ext_depth(inode);
2798                 /*
2799                  * update the extent length after successful insert of the
2800                  * split extent
2801                  */
2802                 ee_len -= ext4_ext_get_actual_len(ex3);
2803                 orig_ex.ee_len = cpu_to_le16(ee_len);
2804                 may_zeroout = ee_block + ee_len <= eof_block;
2805
2806                 depth = newdepth;
2807                 ext4_ext_drop_refs(path);
2808                 path = ext4_ext_find_extent(inode, map->m_lblk, path);
2809                 if (IS_ERR(path)) {
2810                         err = PTR_ERR(path);
2811                         goto out;
2812                 }
2813                 eh = path[depth].p_hdr;
2814                 ex = path[depth].p_ext;
2815                 if (ex2 != &newex)
2816                         ex2 = ex;
2817
2818                 err = ext4_ext_get_access(handle, inode, path + depth);
2819                 if (err)
2820                         goto out;
2821
2822                 allocated = map->m_len;
2823
2824                 /* If extent has less than EXT4_EXT_ZERO_LEN and we are trying
2825                  * to insert a extent in the middle zerout directly
2826                  * otherwise give the extent a chance to merge to left
2827                  */
2828                 if (le16_to_cpu(orig_ex.ee_len) <= EXT4_EXT_ZERO_LEN &&
2829                         map->m_lblk != ee_block && may_zeroout) {
2830                         err =  ext4_ext_zeroout(inode, &orig_ex);
2831                         if (err)
2832                                 goto fix_extent_len;
2833                         /* update the extent length and mark as initialized */
2834                         ex->ee_block = orig_ex.ee_block;
2835                         ex->ee_len   = orig_ex.ee_len;
2836                         ext4_ext_store_pblock(ex, ext_pblock(&orig_ex));
2837                         ext4_ext_dirty(handle, inode, path + depth);
2838                         /* zero out the first half */
2839                         /* blocks available from map->m_lblk */
2840                         return allocated;
2841                 }
2842         }
2843         /*
2844          * If there was a change of depth as part of the
2845          * insertion of ex3 above, we need to update the length
2846          * of the ex1 extent again here
2847          */
2848         if (ex1 && ex1 != ex) {
2849                 ex1 = ex;
2850                 ex1->ee_len = cpu_to_le16(map->m_lblk - ee_block);
2851                 ext4_ext_mark_uninitialized(ex1);
2852                 ex2 = &newex;
2853         }
2854         /* ex2: map->m_lblk to map->m_lblk + maxblocks-1 : initialised */
2855         ex2->ee_block = cpu_to_le32(map->m_lblk);
2856         ext4_ext_store_pblock(ex2, newblock);
2857         ex2->ee_len = cpu_to_le16(allocated);
2858         if (ex2 != ex)
2859                 goto insert;
2860         /*
2861          * New (initialized) extent starts from the first block
2862          * in the current extent. i.e., ex2 == ex
2863          * We have to see if it can be merged with the extent
2864          * on the left.
2865          */
2866         if (ex2 > EXT_FIRST_EXTENT(eh)) {
2867                 /*
2868                  * To merge left, pass "ex2 - 1" to try_to_merge(),
2869                  * since it merges towards right _only_.
2870                  */
2871                 ret = ext4_ext_try_to_merge(inode, path, ex2 - 1);
2872                 if (ret) {
2873                         err = ext4_ext_correct_indexes(handle, inode, path);
2874                         if (err)
2875                                 goto out;
2876                         depth = ext_depth(inode);
2877                         ex2--;
2878                 }
2879         }
2880         /*
2881          * Try to Merge towards right. This might be required
2882          * only when the whole extent is being written to.
2883          * i.e. ex2 == ex and ex3 == NULL.
2884          */
2885         if (!ex3) {
2886                 ret = ext4_ext_try_to_merge(inode, path, ex2);
2887                 if (ret) {
2888                         err = ext4_ext_correct_indexes(handle, inode, path);
2889                         if (err)
2890                                 goto out;
2891                 }
2892         }
2893         /* Mark modified extent as dirty */
2894         err = ext4_ext_dirty(handle, inode, path + depth);
2895         goto out;
2896 insert:
2897         err = ext4_ext_insert_extent(handle, inode, path, &newex, 0);
2898         if (err == -ENOSPC && may_zeroout) {
2899                 err =  ext4_ext_zeroout(inode, &orig_ex);
2900                 if (err)
2901                         goto fix_extent_len;
2902                 /* update the extent length and mark as initialized */
2903                 ex->ee_block = orig_ex.ee_block;
2904                 ex->ee_len   = orig_ex.ee_len;
2905                 ext4_ext_store_pblock(ex, ext_pblock(&orig_ex));
2906                 ext4_ext_dirty(handle, inode, path + depth);
2907                 /* zero out the first half */
2908                 return allocated;
2909         } else if (err)
2910                 goto fix_extent_len;
2911 out:
2912         ext4_ext_show_leaf(inode, path);
2913         return err ? err : allocated;
2914
2915 fix_extent_len:
2916         ex->ee_block = orig_ex.ee_block;
2917         ex->ee_len   = orig_ex.ee_len;
2918         ext4_ext_store_pblock(ex, ext_pblock(&orig_ex));
2919         ext4_ext_mark_uninitialized(ex);
2920         ext4_ext_dirty(handle, inode, path + depth);
2921         return err;
2922 }
2923
2924 /*
2925  * This function is called by ext4_ext_map_blocks() from
2926  * ext4_get_blocks_dio_write() when DIO to write
2927  * to an uninitialized extent.
2928  *
2929  * Writing to an uninitized extent may result in splitting the uninitialized
2930  * extent into multiple /intialized unintialized extents (up to three)
2931  * There are three possibilities:
2932  *   a> There is no split required: Entire extent should be uninitialized
2933  *   b> Splits in two extents: Write is happening at either end of the extent
2934  *   c> Splits in three extents: Somone is writing in middle of the extent
2935  *
2936  * One of more index blocks maybe needed if the extent tree grow after
2937  * the unintialized extent split. To prevent ENOSPC occur at the IO
2938  * complete, we need to split the uninitialized extent before DIO submit
2939  * the IO. The uninitialized extent called at this time will be split
2940  * into three uninitialized extent(at most). After IO complete, the part
2941  * being filled will be convert to initialized by the end_io callback function
2942  * via ext4_convert_unwritten_extents().
2943  *
2944  * Returns the size of uninitialized extent to be written on success.
2945  */
2946 static int ext4_split_unwritten_extents(handle_t *handle,
2947                                         struct inode *inode,
2948                                         struct ext4_map_blocks *map,
2949                                         struct ext4_ext_path *path,
2950                                         int flags)
2951 {
2952         struct ext4_extent *ex, newex, orig_ex;
2953         struct ext4_extent *ex1 = NULL;
2954         struct ext4_extent *ex2 = NULL;
2955         struct ext4_extent *ex3 = NULL;
2956         ext4_lblk_t ee_block, eof_block;
2957         unsigned int allocated, ee_len, depth;
2958         ext4_fsblk_t newblock;
2959         int err = 0;
2960         int may_zeroout;
2961
2962         ext_debug("ext4_split_unwritten_extents: inode %lu, logical"
2963                 "block %llu, max_blocks %u\n", inode->i_ino,
2964                 (unsigned long long)map->m_lblk, map->m_len);
2965
2966         eof_block = (inode->i_size + inode->i_sb->s_blocksize - 1) >>
2967                 inode->i_sb->s_blocksize_bits;
2968         if (eof_block < map->m_lblk + map->m_len)
2969                 eof_block = map->m_lblk + map->m_len;
2970
2971         depth = ext_depth(inode);
2972         ex = path[depth].p_ext;
2973         ee_block = le32_to_cpu(ex->ee_block);
2974         ee_len = ext4_ext_get_actual_len(ex);
2975         allocated = ee_len - (map->m_lblk - ee_block);
2976         newblock = map->m_lblk - ee_block + ext_pblock(ex);
2977
2978         ex2 = ex;
2979         orig_ex.ee_block = ex->ee_block;
2980         orig_ex.ee_len   = cpu_to_le16(ee_len);
2981         ext4_ext_store_pblock(&orig_ex, ext_pblock(ex));
2982
2983         /*
2984          * It is safe to convert extent to initialized via explicit
2985          * zeroout only if extent is fully insde i_size or new_size.
2986          */
2987         may_zeroout = ee_block + ee_len <= eof_block;
2988
2989         /*
2990          * If the uninitialized extent begins at the same logical
2991          * block where the write begins, and the write completely
2992          * covers the extent, then we don't need to split it.
2993          */
2994         if ((map->m_lblk == ee_block) && (allocated <= map->m_len))
2995                 return allocated;
2996
2997         err = ext4_ext_get_access(handle, inode, path + depth);
2998         if (err)
2999                 goto out;
3000         /* ex1: ee_block to map->m_lblk - 1 : uninitialized */
3001         if (map->m_lblk > ee_block) {
3002                 ex1 = ex;
3003                 ex1->ee_len = cpu_to_le16(map->m_lblk - ee_block);
3004                 ext4_ext_mark_uninitialized(ex1);
3005                 ex2 = &newex;
3006         }
3007         /*
3008          * for sanity, update the length of the ex2 extent before
3009          * we insert ex3, if ex1 is NULL. This is to avoid temporary
3010          * overlap of blocks.
3011          */
3012         if (!ex1 && allocated > map->m_len)
3013                 ex2->ee_len = cpu_to_le16(map->m_len);
3014         /* ex3: to ee_block + ee_len : uninitialised */
3015         if (allocated > map->m_len) {
3016                 unsigned int newdepth;
3017                 ex3 = &newex;
3018                 ex3->ee_block = cpu_to_le32(map->m_lblk + map->m_len);
3019                 ext4_ext_store_pblock(ex3, newblock + map->m_len);
3020                 ex3->ee_len = cpu_to_le16(allocated - map->m_len);
3021                 ext4_ext_mark_uninitialized(ex3);
3022                 err = ext4_ext_insert_extent(handle, inode, path, ex3, flags);
3023                 if (err == -ENOSPC && may_zeroout) {
3024                         err =  ext4_ext_zeroout(inode, &orig_ex);
3025                         if (err)
3026                                 goto fix_extent_len;
3027                         /* update the extent length and mark as initialized */
3028                         ex->ee_block = orig_ex.ee_block;
3029                         ex->ee_len   = orig_ex.ee_len;
3030                         ext4_ext_store_pblock(ex, ext_pblock(&orig_ex));
3031                         ext4_ext_dirty(handle, inode, path + depth);
3032                         /* zeroed the full extent */
3033                         /* blocks available from map->m_lblk */
3034                         return allocated;
3035
3036                 } else if (err)
3037                         goto fix_extent_len;
3038                 /*
3039                  * The depth, and hence eh & ex might change
3040                  * as part of the insert above.
3041                  */
3042                 newdepth = ext_depth(inode);
3043                 /*
3044                  * update the extent length after successful insert of the
3045                  * split extent
3046                  */
3047                 ee_len -= ext4_ext_get_actual_len(ex3);
3048                 orig_ex.ee_len = cpu_to_le16(ee_len);
3049                 may_zeroout = ee_block + ee_len <= eof_block;
3050
3051                 depth = newdepth;
3052                 ext4_ext_drop_refs(path);
3053                 path = ext4_ext_find_extent(inode, map->m_lblk, path);
3054                 if (IS_ERR(path)) {
3055                         err = PTR_ERR(path);
3056                         goto out;
3057                 }
3058                 ex = path[depth].p_ext;
3059                 if (ex2 != &newex)
3060                         ex2 = ex;
3061
3062                 err = ext4_ext_get_access(handle, inode, path + depth);
3063                 if (err)
3064                         goto out;
3065
3066                 allocated = map->m_len;
3067         }
3068         /*
3069          * If there was a change of depth as part of the
3070          * insertion of ex3 above, we need to update the length
3071          * of the ex1 extent again here
3072          */
3073         if (ex1 && ex1 != ex) {
3074                 ex1 = ex;
3075                 ex1->ee_len = cpu_to_le16(map->m_lblk - ee_block);
3076                 ext4_ext_mark_uninitialized(ex1);
3077                 ex2 = &newex;
3078         }
3079         /*
3080          * ex2: map->m_lblk to map->m_lblk + map->m_len-1 : to be written
3081          * using direct I/O, uninitialised still.
3082          */
3083         ex2->ee_block = cpu_to_le32(map->m_lblk);
3084         ext4_ext_store_pblock(ex2, newblock);
3085         ex2->ee_len = cpu_to_le16(allocated);
3086         ext4_ext_mark_uninitialized(ex2);
3087         if (ex2 != ex)
3088                 goto insert;
3089         /* Mark modified extent as dirty */
3090         err = ext4_ext_dirty(handle, inode, path + depth);
3091         ext_debug("out here\n");
3092         goto out;
3093 insert:
3094         err = ext4_ext_insert_extent(handle, inode, path, &newex, flags);
3095         if (err == -ENOSPC && may_zeroout) {
3096                 err =  ext4_ext_zeroout(inode, &orig_ex);
3097                 if (err)
3098                         goto fix_extent_len;
3099                 /* update the extent length and mark as initialized */
3100                 ex->ee_block = orig_ex.ee_block;
3101                 ex->ee_len   = orig_ex.ee_len;
3102                 ext4_ext_store_pblock(ex, ext_pblock(&orig_ex));
3103                 ext4_ext_dirty(handle, inode, path + depth);
3104                 /* zero out the first half */
3105                 return allocated;
3106         } else if (err)
3107                 goto fix_extent_len;
3108 out:
3109         ext4_ext_show_leaf(inode, path);
3110         return err ? err : allocated;
3111
3112 fix_extent_len:
3113         ex->ee_block = orig_ex.ee_block;
3114         ex->ee_len   = orig_ex.ee_len;
3115         ext4_ext_store_pblock(ex, ext_pblock(&orig_ex));
3116         ext4_ext_mark_uninitialized(ex);
3117         ext4_ext_dirty(handle, inode, path + depth);
3118         return err;
3119 }
3120 static int ext4_convert_unwritten_extents_endio(handle_t *handle,
3121                                               struct inode *inode,
3122                                               struct ext4_ext_path *path)
3123 {
3124         struct ext4_extent *ex;
3125         struct ext4_extent_header *eh;
3126         int depth;
3127         int err = 0;
3128         int ret = 0;
3129
3130         depth = ext_depth(inode);
3131         eh = path[depth].p_hdr;
3132         ex = path[depth].p_ext;
3133
3134         err = ext4_ext_get_access(handle, inode, path + depth);
3135         if (err)
3136                 goto out;
3137         /* first mark the extent as initialized */
3138         ext4_ext_mark_initialized(ex);
3139
3140         /*
3141          * We have to see if it can be merged with the extent
3142          * on the left.
3143          */
3144         if (ex > EXT_FIRST_EXTENT(eh)) {
3145                 /*
3146                  * To merge left, pass "ex - 1" to try_to_merge(),
3147                  * since it merges towards right _only_.
3148                  */
3149                 ret = ext4_ext_try_to_merge(inode, path, ex - 1);
3150                 if (ret) {
3151                         err = ext4_ext_correct_indexes(handle, inode, path);
3152                         if (err)
3153                                 goto out;
3154                         depth = ext_depth(inode);
3155                         ex--;
3156                 }
3157         }
3158         /*
3159          * Try to Merge towards right.
3160          */
3161         ret = ext4_ext_try_to_merge(inode, path, ex);
3162         if (ret) {
3163                 err = ext4_ext_correct_indexes(handle, inode, path);
3164                 if (err)
3165                         goto out;
3166                 depth = ext_depth(inode);
3167         }
3168         /* Mark modified extent as dirty */
3169         err = ext4_ext_dirty(handle, inode, path + depth);
3170 out:
3171         ext4_ext_show_leaf(inode, path);
3172         return err;
3173 }
3174
3175 static void unmap_underlying_metadata_blocks(struct block_device *bdev,
3176                         sector_t block, int count)
3177 {
3178         int i;
3179         for (i = 0; i < count; i++)
3180                 unmap_underlying_metadata(bdev, block + i);
3181 }
3182
3183 static int
3184 ext4_ext_handle_uninitialized_extents(handle_t *handle, struct inode *inode,
3185                         struct ext4_map_blocks *map,
3186                         struct ext4_ext_path *path, int flags,
3187                         unsigned int allocated, ext4_fsblk_t newblock)
3188 {
3189         int ret = 0;
3190         int err = 0;
3191         ext4_io_end_t *io = EXT4_I(inode)->cur_aio_dio;
3192
3193         ext_debug("ext4_ext_handle_uninitialized_extents: inode %lu, logical"
3194                   "block %llu, max_blocks %u, flags %d, allocated %u",
3195                   inode->i_ino, (unsigned long long)map->m_lblk, map->m_len,
3196                   flags, allocated);
3197         ext4_ext_show_leaf(inode, path);
3198
3199         /* get_block() before submit the IO, split the extent */
3200         if ((flags & EXT4_GET_BLOCKS_PRE_IO)) {
3201                 ret = ext4_split_unwritten_extents(handle, inode, map,
3202                                                    path, flags);
3203                 /*
3204                  * Flag the inode(non aio case) or end_io struct (aio case)
3205                  * that this IO needs to convertion to written when IO is
3206                  * completed
3207                  */
3208                 if (io)
3209                         io->flag = EXT4_IO_UNWRITTEN;
3210                 else
3211                         ext4_set_inode_state(inode, EXT4_STATE_DIO_UNWRITTEN);
3212                 if (ext4_should_dioread_nolock(inode))
3213                         map->m_flags |= EXT4_MAP_UNINIT;
3214                 goto out;
3215         }
3216         /* IO end_io complete, convert the filled extent to written */
3217         if ((flags & EXT4_GET_BLOCKS_CONVERT)) {
3218                 ret = ext4_convert_unwritten_extents_endio(handle, inode,
3219                                                         path);
3220                 if (ret >= 0)
3221                         ext4_update_inode_fsync_trans(handle, inode, 1);
3222                 goto out2;
3223         }
3224         /* buffered IO case */
3225         /*
3226          * repeat fallocate creation request
3227          * we already have an unwritten extent
3228          */
3229         if (flags & EXT4_GET_BLOCKS_UNINIT_EXT)
3230                 goto map_out;
3231
3232         /* buffered READ or buffered write_begin() lookup */
3233         if ((flags & EXT4_GET_BLOCKS_CREATE) == 0) {
3234                 /*
3235                  * We have blocks reserved already.  We
3236                  * return allocated blocks so that delalloc
3237                  * won't do block reservation for us.  But
3238                  * the buffer head will be unmapped so that
3239                  * a read from the block returns 0s.
3240                  */
3241                 map->m_flags |= EXT4_MAP_UNWRITTEN;
3242                 goto out1;
3243         }
3244
3245         /* buffered write, writepage time, convert*/
3246         ret = ext4_ext_convert_to_initialized(handle, inode, map, path);
3247         if (ret >= 0)
3248                 ext4_update_inode_fsync_trans(handle, inode, 1);
3249 out:
3250         if (ret <= 0) {
3251                 err = ret;
3252                 goto out2;
3253         } else
3254                 allocated = ret;
3255         map->m_flags |= EXT4_MAP_NEW;
3256         /*
3257          * if we allocated more blocks than requested
3258          * we need to make sure we unmap the extra block
3259          * allocated. The actual needed block will get
3260          * unmapped later when we find the buffer_head marked
3261          * new.
3262          */
3263         if (allocated > map->m_len) {
3264                 unmap_underlying_metadata_blocks(inode->i_sb->s_bdev,
3265                                         newblock + map->m_len,
3266                                         allocated - map->m_len);
3267                 allocated = map->m_len;
3268         }
3269
3270         /*
3271          * If we have done fallocate with the offset that is already
3272          * delayed allocated, we would have block reservation
3273          * and quota reservation done in the delayed write path.
3274          * But fallocate would have already updated quota and block
3275          * count for this offset. So cancel these reservation
3276          */
3277         if (flags & EXT4_GET_BLOCKS_DELALLOC_RESERVE)
3278                 ext4_da_update_reserve_space(inode, allocated, 0);
3279
3280 map_out:
3281         map->m_flags |= EXT4_MAP_MAPPED;
3282 out1:
3283         if (allocated > map->m_len)
3284                 allocated = map->m_len;
3285         ext4_ext_show_leaf(inode, path);
3286         map->m_pblk = newblock;
3287         map->m_len = allocated;
3288 out2:
3289         if (path) {
3290                 ext4_ext_drop_refs(path);
3291                 kfree(path);
3292         }
3293         return err ? err : allocated;
3294 }
3295 /*
3296  * Block allocation/map/preallocation routine for extents based files
3297  *
3298  *
3299  * Need to be called with
3300  * down_read(&EXT4_I(inode)->i_data_sem) if not allocating file system block
3301  * (ie, create is zero). Otherwise down_write(&EXT4_I(inode)->i_data_sem)
3302  *
3303  * return > 0, number of of blocks already mapped/allocated
3304  *          if create == 0 and these are pre-allocated blocks
3305  *              buffer head is unmapped
3306  *          otherwise blocks are mapped
3307  *
3308  * return = 0, if plain look up failed (blocks have not been allocated)
3309  *          buffer head is unmapped
3310  *
3311  * return < 0, error case.
3312  */
3313 int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
3314                         struct ext4_map_blocks *map, int flags)
3315 {
3316         struct ext4_ext_path *path = NULL;
3317         struct ext4_extent_header *eh;
3318         struct ext4_extent newex, *ex, *last_ex;
3319         ext4_fsblk_t newblock;
3320         int i, err = 0, depth, ret, cache_type;
3321         unsigned int allocated = 0;
3322         struct ext4_allocation_request ar;
3323         ext4_io_end_t *io = EXT4_I(inode)->cur_aio_dio;
3324
3325         ext_debug("blocks %u/%u requested for inode %lu\n",
3326                   map->m_lblk, map->m_len, inode->i_ino);
3327
3328         /* check in cache */
3329         cache_type = ext4_ext_in_cache(inode, map->m_lblk, &newex);
3330         if (cache_type) {
3331                 if (cache_type == EXT4_EXT_CACHE_GAP) {
3332                         if ((flags & EXT4_GET_BLOCKS_CREATE) == 0) {
3333                                 /*
3334                                  * block isn't allocated yet and
3335                                  * user doesn't want to allocate it
3336                                  */
3337                                 goto out2;
3338                         }
3339                         /* we should allocate requested block */
3340                 } else if (cache_type == EXT4_EXT_CACHE_EXTENT) {
3341                         /* block is already allocated */
3342                         newblock = map->m_lblk
3343                                    - le32_to_cpu(newex.ee_block)
3344                                    + ext_pblock(&newex);
3345                         /* number of remaining blocks in the extent */
3346                         allocated = ext4_ext_get_actual_len(&newex) -
3347                                 (map->m_lblk - le32_to_cpu(newex.ee_block));
3348                         goto out;
3349                 } else {
3350                         BUG();
3351                 }
3352         }
3353
3354         /* find extent for this block */
3355         path = ext4_ext_find_extent(inode, map->m_lblk, NULL);
3356         if (IS_ERR(path)) {
3357                 err = PTR_ERR(path);
3358                 path = NULL;
3359                 goto out2;
3360         }
3361
3362         depth = ext_depth(inode);
3363
3364         /*
3365          * consistent leaf must not be empty;
3366          * this situation is possible, though, _during_ tree modification;
3367          * this is why assert can't be put in ext4_ext_find_extent()
3368          */
3369         if (unlikely(path[depth].p_ext == NULL && depth != 0)) {
3370                 EXT4_ERROR_INODE(inode, "bad extent address "
3371                                  "lblock: %lu, depth: %d pblock %lld",
3372                                  (unsigned long) map->m_lblk, depth,
3373                                  path[depth].p_block);
3374                 err = -EIO;
3375                 goto out2;
3376         }
3377         eh = path[depth].p_hdr;
3378
3379         ex = path[depth].p_ext;
3380         if (ex) {
3381                 ext4_lblk_t ee_block = le32_to_cpu(ex->ee_block);
3382                 ext4_fsblk_t ee_start = ext_pblock(ex);
3383                 unsigned short ee_len;
3384
3385                 /*
3386                  * Uninitialized extents are treated as holes, except that
3387                  * we split out initialized portions during a write.
3388                  */
3389                 ee_len = ext4_ext_get_actual_len(ex);
3390                 /* if found extent covers block, simply return it */
3391                 if (in_range(map->m_lblk, ee_block, ee_len)) {
3392                         newblock = map->m_lblk - ee_block + ee_start;
3393                         /* number of remaining blocks in the extent */
3394                         allocated = ee_len - (map->m_lblk - ee_block);
3395                         ext_debug("%u fit into %u:%d -> %llu\n", map->m_lblk,
3396                                   ee_block, ee_len, newblock);
3397
3398                         /* Do not put uninitialized extent in the cache */
3399                         if (!ext4_ext_is_uninitialized(ex)) {
3400                                 ext4_ext_put_in_cache(inode, ee_block,
3401                                                         ee_len, ee_start,
3402                                                         EXT4_EXT_CACHE_EXTENT);
3403                                 goto out;
3404                         }
3405                         ret = ext4_ext_handle_uninitialized_extents(handle,
3406                                         inode, map, path, flags, allocated,
3407                                         newblock);
3408                         return ret;
3409                 }
3410         }
3411
3412         /*
3413          * requested block isn't allocated yet;
3414          * we couldn't try to create block if create flag is zero
3415          */
3416         if ((flags & EXT4_GET_BLOCKS_CREATE) == 0) {
3417                 /*
3418                  * put just found gap into cache to speed up
3419                  * subsequent requests
3420                  */
3421                 ext4_ext_put_gap_in_cache(inode, path, map->m_lblk);
3422                 goto out2;
3423         }
3424         /*
3425          * Okay, we need to do block allocation.
3426          */
3427
3428         /* find neighbour allocated blocks */
3429         ar.lleft = map->m_lblk;
3430         err = ext4_ext_search_left(inode, path, &ar.lleft, &ar.pleft);
3431         if (err)
3432                 goto out2;
3433         ar.lright = map->m_lblk;
3434         err = ext4_ext_search_right(inode, path, &ar.lright, &ar.pright);
3435         if (err)
3436                 goto out2;
3437
3438         /*
3439          * See if request is beyond maximum number of blocks we can have in
3440          * a single extent. For an initialized extent this limit is
3441          * EXT_INIT_MAX_LEN and for an uninitialized extent this limit is
3442          * EXT_UNINIT_MAX_LEN.
3443          */
3444         if (map->m_len > EXT_INIT_MAX_LEN &&
3445             !(flags & EXT4_GET_BLOCKS_UNINIT_EXT))
3446                 map->m_len = EXT_INIT_MAX_LEN;
3447         else if (map->m_len > EXT_UNINIT_MAX_LEN &&
3448                  (flags & EXT4_GET_BLOCKS_UNINIT_EXT))
3449                 map->m_len = EXT_UNINIT_MAX_LEN;
3450
3451         /* Check if we can really insert (m_lblk)::(m_lblk + m_len) extent */
3452         newex.ee_block = cpu_to_le32(map->m_lblk);
3453         newex.ee_len = cpu_to_le16(map->m_len);
3454         err = ext4_ext_check_overlap(inode, &newex, path);
3455         if (err)
3456                 allocated = ext4_ext_get_actual_len(&newex);
3457         else
3458                 allocated = map->m_len;
3459
3460         /* allocate new block */
3461         ar.inode = inode;
3462         ar.goal = ext4_ext_find_goal(inode, path, map->m_lblk);
3463         ar.logical = map->m_lblk;
3464         ar.len = allocated;
3465         if (S_ISREG(inode->i_mode))
3466                 ar.flags = EXT4_MB_HINT_DATA;
3467         else
3468                 /* disable in-core preallocation for non-regular files */
3469                 ar.flags = 0;
3470         newblock = ext4_mb_new_blocks(handle, &ar, &err);
3471         if (!newblock)
3472                 goto out2;
3473         ext_debug("allocate new block: goal %llu, found %llu/%u\n",
3474                   ar.goal, newblock, allocated);
3475
3476         /* try to insert new extent into found leaf and return */
3477         ext4_ext_store_pblock(&newex, newblock);
3478         newex.ee_len = cpu_to_le16(ar.len);
3479         /* Mark uninitialized */
3480         if (flags & EXT4_GET_BLOCKS_UNINIT_EXT){
3481                 ext4_ext_mark_uninitialized(&newex);
3482                 /*
3483                  * io_end structure was created for every IO write to an
3484                  * uninitialized extent. To avoid unecessary conversion,
3485                  * here we flag the IO that really needs the conversion.
3486                  * For non asycn direct IO case, flag the inode state
3487                  * that we need to perform convertion when IO is done.
3488                  */
3489                 if ((flags & EXT4_GET_BLOCKS_PRE_IO)) {
3490                         if (io)
3491                                 io->flag = EXT4_IO_UNWRITTEN;
3492                         else
3493                                 ext4_set_inode_state(inode,
3494                                                      EXT4_STATE_DIO_UNWRITTEN);
3495                 }
3496                 if (ext4_should_dioread_nolock(inode))
3497                         map->m_flags |= EXT4_MAP_UNINIT;
3498         }
3499
3500         if (unlikely(ext4_test_inode_flag(inode, EXT4_INODE_EOFBLOCKS))) {
3501                 if (unlikely(!eh->eh_entries)) {
3502                         EXT4_ERROR_INODE(inode,
3503                                          "eh->eh_entries == 0 and "
3504                                          "EOFBLOCKS_FL set");
3505                         err = -EIO;
3506                         goto out2;
3507                 }
3508                 last_ex = EXT_LAST_EXTENT(eh);
3509                 /*
3510                  * If the current leaf block was reached by looking at
3511                  * the last index block all the way down the tree, and
3512                  * we are extending the inode beyond the last extent
3513                  * in the current leaf block, then clear the
3514                  * EOFBLOCKS_FL flag.
3515                  */
3516                 for (i = depth-1; i >= 0; i--) {
3517                         if (path[i].p_idx != EXT_LAST_INDEX(path[i].p_hdr))
3518                                 break;
3519                 }
3520                 if ((i < 0) &&
3521                     (map->m_lblk + ar.len > le32_to_cpu(last_ex->ee_block) +
3522                      ext4_ext_get_actual_len(last_ex)))
3523                         ext4_clear_inode_flag(inode, EXT4_INODE_EOFBLOCKS);
3524         }
3525         err = ext4_ext_insert_extent(handle, inode, path, &newex, flags);
3526         if (err) {
3527                 /* free data blocks we just allocated */
3528                 /* not a good idea to call discard here directly,
3529                  * but otherwise we'd need to call it every free() */
3530                 ext4_discard_preallocations(inode);
3531                 ext4_free_blocks(handle, inode, 0, ext_pblock(&newex),
3532                                  ext4_ext_get_actual_len(&newex), 0);
3533                 goto out2;
3534         }
3535
3536         /* previous routine could use block we allocated */
3537         newblock = ext_pblock(&newex);
3538         allocated = ext4_ext_get_actual_len(&newex);
3539         if (allocated > map->m_len)
3540                 allocated = map->m_len;
3541         map->m_flags |= EXT4_MAP_NEW;
3542
3543         /*
3544          * Update reserved blocks/metadata blocks after successful
3545          * block allocation which had been deferred till now.
3546          */
3547         if (flags & EXT4_GET_BLOCKS_DELALLOC_RESERVE)
3548                 ext4_da_update_reserve_space(inode, allocated, 1);
3549
3550         /*
3551          * Cache the extent and update transaction to commit on fdatasync only
3552          * when it is _not_ an uninitialized extent.
3553          */
3554         if ((flags & EXT4_GET_BLOCKS_UNINIT_EXT) == 0) {
3555                 ext4_ext_put_in_cache(inode, map->m_lblk, allocated, newblock,
3556                                                 EXT4_EXT_CACHE_EXTENT);
3557                 ext4_update_inode_fsync_trans(handle, inode, 1);
3558         } else
3559                 ext4_update_inode_fsync_trans(handle, inode, 0);
3560 out:
3561         if (allocated > map->m_len)
3562                 allocated = map->m_len;
3563         ext4_ext_show_leaf(inode, path);
3564         map->m_flags |= EXT4_MAP_MAPPED;
3565         map->m_pblk = newblock;
3566         map->m_len = allocated;
3567 out2:
3568         if (path) {
3569                 ext4_ext_drop_refs(path);
3570                 kfree(path);
3571         }
3572         return err ? err : allocated;
3573 }
3574
3575 void ext4_ext_truncate(struct inode *inode)
3576 {
3577         struct address_space *mapping = inode->i_mapping;
3578         struct super_block *sb = inode->i_sb;
3579         ext4_lblk_t last_block;
3580         handle_t *handle;
3581         int err = 0;
3582
3583         /*
3584          * probably first extent we're gonna free will be last in block
3585          */
3586         err = ext4_writepage_trans_blocks(inode);
3587         handle = ext4_journal_start(inode, err);
3588         if (IS_ERR(handle))
3589                 return;
3590
3591         if (inode->i_size & (sb->s_blocksize - 1))
3592                 ext4_block_truncate_page(handle, mapping, inode->i_size);
3593
3594         if (ext4_orphan_add(handle, inode))
3595                 goto out_stop;
3596
3597         down_write(&EXT4_I(inode)->i_data_sem);
3598         ext4_ext_invalidate_cache(inode);
3599
3600         ext4_discard_preallocations(inode);
3601
3602         /*
3603          * TODO: optimization is possible here.
3604          * Probably we need not scan at all,
3605          * because page truncation is enough.
3606          */
3607
3608         /* we have to know where to truncate from in crash case */
3609         EXT4_I(inode)->i_disksize = inode->i_size;
3610         ext4_mark_inode_dirty(handle, inode);
3611
3612         last_block = (inode->i_size + sb->s_blocksize - 1)
3613                         >> EXT4_BLOCK_SIZE_BITS(sb);
3614         err = ext4_ext_remove_space(inode, last_block);
3615
3616         /* In a multi-transaction truncate, we only make the final
3617          * transaction synchronous.
3618          */
3619         if (IS_SYNC(inode))
3620                 ext4_handle_sync(handle);
3621
3622 out_stop:
3623         up_write(&EXT4_I(inode)->i_data_sem);
3624         /*
3625          * If this was a simple ftruncate() and the file will remain alive,
3626          * then we need to clear up the orphan record which we created above.
3627          * However, if this was a real unlink then we were called by
3628          * ext4_delete_inode(), and we allow that function to clean up the
3629          * orphan info for us.
3630          */
3631         if (inode->i_nlink)
3632                 ext4_orphan_del(handle, inode);
3633
3634         inode->i_mtime = inode->i_ctime = ext4_current_time(inode);
3635         ext4_mark_inode_dirty(handle, inode);
3636         ext4_journal_stop(handle);
3637 }
3638
3639 static void ext4_falloc_update_inode(struct inode *inode,
3640                                 int mode, loff_t new_size, int update_ctime)
3641 {
3642         struct timespec now;
3643
3644         if (update_ctime) {
3645                 now = current_fs_time(inode->i_sb);
3646                 if (!timespec_equal(&inode->i_ctime, &now))
3647                         inode->i_ctime = now;
3648         }
3649         /*
3650          * Update only when preallocation was requested beyond
3651          * the file size.
3652          */
3653         if (!(mode & FALLOC_FL_KEEP_SIZE)) {
3654                 if (new_size > i_size_read(inode))
3655                         i_size_write(inode, new_size);
3656                 if (new_size > EXT4_I(inode)->i_disksize)
3657                         ext4_update_i_disksize(inode, new_size);
3658         } else {
3659                 /*
3660                  * Mark that we allocate beyond EOF so the subsequent truncate
3661                  * can proceed even if the new size is the same as i_size.
3662                  */
3663                 if (new_size > i_size_read(inode))
3664                         ext4_set_inode_flag(inode, EXT4_INODE_EOFBLOCKS);
3665         }
3666
3667 }
3668
3669 /*
3670  * preallocate space for a file. This implements ext4's fallocate inode
3671  * operation, which gets called from sys_fallocate system call.
3672  * For block-mapped files, posix_fallocate should fall back to the method
3673  * of writing zeroes to the required new blocks (the same behavior which is
3674  * expected for file systems which do not support fallocate() system call).
3675  */
3676 long ext4_fallocate(struct inode *inode, int mode, loff_t offset, loff_t len)
3677 {
3678         handle_t *handle;
3679         loff_t new_size;
3680         unsigned int max_blocks;
3681         int ret = 0;
3682         int ret2 = 0;
3683         int retries = 0;
3684         struct ext4_map_blocks map;
3685         unsigned int credits, blkbits = inode->i_blkbits;
3686
3687         /*
3688          * currently supporting (pre)allocate mode for extent-based
3689          * files _only_
3690          */
3691         if (!(ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS)))
3692                 return -EOPNOTSUPP;
3693
3694         /* preallocation to directories is currently not supported */
3695         if (S_ISDIR(inode->i_mode))
3696                 return -ENODEV;
3697
3698         map.m_lblk = offset >> blkbits;
3699         /*
3700          * We can't just convert len to max_blocks because
3701          * If blocksize = 4096 offset = 3072 and len = 2048
3702          */
3703         max_blocks = (EXT4_BLOCK_ALIGN(len + offset, blkbits) >> blkbits)
3704                 - map.m_lblk;
3705         /*
3706          * credits to insert 1 extent into extent tree
3707          */
3708         credits = ext4_chunk_trans_blocks(inode, max_blocks);
3709         mutex_lock(&inode->i_mutex);
3710         ret = inode_newsize_ok(inode, (len + offset));
3711         if (ret) {
3712                 mutex_unlock(&inode->i_mutex);
3713                 return ret;
3714         }
3715 retry:
3716         while (ret >= 0 && ret < max_blocks) {
3717                 map.m_lblk = map.m_lblk + ret;
3718                 map.m_len = max_blocks = max_blocks - ret;
3719                 handle = ext4_journal_start(inode, credits);
3720                 if (IS_ERR(handle)) {
3721                         ret = PTR_ERR(handle);
3722                         break;
3723                 }
3724                 ret = ext4_map_blocks(handle, inode, &map,
3725                                       EXT4_GET_BLOCKS_CREATE_UNINIT_EXT);
3726                 if (ret <= 0) {
3727 #ifdef EXT4FS_DEBUG
3728                         WARN_ON(ret <= 0);
3729                         printk(KERN_ERR "%s: ext4_ext_map_blocks "
3730                                     "returned error inode#%lu, block=%u, "
3731                                     "max_blocks=%u", __func__,
3732                                     inode->i_ino, block, max_blocks);
3733 #endif
3734                         ext4_mark_inode_dirty(handle, inode);
3735                         ret2 = ext4_journal_stop(handle);
3736                         break;
3737                 }
3738                 if ((map.m_lblk + ret) >= (EXT4_BLOCK_ALIGN(offset + len,
3739                                                 blkbits) >> blkbits))
3740                         new_size = offset + len;
3741                 else
3742                         new_size = (map.m_lblk + ret) << blkbits;
3743
3744                 ext4_falloc_update_inode(inode, mode, new_size,
3745                                          (map.m_flags & EXT4_MAP_NEW));
3746                 ext4_mark_inode_dirty(handle, inode);
3747                 ret2 = ext4_journal_stop(handle);
3748                 if (ret2)
3749                         break;
3750         }
3751         if (ret == -ENOSPC &&
3752                         ext4_should_retry_alloc(inode->i_sb, &retries)) {
3753                 ret = 0;
3754                 goto retry;
3755         }
3756         mutex_unlock(&inode->i_mutex);
3757         return ret > 0 ? ret2 : ret;
3758 }
3759
3760 /*
3761  * This function convert a range of blocks to written extents
3762  * The caller of this function will pass the start offset and the size.
3763  * all unwritten extents within this range will be converted to
3764  * written extents.
3765  *
3766  * This function is called from the direct IO end io call back
3767  * function, to convert the fallocated extents after IO is completed.
3768  * Returns 0 on success.
3769  */
3770 int ext4_convert_unwritten_extents(struct inode *inode, loff_t offset,
3771                                     ssize_t len)
3772 {
3773         handle_t *handle;
3774         unsigned int max_blocks;
3775         int ret = 0;
3776         int ret2 = 0;
3777         struct ext4_map_blocks map;
3778         unsigned int credits, blkbits = inode->i_blkbits;
3779
3780         map.m_lblk = offset >> blkbits;
3781         /*
3782          * We can't just convert len to max_blocks because
3783          * If blocksize = 4096 offset = 3072 and len = 2048
3784          */
3785         max_blocks = ((EXT4_BLOCK_ALIGN(len + offset, blkbits) >> blkbits) -
3786                       map.m_lblk);
3787         /*
3788          * credits to insert 1 extent into extent tree
3789          */
3790         credits = ext4_chunk_trans_blocks(inode, max_blocks);
3791         while (ret >= 0 && ret < max_blocks) {
3792                 map.m_lblk += ret;
3793                 map.m_len = (max_blocks -= ret);
3794                 handle = ext4_journal_start(inode, credits);
3795                 if (IS_ERR(handle)) {
3796                         ret = PTR_ERR(handle);
3797                         break;
3798                 }
3799                 ret = ext4_map_blocks(handle, inode, &map,
3800                                       EXT4_GET_BLOCKS_IO_CONVERT_EXT);
3801                 if (ret <= 0) {
3802                         WARN_ON(ret <= 0);
3803                         printk(KERN_ERR "%s: ext4_ext_map_blocks "
3804                                     "returned error inode#%lu, block=%u, "
3805                                     "max_blocks=%u", __func__,
3806                                     inode->i_ino, map.m_lblk, map.m_len);
3807                 }
3808                 ext4_mark_inode_dirty(handle, inode);
3809                 ret2 = ext4_journal_stop(handle);
3810                 if (ret <= 0 || ret2 )
3811                         break;
3812         }
3813         return ret > 0 ? ret2 : ret;
3814 }
3815 /*
3816  * Callback function called for each extent to gather FIEMAP information.
3817  */
3818 static int ext4_ext_fiemap_cb(struct inode *inode, struct ext4_ext_path *path,
3819                        struct ext4_ext_cache *newex, struct ext4_extent *ex,
3820                        void *data)
3821 {
3822         struct fiemap_extent_info *fieinfo = data;
3823         unsigned char blksize_bits = inode->i_sb->s_blocksize_bits;
3824         __u64   logical;
3825         __u64   physical;
3826         __u64   length;
3827         __u32   flags = 0;
3828         int     error;
3829
3830         logical =  (__u64)newex->ec_block << blksize_bits;
3831
3832         if (newex->ec_type == EXT4_EXT_CACHE_GAP) {
3833                 pgoff_t offset;
3834                 struct page *page;
3835                 struct buffer_head *bh = NULL;
3836
3837                 offset = logical >> PAGE_SHIFT;
3838                 page = find_get_page(inode->i_mapping, offset);
3839                 if (!page || !page_has_buffers(page))
3840                         return EXT_CONTINUE;
3841
3842                 bh = page_buffers(page);
3843
3844                 if (!bh)
3845                         return EXT_CONTINUE;
3846
3847                 if (buffer_delay(bh)) {
3848                         flags |= FIEMAP_EXTENT_DELALLOC;
3849                         page_cache_release(page);
3850                 } else {
3851                         page_cache_release(page);
3852                         return EXT_CONTINUE;
3853                 }
3854         }
3855
3856         physical = (__u64)newex->ec_start << blksize_bits;
3857         length =   (__u64)newex->ec_len << blksize_bits;
3858
3859         if (ex && ext4_ext_is_uninitialized(ex))
3860                 flags |= FIEMAP_EXTENT_UNWRITTEN;
3861
3862         /*
3863          * If this extent reaches EXT_MAX_BLOCK, it must be last.
3864          *
3865          * Or if ext4_ext_next_allocated_block is EXT_MAX_BLOCK,
3866          * this also indicates no more allocated blocks.
3867          *
3868          * XXX this might miss a single-block extent at EXT_MAX_BLOCK
3869          */
3870         if (ext4_ext_next_allocated_block(path) == EXT_MAX_BLOCK ||
3871             newex->ec_block + newex->ec_len - 1 == EXT_MAX_BLOCK) {
3872                 loff_t size = i_size_read(inode);
3873                 loff_t bs = EXT4_BLOCK_SIZE(inode->i_sb);
3874
3875                 flags |= FIEMAP_EXTENT_LAST;
3876                 if ((flags & FIEMAP_EXTENT_DELALLOC) &&
3877                     logical+length > size)
3878                         length = (size - logical + bs - 1) & ~(bs-1);
3879         }
3880
3881         error = fiemap_fill_next_extent(fieinfo, logical, physical,
3882                                         length, flags);
3883         if (error < 0)
3884                 return error;
3885         if (error == 1)
3886                 return EXT_BREAK;
3887
3888         return EXT_CONTINUE;
3889 }
3890
3891 /* fiemap flags we can handle specified here */
3892 #define EXT4_FIEMAP_FLAGS       (FIEMAP_FLAG_SYNC|FIEMAP_FLAG_XATTR)
3893
3894 static int ext4_xattr_fiemap(struct inode *inode,
3895                                 struct fiemap_extent_info *fieinfo)
3896 {
3897         __u64 physical = 0;
3898         __u64 length;
3899         __u32 flags = FIEMAP_EXTENT_LAST;
3900         int blockbits = inode->i_sb->s_blocksize_bits;
3901         int error = 0;
3902
3903         /* in-inode? */
3904         if (ext4_test_inode_state(inode, EXT4_STATE_XATTR)) {
3905                 struct ext4_iloc iloc;
3906                 int offset;     /* offset of xattr in inode */
3907
3908                 error = ext4_get_inode_loc(inode, &iloc);
3909                 if (error)
3910                         return error;
3911                 physical = iloc.bh->b_blocknr << blockbits;
3912                 offset = EXT4_GOOD_OLD_INODE_SIZE +
3913                                 EXT4_I(inode)->i_extra_isize;
3914                 physical += offset;
3915                 length = EXT4_SB(inode->i_sb)->s_inode_size - offset;
3916                 flags |= FIEMAP_EXTENT_DATA_INLINE;
3917                 brelse(iloc.bh);
3918         } else { /* external block */
3919                 physical = EXT4_I(inode)->i_file_acl << blockbits;
3920                 length = inode->i_sb->s_blocksize;
3921         }
3922
3923         if (physical)
3924                 error = fiemap_fill_next_extent(fieinfo, 0, physical,
3925                                                 length, flags);
3926         return (error < 0 ? error : 0);
3927 }
3928
3929 int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
3930                 __u64 start, __u64 len)
3931 {
3932         ext4_lblk_t start_blk;
3933         int error = 0;
3934
3935         /* fallback to generic here if not in extents fmt */
3936         if (!(ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS)))
3937                 return generic_block_fiemap(inode, fieinfo, start, len,
3938                         ext4_get_block);
3939
3940         if (fiemap_check_flags(fieinfo, EXT4_FIEMAP_FLAGS))
3941                 return -EBADR;
3942
3943         if (fieinfo->fi_flags & FIEMAP_FLAG_XATTR) {
3944                 error = ext4_xattr_fiemap(inode, fieinfo);
3945         } else {
3946                 ext4_lblk_t len_blks;
3947                 __u64 last_blk;
3948
3949                 start_blk = start >> inode->i_sb->s_blocksize_bits;
3950                 last_blk = (start + len - 1) >> inode->i_sb->s_blocksize_bits;
3951                 if (last_blk >= EXT_MAX_BLOCK)
3952                         last_blk = EXT_MAX_BLOCK-1;
3953                 len_blks = ((ext4_lblk_t) last_blk) - start_blk + 1;
3954
3955                 /*
3956                  * Walk the extent tree gathering extent information.
3957                  * ext4_ext_fiemap_cb will push extents back to user.
3958                  */
3959                 error = ext4_ext_walk_space(inode, start_blk, len_blks,
3960                                           ext4_ext_fiemap_cb, fieinfo);
3961         }
3962
3963         return error;
3964 }
3965