fs/xfs/libxfs/xfs_refcount.c

   1 /*
   2  * Copyright (C) 2016 Oracle.  All Rights Reserved.
   3  *
   4  * Author: Darrick J. Wong <darrick.wong@oracle.com>
   5  *
   6  * This program is free software; you can redistribute it and/or
   7  * modify it under the terms of the GNU General Public License
   8  * as published by the Free Software Foundation; either version 2
   9  * of the License, or (at your option) any later version.
  10  *
  11  * This program is distributed in the hope that it would be useful,
  12  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  13  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  14  * GNU General Public License for more details.
  15  *
  16  * You should have received a copy of the GNU General Public License
  17  * along with this program; if not, write the Free Software Foundation,
  18  * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301, USA.
  19  */
  20 #include "xfs.h"
  21 #include "xfs_fs.h"
  22 #include "xfs_shared.h"
  23 #include "xfs_format.h"
  24 #include "xfs_log_format.h"
  25 #include "xfs_trans_resv.h"
  26 #include "xfs_sb.h"
  27 #include "xfs_mount.h"
  28 #include "xfs_defer.h"
  29 #include "xfs_btree.h"
  30 #include "xfs_bmap.h"
  31 #include "xfs_refcount_btree.h"
  32 #include "xfs_alloc.h"
  33 #include "xfs_errortag.h"
  34 #include "xfs_error.h"
  35 #include "xfs_trace.h"
  36 #include "xfs_cksum.h"
  37 #include "xfs_trans.h"
  38 #include "xfs_bit.h"
  39 #include "xfs_refcount.h"
  40 #include "xfs_rmap.h"
  41
  42 /* Allowable refcount adjustment amounts. */
  43 enum xfs_refc_adjust_op {
  44         XFS_REFCOUNT_ADJUST_INCREASE    = 1,
  45         XFS_REFCOUNT_ADJUST_DECREASE    = -1,
  46         XFS_REFCOUNT_ADJUST_COW_ALLOC   = 0,
  47         XFS_REFCOUNT_ADJUST_COW_FREE    = -1,
  48 };
  49
  50 STATIC int __xfs_refcount_cow_alloc(struct xfs_btree_cur *rcur,
  51                 xfs_agblock_t agbno, xfs_extlen_t aglen,
  52                 struct xfs_defer_ops *dfops);
  53 STATIC int __xfs_refcount_cow_free(struct xfs_btree_cur *rcur,
  54                 xfs_agblock_t agbno, xfs_extlen_t aglen,
  55                 struct xfs_defer_ops *dfops);
  56
  57 /*
  58  * Look up the first record less than or equal to [bno, len] in the btree
  59  * given by cur.
  60  */
  61 int
  62 xfs_refcount_lookup_le(
  63         struct xfs_btree_cur    *cur,
  64         xfs_agblock_t           bno,
  65         int                     *stat)
  66 {
  67         trace_xfs_refcount_lookup(cur->bc_mp, cur->bc_private.a.agno, bno,
  68                         XFS_LOOKUP_LE);
  69         cur->bc_rec.rc.rc_startblock = bno;
  70         cur->bc_rec.rc.rc_blockcount = 0;
  71         return xfs_btree_lookup(cur, XFS_LOOKUP_LE, stat);
  72 }
  73
  74 /*
  75  * Look up the first record greater than or equal to [bno, len] in the btree
  76  * given by cur.
  77  */
  78 int
  79 xfs_refcount_lookup_ge(
  80         struct xfs_btree_cur    *cur,
  81         xfs_agblock_t           bno,
  82         int                     *stat)
  83 {
  84         trace_xfs_refcount_lookup(cur->bc_mp, cur->bc_private.a.agno, bno,
  85                         XFS_LOOKUP_GE);
  86         cur->bc_rec.rc.rc_startblock = bno;
  87         cur->bc_rec.rc.rc_blockcount = 0;
  88         return xfs_btree_lookup(cur, XFS_LOOKUP_GE, stat);
  89 }
  90
  91 /* Convert on-disk record to in-core format. */
  92 static inline void
  93 xfs_refcount_btrec_to_irec(
  94         union xfs_btree_rec             *rec,
  95         struct xfs_refcount_irec        *irec)
  96 {
  97         irec->rc_startblock = be32_to_cpu(rec->refc.rc_startblock);
  98         irec->rc_blockcount = be32_to_cpu(rec->refc.rc_blockcount);
  99         irec->rc_refcount = be32_to_cpu(rec->refc.rc_refcount);
 100 }
 101
 102 /*
 103  * Get the data from the pointed-to record.
 104  */
 105 int
 106 xfs_refcount_get_rec(
 107         struct xfs_btree_cur            *cur,
 108         struct xfs_refcount_irec        *irec,
 109         int                             *stat)
 110 {
 111         union xfs_btree_rec             *rec;
 112         int                             error;
 113
 114         error = xfs_btree_get_rec(cur, &rec, stat);
 115         if (!error && *stat == 1) {
 116                 xfs_refcount_btrec_to_irec(rec, irec);
 117                 trace_xfs_refcount_get(cur->bc_mp, cur->bc_private.a.agno,
 118                                 irec);
 119         }
 120         return error;
 121 }
 122
 123 /*
 124  * Update the record referred to by cur to the value given
 125  * by [bno, len, refcount].
 126  * This either works (return 0) or gets an EFSCORRUPTED error.
 127  */
 128 STATIC int
 129 xfs_refcount_update(
 130         struct xfs_btree_cur            *cur,
 131         struct xfs_refcount_irec        *irec)
 132 {
 133         union xfs_btree_rec     rec;
 134         int                     error;
 135
 136         trace_xfs_refcount_update(cur->bc_mp, cur->bc_private.a.agno, irec);
 137         rec.refc.rc_startblock = cpu_to_be32(irec->rc_startblock);
 138         rec.refc.rc_blockcount = cpu_to_be32(irec->rc_blockcount);
 139         rec.refc.rc_refcount = cpu_to_be32(irec->rc_refcount);
 140         error = xfs_btree_update(cur, &rec);
 141         if (error)
 142                 trace_xfs_refcount_update_error(cur->bc_mp,
 143                                 cur->bc_private.a.agno, error, _RET_IP_);
 144         return error;
 145 }
 146
 147 /*
 148  * Insert the record referred to by cur to the value given
 149  * by [bno, len, refcount].
 150  * This either works (return 0) or gets an EFSCORRUPTED error.
 151  */
 152 STATIC int
 153 xfs_refcount_insert(
 154         struct xfs_btree_cur            *cur,
 155         struct xfs_refcount_irec        *irec,
 156         int                             *i)
 157 {
 158         int                             error;
 159
 160         trace_xfs_refcount_insert(cur->bc_mp, cur->bc_private.a.agno, irec);
 161         cur->bc_rec.rc.rc_startblock = irec->rc_startblock;
 162         cur->bc_rec.rc.rc_blockcount = irec->rc_blockcount;
 163         cur->bc_rec.rc.rc_refcount = irec->rc_refcount;
 164         error = xfs_btree_insert(cur, i);
 165         XFS_WANT_CORRUPTED_GOTO(cur->bc_mp, *i == 1, out_error);
 166 out_error:
 167         if (error)
 168                 trace_xfs_refcount_insert_error(cur->bc_mp,
 169                                 cur->bc_private.a.agno, error, _RET_IP_);
 170         return error;
 171 }
 172
 173 /*
 174  * Remove the record referred to by cur, then set the pointer to the spot
 175  * where the record could be re-inserted, in case we want to increment or
 176  * decrement the cursor.
 177  * This either works (return 0) or gets an EFSCORRUPTED error.
 178  */
 179 STATIC int
 180 xfs_refcount_delete(
 181         struct xfs_btree_cur    *cur,
 182         int                     *i)
 183 {
 184         struct xfs_refcount_irec        irec;
 185         int                     found_rec;
 186         int                     error;
 187
 188         error = xfs_refcount_get_rec(cur, &irec, &found_rec);
 189         if (error)
 190                 goto out_error;
 191         XFS_WANT_CORRUPTED_GOTO(cur->bc_mp, found_rec == 1, out_error);
 192         trace_xfs_refcount_delete(cur->bc_mp, cur->bc_private.a.agno, &irec);
 193         error = xfs_btree_delete(cur, i);
 194         XFS_WANT_CORRUPTED_GOTO(cur->bc_mp, *i == 1, out_error);
 195         if (error)
 196                 goto out_error;
 197         error = xfs_refcount_lookup_ge(cur, irec.rc_startblock, &found_rec);
 198 out_error:
 199         if (error)
 200                 trace_xfs_refcount_delete_error(cur->bc_mp,
 201                                 cur->bc_private.a.agno, error, _RET_IP_);
 202         return error;
 203 }
 204
 205 /*
 206  * Adjusting the Reference Count
 207  *
 208  * As stated elsewhere, the reference count btree (refcbt) stores
 209  * >1 reference counts for extents of physical blocks.  In this
 210  * operation, we're either raising or lowering the reference count of
 211  * some subrange stored in the tree:
 212  *
 213  *      <------ adjustment range ------>
 214  * ----+   +---+-----+ +--+--------+---------
 215  *  2  |   | 3 |  4  | |17|   55   |   10
 216  * ----+   +---+-----+ +--+--------+---------
 217  * X axis is physical blocks number;
 218  * reference counts are the numbers inside the rectangles
 219  *
 220  * The first thing we need to do is to ensure that there are no
 221  * refcount extents crossing either boundary of the range to be
 222  * adjusted.  For any extent that does cross a boundary, split it into
 223  * two extents so that we can increment the refcount of one of the
 224  * pieces later:
 225  *
 226  *      <------ adjustment range ------>
 227  * ----+   +---+-----+ +--+--------+----+----
 228  *  2  |   | 3 |  2  | |17|   55   | 10 | 10
 229  * ----+   +---+-----+ +--+--------+----+----
 230  *
 231  * For this next step, let's assume that all the physical blocks in
 232  * the adjustment range are mapped to a file and are therefore in use
 233  * at least once.  Therefore, we can infer that any gap in the
 234  * refcount tree within the adjustment range represents a physical
 235  * extent with refcount == 1:
 236  *
 237  *      <------ adjustment range ------>
 238  * ----+---+---+-----+-+--+--------+----+----
 239  *  2  |"1"| 3 |  2  |1|17|   55   | 10 | 10
 240  * ----+---+---+-----+-+--+--------+----+----
 241  *      ^
 242  *
 243  * For each extent that falls within the interval range, figure out
 244  * which extent is to the left or the right of that extent.  Now we
 245  * have a left, current, and right extent.  If the new reference count
 246  * of the center extent enables us to merge left, center, and right
 247  * into one record covering all three, do so.  If the center extent is
 248  * at the left end of the range, abuts the left extent, and its new
 249  * reference count matches the left extent's record, then merge them.
 250  * If the center extent is at the right end of the range, abuts the
 251  * right extent, and the reference counts match, merge those.  In the
 252  * example, we can left merge (assuming an increment operation):
 253  *
 254  *      <------ adjustment range ------>
 255  * --------+---+-----+-+--+--------+----+----
 256  *    2    | 3 |  2  |1|17|   55   | 10 | 10
 257  * --------+---+-----+-+--+--------+----+----
 258  *          ^
 259  *
 260  * For all other extents within the range, adjust the reference count
 261  * or delete it if the refcount falls below 2.  If we were
 262  * incrementing, the end result looks like this:
 263  *
 264  *      <------ adjustment range ------>
 265  * --------+---+-----+-+--+--------+----+----
 266  *    2    | 4 |  3  |2|18|   56   | 11 | 10
 267  * --------+---+-----+-+--+--------+----+----
 268  *
 269  * The result of a decrement operation looks as such:
 270  *
 271  *      <------ adjustment range ------>
 272  * ----+   +---+       +--+--------+----+----
 273  *  2  |   | 2 |       |16|   54   |  9 | 10
 274  * ----+   +---+       +--+--------+----+----
 275  *      DDDD    111111DD
 276  *
 277  * The blocks marked "D" are freed; the blocks marked "1" are only
 278  * referenced once and therefore the record is removed from the
 279  * refcount btree.
 280  */
 281
 282 /* Next block after this extent. */
 283 static inline xfs_agblock_t
 284 xfs_refc_next(
 285         struct xfs_refcount_irec        *rc)
 286 {
 287         return rc->rc_startblock + rc->rc_blockcount;
 288 }
 289
 290 /*
 291  * Split a refcount extent that crosses agbno.
 292  */
 293 STATIC int
 294 xfs_refcount_split_extent(
 295         struct xfs_btree_cur            *cur,
 296         xfs_agblock_t                   agbno,
 297         bool                            *shape_changed)
 298 {
 299         struct xfs_refcount_irec        rcext, tmp;
 300         int                             found_rec;
 301         int                             error;
 302
 303         *shape_changed = false;
 304         error = xfs_refcount_lookup_le(cur, agbno, &found_rec);
 305         if (error)
 306                 goto out_error;
 307         if (!found_rec)
 308                 return 0;
 309
 310         error = xfs_refcount_get_rec(cur, &rcext, &found_rec);
 311         if (error)
 312                 goto out_error;
 313         XFS_WANT_CORRUPTED_GOTO(cur->bc_mp, found_rec == 1, out_error);
 314         if (rcext.rc_startblock == agbno || xfs_refc_next(&rcext) <= agbno)
 315                 return 0;
 316
 317         *shape_changed = true;
 318         trace_xfs_refcount_split_extent(cur->bc_mp, cur->bc_private.a.agno,
 319                         &rcext, agbno);
 320
 321         /* Establish the right extent. */
 322         tmp = rcext;
 323         tmp.rc_startblock = agbno;
 324         tmp.rc_blockcount -= (agbno - rcext.rc_startblock);
 325         error = xfs_refcount_update(cur, &tmp);
 326         if (error)
 327                 goto out_error;
 328
 329         /* Insert the left extent. */
 330         tmp = rcext;
 331         tmp.rc_blockcount = agbno - rcext.rc_startblock;
 332         error = xfs_refcount_insert(cur, &tmp, &found_rec);
 333         if (error)
 334                 goto out_error;
 335         XFS_WANT_CORRUPTED_GOTO(cur->bc_mp, found_rec == 1, out_error);
 336         return error;
 337
 338 out_error:
 339         trace_xfs_refcount_split_extent_error(cur->bc_mp,
 340                         cur->bc_private.a.agno, error, _RET_IP_);
 341         return error;
 342 }
 343
 344 /*
 345  * Merge the left, center, and right extents.
 346  */
 347 STATIC int
 348 xfs_refcount_merge_center_extents(
 349         struct xfs_btree_cur            *cur,
 350         struct xfs_refcount_irec        *left,
 351         struct xfs_refcount_irec        *center,
 352         struct xfs_refcount_irec        *right,
 353         unsigned long long              extlen,
 354         xfs_agblock_t                   *agbno,
 355         xfs_extlen_t                    *aglen)
 356 {
 357         int                             error;
 358         int                             found_rec;
 359
 360         trace_xfs_refcount_merge_center_extents(cur->bc_mp,
 361                         cur->bc_private.a.agno, left, center, right);
 362
 363         /*
 364          * Make sure the center and right extents are not in the btree.
 365          * If the center extent was synthesized, the first delete call
 366          * removes the right extent and we skip the second deletion.
 367          * If center and right were in the btree, then the first delete
 368          * call removes the center and the second one removes the right
 369          * extent.
 370          */
 371         error = xfs_refcount_lookup_ge(cur, center->rc_startblock,
 372                         &found_rec);
 373         if (error)
 374                 goto out_error;
 375         XFS_WANT_CORRUPTED_GOTO(cur->bc_mp, found_rec == 1, out_error);
 376
 377         error = xfs_refcount_delete(cur, &found_rec);
 378         if (error)
 379                 goto out_error;
 380         XFS_WANT_CORRUPTED_GOTO(cur->bc_mp, found_rec == 1, out_error);
 381
 382         if (center->rc_refcount > 1) {
 383                 error = xfs_refcount_delete(cur, &found_rec);
 384                 if (error)
 385                         goto out_error;
 386                 XFS_WANT_CORRUPTED_GOTO(cur->bc_mp, found_rec == 1,
 387                                 out_error);
 388         }
 389
 390         /* Enlarge the left extent. */
 391         error = xfs_refcount_lookup_le(cur, left->rc_startblock,
 392                         &found_rec);
 393         if (error)
 394                 goto out_error;
 395         XFS_WANT_CORRUPTED_GOTO(cur->bc_mp, found_rec == 1, out_error);
 396
 397         left->rc_blockcount = extlen;
 398         error = xfs_refcount_update(cur, left);
 399         if (error)
 400                 goto out_error;
 401
 402         *aglen = 0;
 403         return error;
 404
 405 out_error:
 406         trace_xfs_refcount_merge_center_extents_error(cur->bc_mp,
 407                         cur->bc_private.a.agno, error, _RET_IP_);
 408         return error;
 409 }
 410
 411 /*
 412  * Merge with the left extent.
 413  */
 414 STATIC int
 415 xfs_refcount_merge_left_extent(
 416         struct xfs_btree_cur            *cur,
 417         struct xfs_refcount_irec        *left,
 418         struct xfs_refcount_irec        *cleft,
 419         xfs_agblock_t                   *agbno,
 420         xfs_extlen_t                    *aglen)
 421 {
 422         int                             error;
 423         int                             found_rec;
 424
 425         trace_xfs_refcount_merge_left_extent(cur->bc_mp,
 426                         cur->bc_private.a.agno, left, cleft);
 427
 428         /* If the extent at agbno (cleft) wasn't synthesized, remove it. */
 429         if (cleft->rc_refcount > 1) {
 430                 error = xfs_refcount_lookup_le(cur, cleft->rc_startblock,
 431                                 &found_rec);
 432                 if (error)
 433                         goto out_error;
 434                 XFS_WANT_CORRUPTED_GOTO(cur->bc_mp, found_rec == 1,
 435                                 out_error);
 436
 437                 error = xfs_refcount_delete(cur, &found_rec);
 438                 if (error)
 439                         goto out_error;
 440                 XFS_WANT_CORRUPTED_GOTO(cur->bc_mp, found_rec == 1,
 441                                 out_error);
 442         }
 443
 444         /* Enlarge the left extent. */
 445         error = xfs_refcount_lookup_le(cur, left->rc_startblock,
 446                         &found_rec);
 447         if (error)
 448                 goto out_error;
 449         XFS_WANT_CORRUPTED_GOTO(cur->bc_mp, found_rec == 1, out_error);
 450
 451         left->rc_blockcount += cleft->rc_blockcount;
 452         error = xfs_refcount_update(cur, left);
 453         if (error)
 454                 goto out_error;
 455
 456         *agbno += cleft->rc_blockcount;
 457         *aglen -= cleft->rc_blockcount;
 458         return error;
 459
 460 out_error:
 461         trace_xfs_refcount_merge_left_extent_error(cur->bc_mp,
 462                         cur->bc_private.a.agno, error, _RET_IP_);
 463         return error;
 464 }
 465
 466 /*
 467  * Merge with the right extent.
 468  */
 469 STATIC int
 470 xfs_refcount_merge_right_extent(
 471         struct xfs_btree_cur            *cur,
 472         struct xfs_refcount_irec        *right,
 473         struct xfs_refcount_irec        *cright,
 474         xfs_agblock_t                   *agbno,
 475         xfs_extlen_t                    *aglen)
 476 {
 477         int                             error;
 478         int                             found_rec;
 479
 480         trace_xfs_refcount_merge_right_extent(cur->bc_mp,
 481                         cur->bc_private.a.agno, cright, right);
 482
 483         /*
 484          * If the extent ending at agbno+aglen (cright) wasn't synthesized,
 485          * remove it.
 486          */
 487         if (cright->rc_refcount > 1) {
 488                 error = xfs_refcount_lookup_le(cur, cright->rc_startblock,
 489                         &found_rec);
 490                 if (error)
 491                         goto out_error;
 492                 XFS_WANT_CORRUPTED_GOTO(cur->bc_mp, found_rec == 1,
 493                                 out_error);
 494
 495                 error = xfs_refcount_delete(cur, &found_rec);
 496                 if (error)
 497                         goto out_error;
 498                 XFS_WANT_CORRUPTED_GOTO(cur->bc_mp, found_rec == 1,
 499                                 out_error);
 500         }
 501
 502         /* Enlarge the right extent. */
 503         error = xfs_refcount_lookup_le(cur, right->rc_startblock,
 504                         &found_rec);
 505         if (error)
 506                 goto out_error;
 507         XFS_WANT_CORRUPTED_GOTO(cur->bc_mp, found_rec == 1, out_error);
 508
 509         right->rc_startblock -= cright->rc_blockcount;
 510         right->rc_blockcount += cright->rc_blockcount;
 511         error = xfs_refcount_update(cur, right);
 512         if (error)
 513                 goto out_error;
 514
 515         *aglen -= cright->rc_blockcount;
 516         return error;
 517
 518 out_error:
 519         trace_xfs_refcount_merge_right_extent_error(cur->bc_mp,
 520                         cur->bc_private.a.agno, error, _RET_IP_);
 521         return error;
 522 }
 523
 524 #define XFS_FIND_RCEXT_SHARED   1
 525 #define XFS_FIND_RCEXT_COW      2
 526 /*
 527  * Find the left extent and the one after it (cleft).  This function assumes
 528  * that we've already split any extent crossing agbno.
 529  */
 530 STATIC int
 531 xfs_refcount_find_left_extents(
 532         struct xfs_btree_cur            *cur,
 533         struct xfs_refcount_irec        *left,
 534         struct xfs_refcount_irec        *cleft,
 535         xfs_agblock_t                   agbno,
 536         xfs_extlen_t                    aglen,
 537         int                             flags)
 538 {
 539         struct xfs_refcount_irec        tmp;
 540         int                             error;
 541         int                             found_rec;
 542
 543         left->rc_startblock = cleft->rc_startblock = NULLAGBLOCK;
 544         error = xfs_refcount_lookup_le(cur, agbno - 1, &found_rec);
 545         if (error)
 546                 goto out_error;
 547         if (!found_rec)
 548                 return 0;
 549
 550         error = xfs_refcount_get_rec(cur, &tmp, &found_rec);
 551         if (error)
 552                 goto out_error;
 553         XFS_WANT_CORRUPTED_GOTO(cur->bc_mp, found_rec == 1, out_error);
 554
 555         if (xfs_refc_next(&tmp) != agbno)
 556                 return 0;
 557         if ((flags & XFS_FIND_RCEXT_SHARED) && tmp.rc_refcount < 2)
 558                 return 0;
 559         if ((flags & XFS_FIND_RCEXT_COW) && tmp.rc_refcount > 1)
 560                 return 0;
 561         /* We have a left extent; retrieve (or invent) the next right one */
 562         *left = tmp;
 563
 564         error = xfs_btree_increment(cur, 0, &found_rec);
 565         if (error)
 566                 goto out_error;
 567         if (found_rec) {
 568                 error = xfs_refcount_get_rec(cur, &tmp, &found_rec);
 569                 if (error)
 570                         goto out_error;
 571                 XFS_WANT_CORRUPTED_GOTO(cur->bc_mp, found_rec == 1,
 572                                 out_error);
 573
 574                 /* if tmp starts at the end of our range, just use that */
 575                 if (tmp.rc_startblock == agbno)
 576                         *cleft = tmp;
 577                 else {
 578                         /*
 579                          * There's a gap in the refcntbt at the start of the
 580                          * range we're interested in (refcount == 1) so
 581                          * synthesize the implied extent and pass it back.
 582                          * We assume here that the agbno/aglen range was
 583                          * passed in from a data fork extent mapping and
 584                          * therefore is allocated to exactly one owner.
 585                          */
 586                         cleft->rc_startblock = agbno;
 587                         cleft->rc_blockcount = min(aglen,
 588                                         tmp.rc_startblock - agbno);
 589                         cleft->rc_refcount = 1;
 590                 }
 591         } else {
 592                 /*
 593                  * No extents, so pretend that there's one covering the whole
 594                  * range.
 595                  */
 596                 cleft->rc_startblock = agbno;
 597                 cleft->rc_blockcount = aglen;
 598                 cleft->rc_refcount = 1;
 599         }
 600         trace_xfs_refcount_find_left_extent(cur->bc_mp, cur->bc_private.a.agno,
 601                         left, cleft, agbno);
 602         return error;
 603
 604 out_error:
 605         trace_xfs_refcount_find_left_extent_error(cur->bc_mp,
 606                         cur->bc_private.a.agno, error, _RET_IP_);
 607         return error;
 608 }
 609
 610 /*
 611  * Find the right extent and the one before it (cright).  This function
 612  * assumes that we've already split any extents crossing agbno + aglen.
 613  */
 614 STATIC int
 615 xfs_refcount_find_right_extents(
 616         struct xfs_btree_cur            *cur,
 617         struct xfs_refcount_irec        *right,
 618         struct xfs_refcount_irec        *cright,
 619         xfs_agblock_t                   agbno,
 620         xfs_extlen_t                    aglen,
 621         int                             flags)
 622 {
 623         struct xfs_refcount_irec        tmp;
 624         int                             error;
 625         int                             found_rec;
 626
 627         right->rc_startblock = cright->rc_startblock = NULLAGBLOCK;
 628         error = xfs_refcount_lookup_ge(cur, agbno + aglen, &found_rec);
 629         if (error)
 630                 goto out_error;
 631         if (!found_rec)
 632                 return 0;
 633
 634         error = xfs_refcount_get_rec(cur, &tmp, &found_rec);
 635         if (error)
 636                 goto out_error;
 637         XFS_WANT_CORRUPTED_GOTO(cur->bc_mp, found_rec == 1, out_error);
 638
 639         if (tmp.rc_startblock != agbno + aglen)
 640                 return 0;
 641         if ((flags & XFS_FIND_RCEXT_SHARED) && tmp.rc_refcount < 2)
 642                 return 0;
 643         if ((flags & XFS_FIND_RCEXT_COW) && tmp.rc_refcount > 1)
 644                 return 0;
 645         /* We have a right extent; retrieve (or invent) the next left one */
 646         *right = tmp;
 647
 648         error = xfs_btree_decrement(cur, 0, &found_rec);
 649         if (error)
 650                 goto out_error;
 651         if (found_rec) {
 652                 error = xfs_refcount_get_rec(cur, &tmp, &found_rec);
 653                 if (error)
 654                         goto out_error;
 655                 XFS_WANT_CORRUPTED_GOTO(cur->bc_mp, found_rec == 1,
 656                                 out_error);
 657
 658                 /* if tmp ends at the end of our range, just use that */
 659                 if (xfs_refc_next(&tmp) == agbno + aglen)
 660                         *cright = tmp;
 661                 else {
 662                         /*
 663                          * There's a gap in the refcntbt at the end of the
 664                          * range we're interested in (refcount == 1) so
 665                          * create the implied extent and pass it back.
 666                          * We assume here that the agbno/aglen range was
 667                          * passed in from a data fork extent mapping and
 668                          * therefore is allocated to exactly one owner.
 669                          */
 670                         cright->rc_startblock = max(agbno, xfs_refc_next(&tmp));
 671                         cright->rc_blockcount = right->rc_startblock -
 672                                         cright->rc_startblock;
 673                         cright->rc_refcount = 1;
 674                 }
 675         } else {
 676                 /*
 677                  * No extents, so pretend that there's one covering the whole
 678                  * range.
 679                  */
 680                 cright->rc_startblock = agbno;
 681                 cright->rc_blockcount = aglen;
 682                 cright->rc_refcount = 1;
 683         }
 684         trace_xfs_refcount_find_right_extent(cur->bc_mp, cur->bc_private.a.agno,
 685                         cright, right, agbno + aglen);
 686         return error;
 687
 688 out_error:
 689         trace_xfs_refcount_find_right_extent_error(cur->bc_mp,
 690                         cur->bc_private.a.agno, error, _RET_IP_);
 691         return error;
 692 }
 693
 694 /* Is this extent valid? */
 695 static inline bool
 696 xfs_refc_valid(
 697         struct xfs_refcount_irec        *rc)
 698 {
 699         return rc->rc_startblock != NULLAGBLOCK;
 700 }
 701
 702 /*
 703  * Try to merge with any extents on the boundaries of the adjustment range.
 704  */
 705 STATIC int
 706 xfs_refcount_merge_extents(
 707         struct xfs_btree_cur    *cur,
 708         xfs_agblock_t           *agbno,
 709         xfs_extlen_t            *aglen,
 710         enum xfs_refc_adjust_op adjust,
 711         int                     flags,
 712         bool                    *shape_changed)
 713 {
 714         struct xfs_refcount_irec        left = {0}, cleft = {0};
 715         struct xfs_refcount_irec        cright = {0}, right = {0};
 716         int                             error;
 717         unsigned long long              ulen;
 718         bool                            cequal;
 719
 720         *shape_changed = false;
 721         /*
 722          * Find the extent just below agbno [left], just above agbno [cleft],
 723          * just below (agbno + aglen) [cright], and just above (agbno + aglen)
 724          * [right].
 725          */
 726         error = xfs_refcount_find_left_extents(cur, &left, &cleft, *agbno,
 727                         *aglen, flags);
 728         if (error)
 729                 return error;
 730         error = xfs_refcount_find_right_extents(cur, &right, &cright, *agbno,
 731                         *aglen, flags);
 732         if (error)
 733                 return error;
 734
 735         /* No left or right extent to merge; exit. */
 736         if (!xfs_refc_valid(&left) && !xfs_refc_valid(&right))
 737                 return 0;
 738
 739         cequal = (cleft.rc_startblock == cright.rc_startblock) &&
 740                  (cleft.rc_blockcount == cright.rc_blockcount);
 741
 742         /* Try to merge left, cleft, and right.  cleft must == cright. */
 743         ulen = (unsigned long long)left.rc_blockcount + cleft.rc_blockcount +
 744                         right.rc_blockcount;
 745         if (xfs_refc_valid(&left) && xfs_refc_valid(&right) &&
 746             xfs_refc_valid(&cleft) && xfs_refc_valid(&cright) && cequal &&
 747             left.rc_refcount == cleft.rc_refcount + adjust &&
 748             right.rc_refcount == cleft.rc_refcount + adjust &&
 749             ulen < MAXREFCEXTLEN) {
 750                 *shape_changed = true;
 751                 return xfs_refcount_merge_center_extents(cur, &left, &cleft,
 752                                 &right, ulen, agbno, aglen);
 753         }
 754
 755         /* Try to merge left and cleft. */
 756         ulen = (unsigned long long)left.rc_blockcount + cleft.rc_blockcount;
 757         if (xfs_refc_valid(&left) && xfs_refc_valid(&cleft) &&
 758             left.rc_refcount == cleft.rc_refcount + adjust &&
 759             ulen < MAXREFCEXTLEN) {
 760                 *shape_changed = true;
 761                 error = xfs_refcount_merge_left_extent(cur, &left, &cleft,
 762                                 agbno, aglen);
 763                 if (error)
 764                         return error;
 765
 766                 /*
 767                  * If we just merged left + cleft and cleft == cright,
 768                  * we no longer have a cright to merge with right.  We're done.
 769                  */
 770                 if (cequal)
 771                         return 0;
 772         }
 773
 774         /* Try to merge cright and right. */
 775         ulen = (unsigned long long)right.rc_blockcount + cright.rc_blockcount;
 776         if (xfs_refc_valid(&right) && xfs_refc_valid(&cright) &&
 777             right.rc_refcount == cright.rc_refcount + adjust &&
 778             ulen < MAXREFCEXTLEN) {
 779                 *shape_changed = true;
 780                 return xfs_refcount_merge_right_extent(cur, &right, &cright,
 781                                 agbno, aglen);
 782         }
 783
 784         return error;
 785 }
 786
 787 /*
 788  * XXX: This is a pretty hand-wavy estimate.  The penalty for guessing
 789  * true incorrectly is a shutdown FS; the penalty for guessing false
 790  * incorrectly is more transaction rolls than might be necessary.
 791  * Be conservative here.
 792  */
 793 static bool
 794 xfs_refcount_still_have_space(
 795         struct xfs_btree_cur            *cur)
 796 {
 797         unsigned long                   overhead;
 798
 799         overhead = cur->bc_private.a.priv.refc.shape_changes *
 800                         xfs_allocfree_log_count(cur->bc_mp, 1);
 801         overhead *= cur->bc_mp->m_sb.sb_blocksize;
 802
 803         /*
 804          * Only allow 2 refcount extent updates per transaction if the
 805          * refcount continue update "error" has been injected.
 806          */
 807         if (cur->bc_private.a.priv.refc.nr_ops > 2 &&
 808             XFS_TEST_ERROR(false, cur->bc_mp,
 809                         XFS_ERRTAG_REFCOUNT_CONTINUE_UPDATE))
 810                 return false;
 811
 812         if (cur->bc_private.a.priv.refc.nr_ops == 0)
 813                 return true;
 814         else if (overhead > cur->bc_tp->t_log_res)
 815                 return false;
 816         return  cur->bc_tp->t_log_res - overhead >
 817                 cur->bc_private.a.priv.refc.nr_ops * XFS_REFCOUNT_ITEM_OVERHEAD;
 818 }
 819
 820 /*
 821  * Adjust the refcounts of middle extents.  At this point we should have
 822  * split extents that crossed the adjustment range; merged with adjacent
 823  * extents; and updated agbno/aglen to reflect the merges.  Therefore,
 824  * all we have to do is update the extents inside [agbno, agbno + aglen].
 825  */
 826 STATIC int
 827 xfs_refcount_adjust_extents(
 828         struct xfs_btree_cur    *cur,
 829         xfs_agblock_t           *agbno,
 830         xfs_extlen_t            *aglen,
 831         enum xfs_refc_adjust_op adj,
 832         struct xfs_defer_ops    *dfops,
 833         struct xfs_owner_info   *oinfo)
 834 {
 835         struct xfs_refcount_irec        ext, tmp;
 836         int                             error;
 837         int                             found_rec, found_tmp;
 838         xfs_fsblock_t                   fsbno;
 839
 840         /* Merging did all the work already. */
 841         if (*aglen == 0)
 842                 return 0;
 843
 844         error = xfs_refcount_lookup_ge(cur, *agbno, &found_rec);
 845         if (error)
 846                 goto out_error;
 847
 848         while (*aglen > 0 && xfs_refcount_still_have_space(cur)) {
 849                 error = xfs_refcount_get_rec(cur, &ext, &found_rec);
 850                 if (error)
 851                         goto out_error;
 852                 if (!found_rec) {
 853                         ext.rc_startblock = cur->bc_mp->m_sb.sb_agblocks;
 854                         ext.rc_blockcount = 0;
 855                         ext.rc_refcount = 0;
 856                 }
 857
 858                 /*
 859                  * Deal with a hole in the refcount tree; if a file maps to
 860                  * these blocks and there's no refcountbt record, pretend that
 861                  * there is one with refcount == 1.
 862                  */
 863                 if (ext.rc_startblock != *agbno) {
 864                         tmp.rc_startblock = *agbno;
 865                         tmp.rc_blockcount = min(*aglen,
 866                                         ext.rc_startblock - *agbno);
 867                         tmp.rc_refcount = 1 + adj;
 868                         trace_xfs_refcount_modify_extent(cur->bc_mp,
 869                                         cur->bc_private.a.agno, &tmp);
 870
 871                         /*
 872                          * Either cover the hole (increment) or
 873                          * delete the range (decrement).
 874                          */
 875                         if (tmp.rc_refcount) {
 876                                 error = xfs_refcount_insert(cur, &tmp,
 877                                                 &found_tmp);
 878                                 if (error)
 879                                         goto out_error;
 880                                 XFS_WANT_CORRUPTED_GOTO(cur->bc_mp,
 881                                                 found_tmp == 1, out_error);
 882                                 cur->bc_private.a.priv.refc.nr_ops++;
 883                         } else {
 884                                 fsbno = XFS_AGB_TO_FSB(cur->bc_mp,
 885                                                 cur->bc_private.a.agno,
 886                                                 tmp.rc_startblock);
 887                                 xfs_bmap_add_free(cur->bc_mp, dfops, fsbno,
 888                                                 tmp.rc_blockcount, oinfo);
 889                         }
 890
 891                         (*agbno) += tmp.rc_blockcount;
 892                         (*aglen) -= tmp.rc_blockcount;
 893
 894                         error = xfs_refcount_lookup_ge(cur, *agbno,
 895                                         &found_rec);
 896                         if (error)
 897                                 goto out_error;
 898                 }
 899
 900                 /* Stop if there's nothing left to modify */
 901                 if (*aglen == 0 || !xfs_refcount_still_have_space(cur))
 902                         break;
 903
 904                 /*
 905                  * Adjust the reference count and either update the tree
 906                  * (incr) or free the blocks (decr).
 907                  */
 908                 if (ext.rc_refcount == MAXREFCOUNT)
 909                         goto skip;
 910                 ext.rc_refcount += adj;
 911                 trace_xfs_refcount_modify_extent(cur->bc_mp,
 912                                 cur->bc_private.a.agno, &ext);
 913                 if (ext.rc_refcount > 1) {
 914                         error = xfs_refcount_update(cur, &ext);
 915                         if (error)
 916                                 goto out_error;
 917                         cur->bc_private.a.priv.refc.nr_ops++;
 918                 } else if (ext.rc_refcount == 1) {
 919                         error = xfs_refcount_delete(cur, &found_rec);
 920                         if (error)
 921                                 goto out_error;
 922                         XFS_WANT_CORRUPTED_GOTO(cur->bc_mp,
 923                                         found_rec == 1, out_error);
 924                         cur->bc_private.a.priv.refc.nr_ops++;
 925                         goto advloop;
 926                 } else {
 927                         fsbno = XFS_AGB_TO_FSB(cur->bc_mp,
 928                                         cur->bc_private.a.agno,
 929                                         ext.rc_startblock);
 930                         xfs_bmap_add_free(cur->bc_mp, dfops, fsbno,
 931                                         ext.rc_blockcount, oinfo);
 932                 }
 933
 934 skip:
 935                 error = xfs_btree_increment(cur, 0, &found_rec);
 936                 if (error)
 937                         goto out_error;
 938
 939 advloop:
 940                 (*agbno) += ext.rc_blockcount;
 941                 (*aglen) -= ext.rc_blockcount;
 942         }
 943
 944         return error;
 945 out_error:
 946         trace_xfs_refcount_modify_extent_error(cur->bc_mp,
 947                         cur->bc_private.a.agno, error, _RET_IP_);
 948         return error;
 949 }
 950
 951 /* Adjust the reference count of a range of AG blocks. */
 952 STATIC int
 953 xfs_refcount_adjust(
 954         struct xfs_btree_cur    *cur,
 955         xfs_agblock_t           agbno,
 956         xfs_extlen_t            aglen,
 957         xfs_agblock_t           *new_agbno,
 958         xfs_extlen_t            *new_aglen,
 959         enum xfs_refc_adjust_op adj,
 960         struct xfs_defer_ops    *dfops,
 961         struct xfs_owner_info   *oinfo)
 962 {
 963         bool                    shape_changed;
 964         int                     shape_changes = 0;
 965         int                     error;
 966
 967         *new_agbno = agbno;
 968         *new_aglen = aglen;
 969         if (adj == XFS_REFCOUNT_ADJUST_INCREASE)
 970                 trace_xfs_refcount_increase(cur->bc_mp, cur->bc_private.a.agno,
 971                                 agbno, aglen);
 972         else
 973                 trace_xfs_refcount_decrease(cur->bc_mp, cur->bc_private.a.agno,
 974                                 agbno, aglen);
 975
 976         /*
 977          * Ensure that no rcextents cross the boundary of the adjustment range.
 978          */
 979         error = xfs_refcount_split_extent(cur, agbno, &shape_changed);
 980         if (error)
 981                 goto out_error;
 982         if (shape_changed)
 983                 shape_changes++;
 984
 985         error = xfs_refcount_split_extent(cur, agbno + aglen, &shape_changed);
 986         if (error)
 987                 goto out_error;
 988         if (shape_changed)
 989                 shape_changes++;
 990
 991         /*
 992          * Try to merge with the left or right extents of the range.
 993          */
 994         error = xfs_refcount_merge_extents(cur, new_agbno, new_aglen, adj,
 995                         XFS_FIND_RCEXT_SHARED, &shape_changed);
 996         if (error)
 997                 goto out_error;
 998         if (shape_changed)
 999                 shape_changes++;
1000         if (shape_changes)
1001                 cur->bc_private.a.priv.refc.shape_changes++;
1002
1003         /* Now that we've taken care of the ends, adjust the middle extents */
1004         error = xfs_refcount_adjust_extents(cur, new_agbno, new_aglen,
1005                         adj, dfops, oinfo);
1006         if (error)
1007                 goto out_error;
1008
1009         return 0;
1010
1011 out_error:
1012         trace_xfs_refcount_adjust_error(cur->bc_mp, cur->bc_private.a.agno,
1013                         error, _RET_IP_);
1014         return error;
1015 }
1016
1017 /* Clean up after calling xfs_refcount_finish_one. */
1018 void
1019 xfs_refcount_finish_one_cleanup(
1020         struct xfs_trans        *tp,
1021         struct xfs_btree_cur    *rcur,
1022         int                     error)
1023 {
1024         struct xfs_buf          *agbp;
1025
1026         if (rcur == NULL)
1027                 return;
1028         agbp = rcur->bc_private.a.agbp;
1029         xfs_btree_del_cursor(rcur, error ? XFS_BTREE_ERROR : XFS_BTREE_NOERROR);
1030         if (error)
1031                 xfs_trans_brelse(tp, agbp);
1032 }
1033
1034 /*
1035  * Process one of the deferred refcount operations.  We pass back the
1036  * btree cursor to maintain our lock on the btree between calls.
1037  * This saves time and eliminates a buffer deadlock between the
1038  * superblock and the AGF because we'll always grab them in the same
1039  * order.
1040  */
1041 int
1042 xfs_refcount_finish_one(
1043         struct xfs_trans                *tp,
1044         struct xfs_defer_ops            *dfops,
1045         enum xfs_refcount_intent_type   type,
1046         xfs_fsblock_t                   startblock,
1047         xfs_extlen_t                    blockcount,
1048         xfs_fsblock_t                   *new_fsb,
1049         xfs_extlen_t                    *new_len,
1050         struct xfs_btree_cur            **pcur)
1051 {
1052         struct xfs_mount                *mp = tp->t_mountp;
1053         struct xfs_btree_cur            *rcur;
1054         struct xfs_buf                  *agbp = NULL;
1055         int                             error = 0;
1056         xfs_agnumber_t                  agno;
1057         xfs_agblock_t                   bno;
1058         xfs_agblock_t                   new_agbno;
1059         unsigned long                   nr_ops = 0;
1060         int                             shape_changes = 0;
1061
1062         agno = XFS_FSB_TO_AGNO(mp, startblock);
1063         ASSERT(agno != NULLAGNUMBER);
1064         bno = XFS_FSB_TO_AGBNO(mp, startblock);
1065
1066         trace_xfs_refcount_deferred(mp, XFS_FSB_TO_AGNO(mp, startblock),
1067                         type, XFS_FSB_TO_AGBNO(mp, startblock),
1068                         blockcount);
1069
1070         if (XFS_TEST_ERROR(false, mp,
1071                         XFS_ERRTAG_REFCOUNT_FINISH_ONE))
1072                 return -EIO;
1073
1074         /*
1075          * If we haven't gotten a cursor or the cursor AG doesn't match
1076          * the startblock, get one now.
1077          */
1078         rcur = *pcur;
1079         if (rcur != NULL && rcur->bc_private.a.agno != agno) {
1080                 nr_ops = rcur->bc_private.a.priv.refc.nr_ops;
1081                 shape_changes = rcur->bc_private.a.priv.refc.shape_changes;
1082                 xfs_refcount_finish_one_cleanup(tp, rcur, 0);
1083                 rcur = NULL;
1084                 *pcur = NULL;
1085         }
1086         if (rcur == NULL) {
1087                 error = xfs_alloc_read_agf(tp->t_mountp, tp, agno,
1088                                 XFS_ALLOC_FLAG_FREEING, &agbp);
1089                 if (error)
1090                         return error;
1091                 if (!agbp)
1092                         return -EFSCORRUPTED;
1093
1094                 rcur = xfs_refcountbt_init_cursor(mp, tp, agbp, agno, dfops);
1095                 if (!rcur) {
1096                         error = -ENOMEM;
1097                         goto out_cur;
1098                 }
1099                 rcur->bc_private.a.priv.refc.nr_ops = nr_ops;
1100                 rcur->bc_private.a.priv.refc.shape_changes = shape_changes;
1101         }
1102         *pcur = rcur;
1103
1104         switch (type) {
1105         case XFS_REFCOUNT_INCREASE:
1106                 error = xfs_refcount_adjust(rcur, bno, blockcount, &new_agbno,
1107                         new_len, XFS_REFCOUNT_ADJUST_INCREASE, dfops, NULL);
1108                 *new_fsb = XFS_AGB_TO_FSB(mp, agno, new_agbno);
1109                 break;
1110         case XFS_REFCOUNT_DECREASE:
1111                 error = xfs_refcount_adjust(rcur, bno, blockcount, &new_agbno,
1112                         new_len, XFS_REFCOUNT_ADJUST_DECREASE, dfops, NULL);
1113                 *new_fsb = XFS_AGB_TO_FSB(mp, agno, new_agbno);
1114                 break;
1115         case XFS_REFCOUNT_ALLOC_COW:
1116                 *new_fsb = startblock + blockcount;
1117                 *new_len = 0;
1118                 error = __xfs_refcount_cow_alloc(rcur, bno, blockcount, dfops);
1119                 break;
1120         case XFS_REFCOUNT_FREE_COW:
1121                 *new_fsb = startblock + blockcount;
1122                 *new_len = 0;
1123                 error = __xfs_refcount_cow_free(rcur, bno, blockcount, dfops);
1124                 break;
1125         default:
1126                 ASSERT(0);
1127                 error = -EFSCORRUPTED;
1128         }
1129         if (!error && *new_len > 0)
1130                 trace_xfs_refcount_finish_one_leftover(mp, agno, type,
1131                                 bno, blockcount, new_agbno, *new_len);
1132         return error;
1133
1134 out_cur:
1135         xfs_trans_brelse(tp, agbp);
1136
1137         return error;
1138 }
1139
1140 /*
1141  * Record a refcount intent for later processing.
1142  */
1143 static int
1144 __xfs_refcount_add(
1145         struct xfs_mount                *mp,
1146         struct xfs_defer_ops            *dfops,
1147         enum xfs_refcount_intent_type   type,
1148         xfs_fsblock_t                   startblock,
1149         xfs_extlen_t                    blockcount)
1150 {
1151         struct xfs_refcount_intent      *ri;
1152
1153         trace_xfs_refcount_defer(mp, XFS_FSB_TO_AGNO(mp, startblock),
1154                         type, XFS_FSB_TO_AGBNO(mp, startblock),
1155                         blockcount);
1156
1157         ri = kmem_alloc(sizeof(struct xfs_refcount_intent),
1158                         KM_SLEEP | KM_NOFS);
1159         INIT_LIST_HEAD(&ri->ri_list);
1160         ri->ri_type = type;
1161         ri->ri_startblock = startblock;
1162         ri->ri_blockcount = blockcount;
1163
1164         xfs_defer_add(dfops, XFS_DEFER_OPS_TYPE_REFCOUNT, &ri->ri_list);
1165         return 0;
1166 }
1167
1168 /*
1169  * Increase the reference count of the blocks backing a file's extent.
1170  */
1171 int
1172 xfs_refcount_increase_extent(
1173         struct xfs_mount                *mp,
1174         struct xfs_defer_ops            *dfops,
1175         struct xfs_bmbt_irec            *PREV)
1176 {
1177         if (!xfs_sb_version_hasreflink(&mp->m_sb))
1178                 return 0;
1179
1180         return __xfs_refcount_add(mp, dfops, XFS_REFCOUNT_INCREASE,
1181                         PREV->br_startblock, PREV->br_blockcount);
1182 }
1183
1184 /*
1185  * Decrease the reference count of the blocks backing a file's extent.
1186  */
1187 int
1188 xfs_refcount_decrease_extent(
1189         struct xfs_mount                *mp,
1190         struct xfs_defer_ops            *dfops,
1191         struct xfs_bmbt_irec            *PREV)
1192 {
1193         if (!xfs_sb_version_hasreflink(&mp->m_sb))
1194                 return 0;
1195
1196         return __xfs_refcount_add(mp, dfops, XFS_REFCOUNT_DECREASE,
1197                         PREV->br_startblock, PREV->br_blockcount);
1198 }
1199
1200 /*
1201  * Given an AG extent, find the lowest-numbered run of shared blocks
1202  * within that range and return the range in fbno/flen.  If
1203  * find_end_of_shared is set, return the longest contiguous extent of
1204  * shared blocks; if not, just return the first extent we find.  If no
1205  * shared blocks are found, fbno and flen will be set to NULLAGBLOCK
1206  * and 0, respectively.
1207  */
1208 int
1209 xfs_refcount_find_shared(
1210         struct xfs_btree_cur            *cur,
1211         xfs_agblock_t                   agbno,
1212         xfs_extlen_t                    aglen,
1213         xfs_agblock_t                   *fbno,
1214         xfs_extlen_t                    *flen,
1215         bool                            find_end_of_shared)
1216 {
1217         struct xfs_refcount_irec        tmp;
1218         int                             i;
1219         int                             have;
1220         int                             error;
1221
1222         trace_xfs_refcount_find_shared(cur->bc_mp, cur->bc_private.a.agno,
1223                         agbno, aglen);
1224
1225         /* By default, skip the whole range */
1226         *fbno = NULLAGBLOCK;
1227         *flen = 0;
1228
1229         /* Try to find a refcount extent that crosses the start */
1230         error = xfs_refcount_lookup_le(cur, agbno, &have);
1231         if (error)
1232                 goto out_error;
1233         if (!have) {
1234                 /* No left extent, look at the next one */
1235                 error = xfs_btree_increment(cur, 0, &have);
1236                 if (error)
1237                         goto out_error;
1238                 if (!have)
1239                         goto done;
1240         }
1241         error = xfs_refcount_get_rec(cur, &tmp, &i);
1242         if (error)
1243                 goto out_error;
1244         XFS_WANT_CORRUPTED_GOTO(cur->bc_mp, i == 1, out_error);
1245
1246         /* If the extent ends before the start, look at the next one */
1247         if (tmp.rc_startblock + tmp.rc_blockcount <= agbno) {
1248                 error = xfs_btree_increment(cur, 0, &have);
1249                 if (error)
1250                         goto out_error;
1251                 if (!have)
1252                         goto done;
1253                 error = xfs_refcount_get_rec(cur, &tmp, &i);
1254                 if (error)
1255                         goto out_error;
1256                 XFS_WANT_CORRUPTED_GOTO(cur->bc_mp, i == 1, out_error);
1257         }
1258
1259         /* If the extent starts after the range we want, bail out */
1260         if (tmp.rc_startblock >= agbno + aglen)
1261                 goto done;
1262
1263         /* We found the start of a shared extent! */
1264         if (tmp.rc_startblock < agbno) {
1265                 tmp.rc_blockcount -= (agbno - tmp.rc_startblock);
1266                 tmp.rc_startblock = agbno;
1267         }
1268
1269         *fbno = tmp.rc_startblock;
1270         *flen = min(tmp.rc_blockcount, agbno + aglen - *fbno);
1271         if (!find_end_of_shared)
1272                 goto done;
1273
1274         /* Otherwise, find the end of this shared extent */
1275         while (*fbno + *flen < agbno + aglen) {
1276                 error = xfs_btree_increment(cur, 0, &have);
1277                 if (error)
1278                         goto out_error;
1279                 if (!have)
1280                         break;
1281                 error = xfs_refcount_get_rec(cur, &tmp, &i);
1282                 if (error)
1283                         goto out_error;
1284                 XFS_WANT_CORRUPTED_GOTO(cur->bc_mp, i == 1, out_error);
1285                 if (tmp.rc_startblock >= agbno + aglen ||
1286                     tmp.rc_startblock != *fbno + *flen)
1287                         break;
1288                 *flen = min(*flen + tmp.rc_blockcount, agbno + aglen - *fbno);
1289         }
1290
1291 done:
1292         trace_xfs_refcount_find_shared_result(cur->bc_mp,
1293                         cur->bc_private.a.agno, *fbno, *flen);
1294
1295 out_error:
1296         if (error)
1297                 trace_xfs_refcount_find_shared_error(cur->bc_mp,
1298                                 cur->bc_private.a.agno, error, _RET_IP_);
1299         return error;
1300 }
1301
1302 /*
1303  * Recovering CoW Blocks After a Crash
1304  *
1305  * Due to the way that the copy on write mechanism works, there's a window of
1306  * opportunity in which we can lose track of allocated blocks during a crash.
1307  * Because CoW uses delayed allocation in the in-core CoW fork, writeback
1308  * causes blocks to be allocated and stored in the CoW fork.  The blocks are
1309  * no longer in the free space btree but are not otherwise recorded anywhere
1310  * until the write completes and the blocks are mapped into the file.  A crash
1311  * in between allocation and remapping results in the replacement blocks being
1312  * lost.  This situation is exacerbated by the CoW extent size hint because
1313  * allocations can hang around for long time.
1314  *
1315  * However, there is a place where we can record these allocations before they
1316  * become mappings -- the reference count btree.  The btree does not record
1317  * extents with refcount == 1, so we can record allocations with a refcount of
1318  * 1.  Blocks being used for CoW writeout cannot be shared, so there should be
1319  * no conflict with shared block records.  These mappings should be created
1320  * when we allocate blocks to the CoW fork and deleted when they're removed
1321  * from the CoW fork.
1322  *
1323  * Minor nit: records for in-progress CoW allocations and records for shared
1324  * extents must never be merged, to preserve the property that (except for CoW
1325  * allocations) there are no refcount btree entries with refcount == 1.  The
1326  * only time this could potentially happen is when unsharing a block that's
1327  * adjacent to CoW allocations, so we must be careful to avoid this.
1328  *
1329  * At mount time we recover lost CoW allocations by searching the refcount
1330  * btree for these refcount == 1 mappings.  These represent CoW allocations
1331  * that were in progress at the time the filesystem went down, so we can free
1332  * them to get the space back.
1333  *
1334  * This mechanism is superior to creating EFIs for unmapped CoW extents for
1335  * several reasons -- first, EFIs pin the tail of the log and would have to be
1336  * periodically relogged to avoid filling up the log.  Second, CoW completions
1337  * will have to file an EFD and create new EFIs for whatever remains in the
1338  * CoW fork; this partially takes care of (1) but extent-size reservations
1339  * will have to periodically relog even if there's no writeout in progress.
1340  * This can happen if the CoW extent size hint is set, which you really want.
1341  * Third, EFIs cannot currently be automatically relogged into newer
1342  * transactions to advance the log tail.  Fourth, stuffing the log full of
1343  * EFIs places an upper bound on the number of CoW allocations that can be
1344  * held filesystem-wide at any given time.  Recording them in the refcount
1345  * btree doesn't require us to maintain any state in memory and doesn't pin
1346  * the log.
1347  */
1348 /*
1349  * Adjust the refcounts of CoW allocations.  These allocations are "magic"
1350  * in that they're not referenced anywhere else in the filesystem, so we
1351  * stash them in the refcount btree with a refcount of 1 until either file
1352  * remapping (or CoW cancellation) happens.
1353  */
1354 STATIC int
1355 xfs_refcount_adjust_cow_extents(
1356         struct xfs_btree_cur    *cur,
1357         xfs_agblock_t           agbno,
1358         xfs_extlen_t            aglen,
1359         enum xfs_refc_adjust_op adj,
1360         struct xfs_defer_ops    *dfops,
1361         struct xfs_owner_info   *oinfo)
1362 {
1363         struct xfs_refcount_irec        ext, tmp;
1364         int                             error;
1365         int                             found_rec, found_tmp;
1366
1367         if (aglen == 0)
1368                 return 0;
1369
1370         /* Find any overlapping refcount records */
1371         error = xfs_refcount_lookup_ge(cur, agbno, &found_rec);
1372         if (error)
1373                 goto out_error;
1374         error = xfs_refcount_get_rec(cur, &ext, &found_rec);
1375         if (error)
1376                 goto out_error;
1377         if (!found_rec) {
1378                 ext.rc_startblock = cur->bc_mp->m_sb.sb_agblocks +
1379                                 XFS_REFC_COW_START;
1380                 ext.rc_blockcount = 0;
1381                 ext.rc_refcount = 0;
1382         }
1383
1384         switch (adj) {
1385         case XFS_REFCOUNT_ADJUST_COW_ALLOC:
1386                 /* Adding a CoW reservation, there should be nothing here. */
1387                 XFS_WANT_CORRUPTED_GOTO(cur->bc_mp,
1388                                 ext.rc_startblock >= agbno + aglen, out_error);
1389
1390                 tmp.rc_startblock = agbno;
1391                 tmp.rc_blockcount = aglen;
1392                 tmp.rc_refcount = 1;
1393                 trace_xfs_refcount_modify_extent(cur->bc_mp,
1394                                 cur->bc_private.a.agno, &tmp);
1395
1396                 error = xfs_refcount_insert(cur, &tmp,
1397                                 &found_tmp);
1398                 if (error)
1399                         goto out_error;
1400                 XFS_WANT_CORRUPTED_GOTO(cur->bc_mp,
1401                                 found_tmp == 1, out_error);
1402                 break;
1403         case XFS_REFCOUNT_ADJUST_COW_FREE:
1404                 /* Removing a CoW reservation, there should be one extent. */
1405                 XFS_WANT_CORRUPTED_GOTO(cur->bc_mp,
1406                         ext.rc_startblock == agbno, out_error);
1407                 XFS_WANT_CORRUPTED_GOTO(cur->bc_mp,
1408                         ext.rc_blockcount == aglen, out_error);
1409                 XFS_WANT_CORRUPTED_GOTO(cur->bc_mp,
1410                         ext.rc_refcount == 1, out_error);
1411
1412                 ext.rc_refcount = 0;
1413                 trace_xfs_refcount_modify_extent(cur->bc_mp,
1414                                 cur->bc_private.a.agno, &ext);
1415                 error = xfs_refcount_delete(cur, &found_rec);
1416                 if (error)
1417                         goto out_error;
1418                 XFS_WANT_CORRUPTED_GOTO(cur->bc_mp,
1419                                 found_rec == 1, out_error);
1420                 break;
1421         default:
1422                 ASSERT(0);
1423         }
1424
1425         return error;
1426 out_error:
1427         trace_xfs_refcount_modify_extent_error(cur->bc_mp,
1428                         cur->bc_private.a.agno, error, _RET_IP_);
1429         return error;
1430 }
1431
1432 /*
1433  * Add or remove refcount btree entries for CoW reservations.
1434  */
1435 STATIC int
1436 xfs_refcount_adjust_cow(
1437         struct xfs_btree_cur    *cur,
1438         xfs_agblock_t           agbno,
1439         xfs_extlen_t            aglen,
1440         enum xfs_refc_adjust_op adj,
1441         struct xfs_defer_ops    *dfops)
1442 {
1443         bool                    shape_changed;
1444         int                     error;
1445
1446         agbno += XFS_REFC_COW_START;
1447
1448         /*
1449          * Ensure that no rcextents cross the boundary of the adjustment range.
1450          */
1451         error = xfs_refcount_split_extent(cur, agbno, &shape_changed);
1452         if (error)
1453                 goto out_error;
1454
1455         error = xfs_refcount_split_extent(cur, agbno + aglen, &shape_changed);
1456         if (error)
1457                 goto out_error;
1458
1459         /*
1460          * Try to merge with the left or right extents of the range.
1461          */
1462         error = xfs_refcount_merge_extents(cur, &agbno, &aglen, adj,
1463                         XFS_FIND_RCEXT_COW, &shape_changed);
1464         if (error)
1465                 goto out_error;
1466
1467         /* Now that we've taken care of the ends, adjust the middle extents */
1468         error = xfs_refcount_adjust_cow_extents(cur, agbno, aglen, adj,
1469                         dfops, NULL);
1470         if (error)
1471                 goto out_error;
1472
1473         return 0;
1474
1475 out_error:
1476         trace_xfs_refcount_adjust_cow_error(cur->bc_mp, cur->bc_private.a.agno,
1477                         error, _RET_IP_);
1478         return error;
1479 }
1480
1481 /*
1482  * Record a CoW allocation in the refcount btree.
1483  */
1484 STATIC int
1485 __xfs_refcount_cow_alloc(
1486         struct xfs_btree_cur    *rcur,
1487         xfs_agblock_t           agbno,
1488         xfs_extlen_t            aglen,
1489         struct xfs_defer_ops    *dfops)
1490 {
1491         trace_xfs_refcount_cow_increase(rcur->bc_mp, rcur->bc_private.a.agno,
1492                         agbno, aglen);
1493
1494         /* Add refcount btree reservation */
1495         return xfs_refcount_adjust_cow(rcur, agbno, aglen,
1496                         XFS_REFCOUNT_ADJUST_COW_ALLOC, dfops);
1497 }
1498
1499 /*
1500  * Remove a CoW allocation from the refcount btree.
1501  */
1502 STATIC int
1503 __xfs_refcount_cow_free(
1504         struct xfs_btree_cur    *rcur,
1505         xfs_agblock_t           agbno,
1506         xfs_extlen_t            aglen,
1507         struct xfs_defer_ops    *dfops)
1508 {
1509         trace_xfs_refcount_cow_decrease(rcur->bc_mp, rcur->bc_private.a.agno,
1510                         agbno, aglen);
1511
1512         /* Remove refcount btree reservation */
1513         return xfs_refcount_adjust_cow(rcur, agbno, aglen,
1514                         XFS_REFCOUNT_ADJUST_COW_FREE, dfops);
1515 }
1516
1517 /* Record a CoW staging extent in the refcount btree. */
1518 int
1519 xfs_refcount_alloc_cow_extent(
1520         struct xfs_mount                *mp,
1521         struct xfs_defer_ops            *dfops,
1522         xfs_fsblock_t                   fsb,
1523         xfs_extlen_t                    len)
1524 {
1525         int                             error;
1526
1527         if (!xfs_sb_version_hasreflink(&mp->m_sb))
1528                 return 0;
1529
1530         error = __xfs_refcount_add(mp, dfops, XFS_REFCOUNT_ALLOC_COW,
1531                         fsb, len);
1532         if (error)
1533                 return error;
1534
1535         /* Add rmap entry */
1536         return xfs_rmap_alloc_extent(mp, dfops, XFS_FSB_TO_AGNO(mp, fsb),
1537                         XFS_FSB_TO_AGBNO(mp, fsb), len, XFS_RMAP_OWN_COW);
1538 }
1539
1540 /* Forget a CoW staging event in the refcount btree. */
1541 int
1542 xfs_refcount_free_cow_extent(
1543         struct xfs_mount                *mp,
1544         struct xfs_defer_ops            *dfops,
1545         xfs_fsblock_t                   fsb,
1546         xfs_extlen_t                    len)
1547 {
1548         int                             error;
1549
1550         if (!xfs_sb_version_hasreflink(&mp->m_sb))
1551                 return 0;
1552
1553         /* Remove rmap entry */
1554         error = xfs_rmap_free_extent(mp, dfops, XFS_FSB_TO_AGNO(mp, fsb),
1555                         XFS_FSB_TO_AGBNO(mp, fsb), len, XFS_RMAP_OWN_COW);
1556         if (error)
1557                 return error;
1558
1559         return __xfs_refcount_add(mp, dfops, XFS_REFCOUNT_FREE_COW,
1560                         fsb, len);
1561 }
1562
1563 struct xfs_refcount_recovery {
1564         struct list_head                rr_list;
1565         struct xfs_refcount_irec        rr_rrec;
1566 };
1567
1568 /* Stuff an extent on the recovery list. */
1569 STATIC int
1570 xfs_refcount_recover_extent(
1571         struct xfs_btree_cur            *cur,
1572         union xfs_btree_rec             *rec,
1573         void                            *priv)
1574 {
1575         struct list_head                *debris = priv;
1576         struct xfs_refcount_recovery    *rr;
1577
1578         if (be32_to_cpu(rec->refc.rc_refcount) != 1)
1579                 return -EFSCORRUPTED;
1580
1581         rr = kmem_alloc(sizeof(struct xfs_refcount_recovery), KM_SLEEP);
1582         xfs_refcount_btrec_to_irec(rec, &rr->rr_rrec);
1583         list_add_tail(&rr->rr_list, debris);
1584
1585         return 0;
1586 }
1587
1588 /* Find and remove leftover CoW reservations. */
1589 int
1590 xfs_refcount_recover_cow_leftovers(
1591         struct xfs_mount                *mp,
1592         xfs_agnumber_t                  agno)
1593 {
1594         struct xfs_trans                *tp;
1595         struct xfs_btree_cur            *cur;
1596         struct xfs_buf                  *agbp;
1597         struct xfs_refcount_recovery    *rr, *n;
1598         struct list_head                debris;
1599         union xfs_btree_irec            low;
1600         union xfs_btree_irec            high;
1601         struct xfs_defer_ops            dfops;
1602         xfs_fsblock_t                   fsb;
1603         xfs_agblock_t                   agbno;
1604         int                             error;
1605
1606         if (mp->m_sb.sb_agblocks >= XFS_REFC_COW_START)
1607                 return -EOPNOTSUPP;
1608
1609         INIT_LIST_HEAD(&debris);
1610
1611         /*
1612          * In this first part, we use an empty transaction to gather up
1613          * all the leftover CoW extents so that we can subsequently
1614          * delete them.  The empty transaction is used to avoid
1615          * a buffer lock deadlock if there happens to be a loop in the
1616          * refcountbt because we're allowed to re-grab a buffer that is
1617          * already attached to our transaction.  When we're done
1618          * recording the CoW debris we cancel the (empty) transaction
1619          * and everything goes away cleanly.
1620          */
1621         error = xfs_trans_alloc_empty(mp, &tp);
1622         if (error)
1623                 return error;
1624
1625         error = xfs_alloc_read_agf(mp, tp, agno, 0, &agbp);
1626         if (error)
1627                 goto out_trans;
1628         if (!agbp) {
1629                 error = -ENOMEM;
1630                 goto out_trans;
1631         }
1632         cur = xfs_refcountbt_init_cursor(mp, tp, agbp, agno, NULL);
1633
1634         /* Find all the leftover CoW staging extents. */
1635         memset(&low, 0, sizeof(low));
1636         memset(&high, 0, sizeof(high));
1637         low.rc.rc_startblock = XFS_REFC_COW_START;
1638         high.rc.rc_startblock = -1U;
1639         error = xfs_btree_query_range(cur, &low, &high,
1640                         xfs_refcount_recover_extent, &debris);
1641         if (error)
1642                 goto out_cursor;
1643         xfs_btree_del_cursor(cur, XFS_BTREE_NOERROR);
1644         xfs_trans_brelse(tp, agbp);
1645         xfs_trans_cancel(tp);
1646
1647         /* Now iterate the list to free the leftovers */
1648         list_for_each_entry_safe(rr, n, &debris, rr_list) {
1649                 /* Set up transaction. */
1650                 error = xfs_trans_alloc(mp, &M_RES(mp)->tr_write, 0, 0, 0, &tp);
1651                 if (error)
1652                         goto out_free;
1653
1654                 trace_xfs_refcount_recover_extent(mp, agno, &rr->rr_rrec);
1655
1656                 /* Free the orphan record */
1657                 xfs_defer_init(&dfops, &fsb);
1658                 agbno = rr->rr_rrec.rc_startblock - XFS_REFC_COW_START;
1659                 fsb = XFS_AGB_TO_FSB(mp, agno, agbno);
1660                 error = xfs_refcount_free_cow_extent(mp, &dfops, fsb,
1661                                 rr->rr_rrec.rc_blockcount);
1662                 if (error)
1663                         goto out_defer;
1664
1665                 /* Free the block. */
1666                 xfs_bmap_add_free(mp, &dfops, fsb,
1667                                 rr->rr_rrec.rc_blockcount, NULL);
1668
1669                 error = xfs_defer_finish(&tp, &dfops);
1670                 if (error)
1671                         goto out_defer;
1672
1673                 error = xfs_trans_commit(tp);
1674                 if (error)
1675                         goto out_free;
1676
1677                 list_del(&rr->rr_list);
1678                 kmem_free(rr);
1679         }
1680
1681         return error;
1682 out_defer:
1683         xfs_defer_cancel(&dfops);
1684 out_trans:
1685         xfs_trans_cancel(tp);
1686 out_free:
1687         /* Free the leftover list */
1688         list_for_each_entry_safe(rr, n, &debris, rr_list) {
1689                 list_del(&rr->rr_list);
1690                 kmem_free(rr);
1691         }
1692         return error;
1693
1694 out_cursor:
1695         xfs_btree_del_cursor(cur, XFS_BTREE_ERROR);
1696         xfs_trans_brelse(tp, agbp);
1697         goto out_trans;
1698 }
1699
1700 /* Is there a record covering a given extent? */
1701 int
1702 xfs_refcount_has_record(
1703         struct xfs_btree_cur    *cur,
1704         xfs_agblock_t           bno,
1705         xfs_extlen_t            len,
1706         bool                    *exists)
1707 {
1708         union xfs_btree_irec    low;
1709         union xfs_btree_irec    high;
1710
1711         memset(&low, 0, sizeof(low));
1712         low.rc.rc_startblock = bno;
1713         memset(&high, 0xFF, sizeof(high));
1714         high.rc.rc_startblock = bno + len - 1;
1715
1716         return xfs_btree_has_record(cur, &low, &high, exists);
1717 }