1// SPDX-License-Identifier: GPL-2.0+ 2/* 3 * Copyright (C) 2016 Oracle. All Rights Reserved. 4 * Author: Darrick J. Wong <darrick.wong@oracle.com> 5 */ 6#include "xfs.h" 7#include "xfs_fs.h" 8#include "xfs_shared.h" 9#include "xfs_format.h" 10#include "xfs_log_format.h" 11#include "xfs_trans_resv.h" 12#include "xfs_mount.h" 13#include "xfs_btree.h" 14#include "xfs_btree_staging.h" 15#include "xfs_refcount_btree.h" 16#include "xfs_refcount.h" 17#include "xfs_alloc.h" 18#include "xfs_error.h" 19#include "xfs_trace.h" 20#include "xfs_trans.h" 21#include "xfs_bit.h" 22#include "xfs_rmap.h" 23#include "xfs_ag.h" 24 25static struct kmem_cache *xfs_refcountbt_cur_cache; 26 27static struct xfs_btree_cur * 28xfs_refcountbt_dup_cursor( 29 struct xfs_btree_cur *cur) 30{ 31 return xfs_refcountbt_init_cursor(cur->bc_mp, cur->bc_tp, 32 cur->bc_ag.agbp, cur->bc_ag.pag); 33} 34 35STATIC void 36xfs_refcountbt_set_root( 37 struct xfs_btree_cur *cur, 38 const union xfs_btree_ptr *ptr, 39 int inc) 40{ 41 struct xfs_buf *agbp = cur->bc_ag.agbp; 42 struct xfs_agf *agf = agbp->b_addr; 43 struct xfs_perag *pag = agbp->b_pag; 44 45 ASSERT(ptr->s != 0); 46 47 agf->agf_refcount_root = ptr->s; 48 be32_add_cpu(&agf->agf_refcount_level, inc); 49 pag->pagf_refcount_level += inc; 50 51 xfs_alloc_log_agf(cur->bc_tp, agbp, 52 XFS_AGF_REFCOUNT_ROOT | XFS_AGF_REFCOUNT_LEVEL); 53} 54 55STATIC int 56xfs_refcountbt_alloc_block( 57 struct xfs_btree_cur *cur, 58 const union xfs_btree_ptr *start, 59 union xfs_btree_ptr *new, 60 int *stat) 61{ 62 struct xfs_buf *agbp = cur->bc_ag.agbp; 63 struct xfs_agf *agf = agbp->b_addr; 64 struct xfs_alloc_arg args; /* block allocation args */ 65 int error; /* error return value */ 66 67 memset(&args, 0, sizeof(args)); 68 args.tp = cur->bc_tp; 69 args.mp = cur->bc_mp; 70 args.pag = cur->bc_ag.pag; 71 args.oinfo = XFS_RMAP_OINFO_REFC; 72 args.minlen = args.maxlen = args.prod = 1; 73 args.resv = XFS_AG_RESV_METADATA; 74 75 error = xfs_alloc_vextent_near_bno(&args, 76 XFS_AGB_TO_FSB(args.mp, args.pag->pag_agno, 77 xfs_refc_block(args.mp))); 78 if (error) 79 goto out_error; 80 trace_xfs_refcountbt_alloc_block(cur->bc_mp, cur->bc_ag.pag->pag_agno, 81 args.agbno, 1); 82 if (args.fsbno == NULLFSBLOCK) { 83 *stat = 0; 84 return 0; 85 } 86 ASSERT(args.agno == cur->bc_ag.pag->pag_agno); 87 ASSERT(args.len == 1); 88 89 new->s = cpu_to_be32(args.agbno); 90 be32_add_cpu(&agf->agf_refcount_blocks, 1); 91 xfs_alloc_log_agf(cur->bc_tp, agbp, XFS_AGF_REFCOUNT_BLOCKS); 92 93 *stat = 1; 94 return 0; 95 96out_error: 97 return error; 98} 99 100STATIC int 101xfs_refcountbt_free_block( 102 struct xfs_btree_cur *cur, 103 struct xfs_buf *bp) 104{ 105 struct xfs_mount *mp = cur->bc_mp; 106 struct xfs_buf *agbp = cur->bc_ag.agbp; 107 struct xfs_agf *agf = agbp->b_addr; 108 xfs_fsblock_t fsbno = XFS_DADDR_TO_FSB(mp, xfs_buf_daddr(bp)); 109 110 trace_xfs_refcountbt_free_block(cur->bc_mp, cur->bc_ag.pag->pag_agno, 111 XFS_FSB_TO_AGBNO(cur->bc_mp, fsbno), 1); 112 be32_add_cpu(&agf->agf_refcount_blocks, -1); 113 xfs_alloc_log_agf(cur->bc_tp, agbp, XFS_AGF_REFCOUNT_BLOCKS); 114 return xfs_free_extent_later(cur->bc_tp, fsbno, 1, 115 &XFS_RMAP_OINFO_REFC, XFS_AG_RESV_METADATA); 116} 117 118STATIC int 119xfs_refcountbt_get_minrecs( 120 struct xfs_btree_cur *cur, 121 int level) 122{ 123 return cur->bc_mp->m_refc_mnr[level != 0]; 124} 125 126STATIC int 127xfs_refcountbt_get_maxrecs( 128 struct xfs_btree_cur *cur, 129 int level) 130{ 131 return cur->bc_mp->m_refc_mxr[level != 0]; 132} 133 134STATIC void 135xfs_refcountbt_init_key_from_rec( 136 union xfs_btree_key *key, 137 const union xfs_btree_rec *rec) 138{ 139 key->refc.rc_startblock = rec->refc.rc_startblock; 140} 141 142STATIC void 143xfs_refcountbt_init_high_key_from_rec( 144 union xfs_btree_key *key, 145 const union xfs_btree_rec *rec) 146{ 147 __u32 x; 148 149 x = be32_to_cpu(rec->refc.rc_startblock); 150 x += be32_to_cpu(rec->refc.rc_blockcount) - 1; 151 key->refc.rc_startblock = cpu_to_be32(x); 152} 153 154STATIC void 155xfs_refcountbt_init_rec_from_cur( 156 struct xfs_btree_cur *cur, 157 union xfs_btree_rec *rec) 158{ 159 const struct xfs_refcount_irec *irec = &cur->bc_rec.rc; 160 uint32_t start; 161 162 start = xfs_refcount_encode_startblock(irec->rc_startblock, 163 irec->rc_domain); 164 rec->refc.rc_startblock = cpu_to_be32(start); 165 rec->refc.rc_blockcount = cpu_to_be32(cur->bc_rec.rc.rc_blockcount); 166 rec->refc.rc_refcount = cpu_to_be32(cur->bc_rec.rc.rc_refcount); 167} 168 169STATIC void 170xfs_refcountbt_init_ptr_from_cur( 171 struct xfs_btree_cur *cur, 172 union xfs_btree_ptr *ptr) 173{ 174 struct xfs_agf *agf = cur->bc_ag.agbp->b_addr; 175 176 ASSERT(cur->bc_ag.pag->pag_agno == be32_to_cpu(agf->agf_seqno)); 177 178 ptr->s = agf->agf_refcount_root; 179} 180 181STATIC int64_t 182xfs_refcountbt_key_diff( 183 struct xfs_btree_cur *cur, 184 const union xfs_btree_key *key) 185{ 186 const struct xfs_refcount_key *kp = &key->refc; 187 const struct xfs_refcount_irec *irec = &cur->bc_rec.rc; 188 uint32_t start; 189 190 start = xfs_refcount_encode_startblock(irec->rc_startblock, 191 irec->rc_domain); 192 return (int64_t)be32_to_cpu(kp->rc_startblock) - start; 193} 194 195STATIC int64_t 196xfs_refcountbt_diff_two_keys( 197 struct xfs_btree_cur *cur, 198 const union xfs_btree_key *k1, 199 const union xfs_btree_key *k2, 200 const union xfs_btree_key *mask) 201{ 202 ASSERT(!mask || mask->refc.rc_startblock); 203 204 return (int64_t)be32_to_cpu(k1->refc.rc_startblock) - 205 be32_to_cpu(k2->refc.rc_startblock); 206} 207 208STATIC xfs_failaddr_t 209xfs_refcountbt_verify( 210 struct xfs_buf *bp) 211{ 212 struct xfs_mount *mp = bp->b_mount; 213 struct xfs_btree_block *block = XFS_BUF_TO_BLOCK(bp); 214 struct xfs_perag *pag = bp->b_pag; 215 xfs_failaddr_t fa; 216 unsigned int level; 217 218 if (!xfs_verify_magic(bp, block->bb_magic)) 219 return __this_address; 220 221 if (!xfs_has_reflink(mp)) 222 return __this_address; 223 fa = xfs_btree_sblock_v5hdr_verify(bp); 224 if (fa) 225 return fa; 226 227 level = be16_to_cpu(block->bb_level); 228 if (pag && xfs_perag_initialised_agf(pag)) { 229 if (level >= pag->pagf_refcount_level) 230 return __this_address; 231 } else if (level >= mp->m_refc_maxlevels) 232 return __this_address; 233 234 return xfs_btree_sblock_verify(bp, mp->m_refc_mxr[level != 0]); 235} 236 237STATIC void 238xfs_refcountbt_read_verify( 239 struct xfs_buf *bp) 240{ 241 xfs_failaddr_t fa; 242 243 if (!xfs_btree_sblock_verify_crc(bp)) 244 xfs_verifier_error(bp, -EFSBADCRC, __this_address); 245 else { 246 fa = xfs_refcountbt_verify(bp); 247 if (fa) 248 xfs_verifier_error(bp, -EFSCORRUPTED, fa); 249 } 250 251 if (bp->b_error) 252 trace_xfs_btree_corrupt(bp, _RET_IP_); 253} 254 255STATIC void 256xfs_refcountbt_write_verify( 257 struct xfs_buf *bp) 258{ 259 xfs_failaddr_t fa; 260 261 fa = xfs_refcountbt_verify(bp); 262 if (fa) { 263 trace_xfs_btree_corrupt(bp, _RET_IP_); 264 xfs_verifier_error(bp, -EFSCORRUPTED, fa); 265 return; 266 } 267 xfs_btree_sblock_calc_crc(bp); 268 269} 270 271const struct xfs_buf_ops xfs_refcountbt_buf_ops = { 272 .name = "xfs_refcountbt", 273 .magic = { 0, cpu_to_be32(XFS_REFC_CRC_MAGIC) }, 274 .verify_read = xfs_refcountbt_read_verify, 275 .verify_write = xfs_refcountbt_write_verify, 276 .verify_struct = xfs_refcountbt_verify, 277}; 278 279STATIC int 280xfs_refcountbt_keys_inorder( 281 struct xfs_btree_cur *cur, 282 const union xfs_btree_key *k1, 283 const union xfs_btree_key *k2) 284{ 285 return be32_to_cpu(k1->refc.rc_startblock) < 286 be32_to_cpu(k2->refc.rc_startblock); 287} 288 289STATIC int 290xfs_refcountbt_recs_inorder( 291 struct xfs_btree_cur *cur, 292 const union xfs_btree_rec *r1, 293 const union xfs_btree_rec *r2) 294{ 295 return be32_to_cpu(r1->refc.rc_startblock) + 296 be32_to_cpu(r1->refc.rc_blockcount) <= 297 be32_to_cpu(r2->refc.rc_startblock); 298} 299 300STATIC enum xbtree_key_contig 301xfs_refcountbt_keys_contiguous( 302 struct xfs_btree_cur *cur, 303 const union xfs_btree_key *key1, 304 const union xfs_btree_key *key2, 305 const union xfs_btree_key *mask) 306{ 307 ASSERT(!mask || mask->refc.rc_startblock); 308 309 return xbtree_key_contig(be32_to_cpu(key1->refc.rc_startblock), 310 be32_to_cpu(key2->refc.rc_startblock)); 311} 312 313static const struct xfs_btree_ops xfs_refcountbt_ops = { 314 .rec_len = sizeof(struct xfs_refcount_rec), 315 .key_len = sizeof(struct xfs_refcount_key), 316 317 .dup_cursor = xfs_refcountbt_dup_cursor, 318 .set_root = xfs_refcountbt_set_root, 319 .alloc_block = xfs_refcountbt_alloc_block, 320 .free_block = xfs_refcountbt_free_block, 321 .get_minrecs = xfs_refcountbt_get_minrecs, 322 .get_maxrecs = xfs_refcountbt_get_maxrecs, 323 .init_key_from_rec = xfs_refcountbt_init_key_from_rec, 324 .init_high_key_from_rec = xfs_refcountbt_init_high_key_from_rec, 325 .init_rec_from_cur = xfs_refcountbt_init_rec_from_cur, 326 .init_ptr_from_cur = xfs_refcountbt_init_ptr_from_cur, 327 .key_diff = xfs_refcountbt_key_diff, 328 .buf_ops = &xfs_refcountbt_buf_ops, 329 .diff_two_keys = xfs_refcountbt_diff_two_keys, 330 .keys_inorder = xfs_refcountbt_keys_inorder, 331 .recs_inorder = xfs_refcountbt_recs_inorder, 332 .keys_contiguous = xfs_refcountbt_keys_contiguous, 333}; 334 335/* 336 * Initialize a new refcount btree cursor. 337 */ 338static struct xfs_btree_cur * 339xfs_refcountbt_init_common( 340 struct xfs_mount *mp, 341 struct xfs_trans *tp, 342 struct xfs_perag *pag) 343{ 344 struct xfs_btree_cur *cur; 345 346 ASSERT(pag->pag_agno < mp->m_sb.sb_agcount); 347 348 cur = xfs_btree_alloc_cursor(mp, tp, XFS_BTNUM_REFC, 349 mp->m_refc_maxlevels, xfs_refcountbt_cur_cache); 350 cur->bc_statoff = XFS_STATS_CALC_INDEX(xs_refcbt_2); 351 352 cur->bc_flags |= XFS_BTREE_CRC_BLOCKS; 353 354 cur->bc_ag.pag = xfs_perag_hold(pag); 355 cur->bc_ag.refc.nr_ops = 0; 356 cur->bc_ag.refc.shape_changes = 0; 357 cur->bc_ops = &xfs_refcountbt_ops; 358 return cur; 359} 360 361/* Create a btree cursor. */ 362struct xfs_btree_cur * 363xfs_refcountbt_init_cursor( 364 struct xfs_mount *mp, 365 struct xfs_trans *tp, 366 struct xfs_buf *agbp, 367 struct xfs_perag *pag) 368{ 369 struct xfs_agf *agf = agbp->b_addr; 370 struct xfs_btree_cur *cur; 371 372 cur = xfs_refcountbt_init_common(mp, tp, pag); 373 cur->bc_nlevels = be32_to_cpu(agf->agf_refcount_level); 374 cur->bc_ag.agbp = agbp; 375 return cur; 376} 377 378/* Create a btree cursor with a fake root for staging. */ 379struct xfs_btree_cur * 380xfs_refcountbt_stage_cursor( 381 struct xfs_mount *mp, 382 struct xbtree_afakeroot *afake, 383 struct xfs_perag *pag) 384{ 385 struct xfs_btree_cur *cur; 386 387 cur = xfs_refcountbt_init_common(mp, NULL, pag); 388 xfs_btree_stage_afakeroot(cur, afake); 389 return cur; 390} 391 392/* 393 * Swap in the new btree root. Once we pass this point the newly rebuilt btree 394 * is in place and we have to kill off all the old btree blocks. 395 */ 396void 397xfs_refcountbt_commit_staged_btree( 398 struct xfs_btree_cur *cur, 399 struct xfs_trans *tp, 400 struct xfs_buf *agbp) 401{ 402 struct xfs_agf *agf = agbp->b_addr; 403 struct xbtree_afakeroot *afake = cur->bc_ag.afake; 404 405 ASSERT(cur->bc_flags & XFS_BTREE_STAGING); 406 407 agf->agf_refcount_root = cpu_to_be32(afake->af_root); 408 agf->agf_refcount_level = cpu_to_be32(afake->af_levels); 409 agf->agf_refcount_blocks = cpu_to_be32(afake->af_blocks); 410 xfs_alloc_log_agf(tp, agbp, XFS_AGF_REFCOUNT_BLOCKS | 411 XFS_AGF_REFCOUNT_ROOT | 412 XFS_AGF_REFCOUNT_LEVEL); 413 xfs_btree_commit_afakeroot(cur, tp, agbp, &xfs_refcountbt_ops); 414} 415 416/* Calculate number of records in a refcount btree block. */ 417static inline unsigned int 418xfs_refcountbt_block_maxrecs( 419 unsigned int blocklen, 420 bool leaf) 421{ 422 if (leaf) 423 return blocklen / sizeof(struct xfs_refcount_rec); 424 return blocklen / (sizeof(struct xfs_refcount_key) + 425 sizeof(xfs_refcount_ptr_t)); 426} 427 428/* 429 * Calculate the number of records in a refcount btree block. 430 */ 431int 432xfs_refcountbt_maxrecs( 433 int blocklen, 434 bool leaf) 435{ 436 blocklen -= XFS_REFCOUNT_BLOCK_LEN; 437 return xfs_refcountbt_block_maxrecs(blocklen, leaf); 438} 439 440/* Compute the max possible height of the maximally sized refcount btree. */ 441unsigned int 442xfs_refcountbt_maxlevels_ondisk(void) 443{ 444 unsigned int minrecs[2]; 445 unsigned int blocklen; 446 447 blocklen = XFS_MIN_CRC_BLOCKSIZE - XFS_BTREE_SBLOCK_CRC_LEN; 448 449 minrecs[0] = xfs_refcountbt_block_maxrecs(blocklen, true) / 2; 450 minrecs[1] = xfs_refcountbt_block_maxrecs(blocklen, false) / 2; 451 452 return xfs_btree_compute_maxlevels(minrecs, XFS_MAX_CRC_AG_BLOCKS); 453} 454 455/* Compute the maximum height of a refcount btree. */ 456void 457xfs_refcountbt_compute_maxlevels( 458 struct xfs_mount *mp) 459{ 460 if (!xfs_has_reflink(mp)) { 461 mp->m_refc_maxlevels = 0; 462 return; 463 } 464 465 mp->m_refc_maxlevels = xfs_btree_compute_maxlevels( 466 mp->m_refc_mnr, mp->m_sb.sb_agblocks); 467 ASSERT(mp->m_refc_maxlevels <= xfs_refcountbt_maxlevels_ondisk()); 468} 469 470/* Calculate the refcount btree size for some records. */ 471xfs_extlen_t 472xfs_refcountbt_calc_size( 473 struct xfs_mount *mp, 474 unsigned long long len) 475{ 476 return xfs_btree_calc_size(mp->m_refc_mnr, len); 477} 478 479/* 480 * Calculate the maximum refcount btree size. 481 */ 482xfs_extlen_t 483xfs_refcountbt_max_size( 484 struct xfs_mount *mp, 485 xfs_agblock_t agblocks) 486{ 487 /* Bail out if we're uninitialized, which can happen in mkfs. */ 488 if (mp->m_refc_mxr[0] == 0) 489 return 0; 490 491 return xfs_refcountbt_calc_size(mp, agblocks); 492} 493 494/* 495 * Figure out how many blocks to reserve and how many are used by this btree. 496 */ 497int 498xfs_refcountbt_calc_reserves( 499 struct xfs_mount *mp, 500 struct xfs_trans *tp, 501 struct xfs_perag *pag, 502 xfs_extlen_t *ask, 503 xfs_extlen_t *used) 504{ 505 struct xfs_buf *agbp; 506 struct xfs_agf *agf; 507 xfs_agblock_t agblocks; 508 xfs_extlen_t tree_len; 509 int error; 510 511 if (!xfs_has_reflink(mp)) 512 return 0; 513 514 error = xfs_alloc_read_agf(pag, tp, 0, &agbp); 515 if (error) 516 return error; 517 518 agf = agbp->b_addr; 519 agblocks = be32_to_cpu(agf->agf_length); 520 tree_len = be32_to_cpu(agf->agf_refcount_blocks); 521 xfs_trans_brelse(tp, agbp); 522 523 /* 524 * The log is permanently allocated, so the space it occupies will 525 * never be available for the kinds of things that would require btree 526 * expansion. We therefore can pretend the space isn't there. 527 */ 528 if (xfs_ag_contains_log(mp, pag->pag_agno)) 529 agblocks -= mp->m_sb.sb_logblocks; 530 531 *ask += xfs_refcountbt_max_size(mp, agblocks); 532 *used += tree_len; 533 534 return error; 535} 536 537int __init 538xfs_refcountbt_init_cur_cache(void) 539{ 540 xfs_refcountbt_cur_cache = kmem_cache_create("xfs_refcbt_cur", 541 xfs_btree_cur_sizeof(xfs_refcountbt_maxlevels_ondisk()), 542 0, 0, NULL); 543 544 if (!xfs_refcountbt_cur_cache) 545 return -ENOMEM; 546 return 0; 547} 548 549void 550xfs_refcountbt_destroy_cur_cache(void) 551{ 552 kmem_cache_destroy(xfs_refcountbt_cur_cache); 553 xfs_refcountbt_cur_cache = NULL; 554} 555