xfs_alloc_btree.c 13.0 KB
Newer Older
D
Dave Chinner 已提交
1
// SPDX-License-Identifier: GPL-2.0
L
Linus Torvalds 已提交
2
/*
3 4
 * Copyright (c) 2000-2001,2005 Silicon Graphics, Inc.
 * All Rights Reserved.
L
Linus Torvalds 已提交
5 6
 */
#include "xfs.h"
7
#include "xfs_fs.h"
8
#include "xfs_shared.h"
9
#include "xfs_format.h"
10 11
#include "xfs_log_format.h"
#include "xfs_trans_resv.h"
L
Linus Torvalds 已提交
12 13 14
#include "xfs_sb.h"
#include "xfs_mount.h"
#include "xfs_btree.h"
15
#include "xfs_alloc_btree.h"
L
Linus Torvalds 已提交
16
#include "xfs_alloc.h"
17
#include "xfs_extent_busy.h"
L
Linus Torvalds 已提交
18
#include "xfs_error.h"
C
Christoph Hellwig 已提交
19
#include "xfs_trace.h"
20
#include "xfs_trans.h"
L
Linus Torvalds 已提交
21 22


23 24 25 26 27 28 29 30 31
STATIC struct xfs_btree_cur *
xfs_allocbt_dup_cursor(
	struct xfs_btree_cur	*cur)
{
	return xfs_allocbt_init_cursor(cur->bc_mp, cur->bc_tp,
			cur->bc_private.a.agbp, cur->bc_private.a.agno,
			cur->bc_btnum);
}

32 33 34 35 36 37 38 39 40 41
STATIC void
xfs_allocbt_set_root(
	struct xfs_btree_cur	*cur,
	union xfs_btree_ptr	*ptr,
	int			inc)
{
	struct xfs_buf		*agbp = cur->bc_private.a.agbp;
	struct xfs_agf		*agf = XFS_BUF_TO_AGF(agbp);
	xfs_agnumber_t		seqno = be32_to_cpu(agf->agf_seqno);
	int			btnum = cur->bc_btnum;
42
	struct xfs_perag	*pag = xfs_perag_get(cur->bc_mp, seqno);
43 44 45 46 47

	ASSERT(ptr->s != 0);

	agf->agf_roots[btnum] = ptr->s;
	be32_add_cpu(&agf->agf_levels[btnum], inc);
48 49
	pag->pagf_levels[btnum] += inc;
	xfs_perag_put(pag);
50 51 52 53

	xfs_alloc_log_agf(cur->bc_tp, agbp, XFS_AGF_ROOTS | XFS_AGF_LEVELS);
}

54 55 56 57 58 59 60 61 62 63 64 65 66
STATIC int
xfs_allocbt_alloc_block(
	struct xfs_btree_cur	*cur,
	union xfs_btree_ptr	*start,
	union xfs_btree_ptr	*new,
	int			*stat)
{
	int			error;
	xfs_agblock_t		bno;

	/* Allocate the new block from the freelist. If we can't, give up.  */
	error = xfs_alloc_get_freelist(cur->bc_tp, cur->bc_private.a.agbp,
				       &bno, 1);
C
Carlos Maiolino 已提交
67
	if (error)
68 69 70 71 72 73
		return error;

	if (bno == NULLAGBLOCK) {
		*stat = 0;
		return 0;
	}
74

D
Dave Chinner 已提交
75
	xfs_extent_busy_reuse(cur->bc_mp, cur->bc_private.a.agno, bno, 1, false);
76 77 78 79 80 81 82 83

	xfs_trans_agbtree_delta(cur->bc_tp, 1);
	new->s = cpu_to_be32(bno);

	*stat = 1;
	return 0;
}

84 85 86 87 88 89 90 91 92 93
STATIC int
xfs_allocbt_free_block(
	struct xfs_btree_cur	*cur,
	struct xfs_buf		*bp)
{
	struct xfs_buf		*agbp = cur->bc_private.a.agbp;
	struct xfs_agf		*agf = XFS_BUF_TO_AGF(agbp);
	xfs_agblock_t		bno;
	int			error;

94
	bno = xfs_daddr_to_agbno(cur->bc_mp, XFS_BUF_ADDR(bp));
95 96 97 98
	error = xfs_alloc_put_freelist(cur->bc_tp, agbp, NULL, bno, 1);
	if (error)
		return error;

D
Dave Chinner 已提交
99 100
	xfs_extent_busy_insert(cur->bc_tp, be32_to_cpu(agf->agf_seqno), bno, 1,
			      XFS_EXTENT_BUSY_SKIP_DISCARD);
101 102 103 104
	xfs_trans_agbtree_delta(cur->bc_tp, -1);
	return 0;
}

L
Linus Torvalds 已提交
105
/*
106
 * Update the longest extent in the AGF
L
Linus Torvalds 已提交
107
 */
108 109 110 111 112 113 114
STATIC void
xfs_allocbt_update_lastrec(
	struct xfs_btree_cur	*cur,
	struct xfs_btree_block	*block,
	union xfs_btree_rec	*rec,
	int			ptr,
	int			reason)
L
Linus Torvalds 已提交
115
{
116 117
	struct xfs_agf		*agf = XFS_BUF_TO_AGF(cur->bc_private.a.agbp);
	xfs_agnumber_t		seqno = be32_to_cpu(agf->agf_seqno);
118
	struct xfs_perag	*pag;
119
	__be32			len;
120
	int			numrecs;
L
Linus Torvalds 已提交
121

122
	ASSERT(cur->bc_btnum == XFS_BTNUM_CNT);
L
Linus Torvalds 已提交
123

124 125
	switch (reason) {
	case LASTREC_UPDATE:
L
Linus Torvalds 已提交
126
		/*
127 128
		 * If this is the last leaf block and it's the last record,
		 * then update the size of the longest extent in the AG.
L
Linus Torvalds 已提交
129
		 */
130 131 132 133
		if (ptr != xfs_btree_get_numrecs(block))
			return;
		len = rec->alloc.ar_blockcount;
		break;
134 135 136 137 138
	case LASTREC_INSREC:
		if (be32_to_cpu(rec->alloc.ar_blockcount) <=
		    be32_to_cpu(agf->agf_longest))
			return;
		len = rec->alloc.ar_blockcount;
139 140 141 142 143 144 145 146 147 148
		break;
	case LASTREC_DELREC:
		numrecs = xfs_btree_get_numrecs(block);
		if (ptr <= numrecs)
			return;
		ASSERT(ptr == numrecs + 1);

		if (numrecs) {
			xfs_alloc_rec_t *rrp;

149
			rrp = XFS_ALLOC_REC_ADDR(cur->bc_mp, block, numrecs);
150 151 152 153 154
			len = rrp->ar_blockcount;
		} else {
			len = 0;
		}

155
		break;
156 157 158
	default:
		ASSERT(0);
		return;
L
Linus Torvalds 已提交
159
	}
160

161
	agf->agf_longest = len;
162 163 164
	pag = xfs_perag_get(cur->bc_mp, seqno);
	pag->pagf_longest = be32_to_cpu(len);
	xfs_perag_put(pag);
165
	xfs_alloc_log_agf(cur->bc_tp, cur->bc_private.a.agbp, XFS_AGF_LONGEST);
166 167
}

168 169 170 171 172 173 174 175
STATIC int
xfs_allocbt_get_minrecs(
	struct xfs_btree_cur	*cur,
	int			level)
{
	return cur->bc_mp->m_alloc_mnr[level != 0];
}

176 177 178 179 180 181 182 183
STATIC int
xfs_allocbt_get_maxrecs(
	struct xfs_btree_cur	*cur,
	int			level)
{
	return cur->bc_mp->m_alloc_mxr[level != 0];
}

184 185 186 187 188 189 190 191 192
STATIC void
xfs_allocbt_init_key_from_rec(
	union xfs_btree_key	*key,
	union xfs_btree_rec	*rec)
{
	key->alloc.ar_startblock = rec->alloc.ar_startblock;
	key->alloc.ar_blockcount = rec->alloc.ar_blockcount;
}

193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214
STATIC void
xfs_bnobt_init_high_key_from_rec(
	union xfs_btree_key	*key,
	union xfs_btree_rec	*rec)
{
	__u32			x;

	x = be32_to_cpu(rec->alloc.ar_startblock);
	x += be32_to_cpu(rec->alloc.ar_blockcount) - 1;
	key->alloc.ar_startblock = cpu_to_be32(x);
	key->alloc.ar_blockcount = 0;
}

STATIC void
xfs_cntbt_init_high_key_from_rec(
	union xfs_btree_key	*key,
	union xfs_btree_rec	*rec)
{
	key->alloc.ar_blockcount = rec->alloc.ar_blockcount;
	key->alloc.ar_startblock = 0;
}

215 216 217 218 219 220 221 222 223
STATIC void
xfs_allocbt_init_rec_from_cur(
	struct xfs_btree_cur	*cur,
	union xfs_btree_rec	*rec)
{
	rec->alloc.ar_startblock = cpu_to_be32(cur->bc_rec.a.ar_startblock);
	rec->alloc.ar_blockcount = cpu_to_be32(cur->bc_rec.a.ar_blockcount);
}

224 225 226 227 228 229 230 231 232 233 234 235
STATIC void
xfs_allocbt_init_ptr_from_cur(
	struct xfs_btree_cur	*cur,
	union xfs_btree_ptr	*ptr)
{
	struct xfs_agf		*agf = XFS_BUF_TO_AGF(cur->bc_private.a.agbp);

	ASSERT(cur->bc_private.a.agno == be32_to_cpu(agf->agf_seqno));

	ptr->s = agf->agf_roots[cur->bc_btnum];
}

236
STATIC int64_t
237
xfs_bnobt_key_diff(
238 239 240 241 242 243
	struct xfs_btree_cur	*cur,
	union xfs_btree_key	*key)
{
	xfs_alloc_rec_incore_t	*rec = &cur->bc_rec.a;
	xfs_alloc_key_t		*kp = &key->alloc;

244
	return (int64_t)be32_to_cpu(kp->ar_startblock) - rec->ar_startblock;
245 246
}

247
STATIC int64_t
248 249 250 251 252 253
xfs_cntbt_key_diff(
	struct xfs_btree_cur	*cur,
	union xfs_btree_key	*key)
{
	xfs_alloc_rec_incore_t	*rec = &cur->bc_rec.a;
	xfs_alloc_key_t		*kp = &key->alloc;
254
	int64_t			diff;
255

256
	diff = (int64_t)be32_to_cpu(kp->ar_blockcount) - rec->ar_blockcount;
257 258 259
	if (diff)
		return diff;

260
	return (int64_t)be32_to_cpu(kp->ar_startblock) - rec->ar_startblock;
261 262
}

263
STATIC int64_t
264 265 266 267 268
xfs_bnobt_diff_two_keys(
	struct xfs_btree_cur	*cur,
	union xfs_btree_key	*k1,
	union xfs_btree_key	*k2)
{
269
	return (int64_t)be32_to_cpu(k1->alloc.ar_startblock) -
270 271 272
			  be32_to_cpu(k2->alloc.ar_startblock);
}

273
STATIC int64_t
274 275 276 277 278
xfs_cntbt_diff_two_keys(
	struct xfs_btree_cur	*cur,
	union xfs_btree_key	*k1,
	union xfs_btree_key	*k2)
{
279
	int64_t			diff;
280 281 282 283 284 285 286 287 288 289

	diff =  be32_to_cpu(k1->alloc.ar_blockcount) -
		be32_to_cpu(k2->alloc.ar_blockcount);
	if (diff)
		return diff;

	return  be32_to_cpu(k1->alloc.ar_startblock) -
		be32_to_cpu(k2->alloc.ar_startblock);
}

290
static xfs_failaddr_t
291
xfs_allocbt_verify(
292 293
	struct xfs_buf		*bp)
{
294
	struct xfs_mount	*mp = bp->b_mount;
295 296
	struct xfs_btree_block	*block = XFS_BUF_TO_BLOCK(bp);
	struct xfs_perag	*pag = bp->b_pag;
297
	xfs_failaddr_t		fa;
298
	unsigned int		level;
299 300 301 302 303 304 305 306 307 308
	xfs_btnum_t		btnum = XFS_BTNUM_BNOi;

	if (!xfs_verify_magic(bp, block->bb_magic))
		return __this_address;

	if (xfs_sb_version_hascrc(&mp->m_sb)) {
		fa = xfs_btree_sblock_v5hdr_verify(bp);
		if (fa)
			return fa;
	}
309 310

	/*
311 312 313
	 * The perag may not be attached during grow operations or fully
	 * initialized from the AGF during log recovery. Therefore we can only
	 * check against maximum tree depth from those contexts.
314
	 *
315 316 317
	 * Otherwise check against the per-tree limit. Peek at one of the
	 * verifier magic values to determine the type of tree we're verifying
	 * against.
318 319
	 */
	level = be16_to_cpu(block->bb_level);
320 321 322 323
	if (bp->b_ops->magic[0] == cpu_to_be32(XFS_ABTC_MAGIC))
		btnum = XFS_BTNUM_CNTi;
	if (pag && pag->pagf_init) {
		if (level >= pag->pagf_levels[btnum])
324
			return __this_address;
325
	} else if (level >= mp->m_ag_maxlevels)
326
		return __this_address;
327

328
	return xfs_btree_sblock_verify(bp, mp->m_alloc_mxr[level != 0]);
329
}
330

331
static void
332
xfs_allocbt_read_verify(
333 334
	struct xfs_buf	*bp)
{
335 336
	xfs_failaddr_t	fa;

337
	if (!xfs_btree_sblock_verify_crc(bp))
338 339 340 341 342 343
		xfs_verifier_error(bp, -EFSBADCRC, __this_address);
	else {
		fa = xfs_allocbt_verify(bp);
		if (fa)
			xfs_verifier_error(bp, -EFSCORRUPTED, fa);
	}
344

345
	if (bp->b_error)
346
		trace_xfs_btree_corrupt(bp, _RET_IP_);
347 348
}

349 350
static void
xfs_allocbt_write_verify(
351 352
	struct xfs_buf	*bp)
{
353 354 355 356
	xfs_failaddr_t	fa;

	fa = xfs_allocbt_verify(bp);
	if (fa) {
357
		trace_xfs_btree_corrupt(bp, _RET_IP_);
358
		xfs_verifier_error(bp, -EFSCORRUPTED, fa);
359
		return;
360 361 362
	}
	xfs_btree_sblock_calc_crc(bp);

363 364
}

365 366
const struct xfs_buf_ops xfs_bnobt_buf_ops = {
	.name = "xfs_bnobt",
367 368
	.magic = { cpu_to_be32(XFS_ABTB_MAGIC),
		   cpu_to_be32(XFS_ABTB_CRC_MAGIC) },
369 370
	.verify_read = xfs_allocbt_read_verify,
	.verify_write = xfs_allocbt_write_verify,
371
	.verify_struct = xfs_allocbt_verify,
372 373
};

374 375
const struct xfs_buf_ops xfs_cntbt_buf_ops = {
	.name = "xfs_cntbt",
376 377
	.magic = { cpu_to_be32(XFS_ABTC_MAGIC),
		   cpu_to_be32(XFS_ABTC_CRC_MAGIC) },
378 379 380 381
	.verify_read = xfs_allocbt_read_verify,
	.verify_write = xfs_allocbt_write_verify,
	.verify_struct = xfs_allocbt_verify,
};
382

383
STATIC int
384
xfs_bnobt_keys_inorder(
385 386 387 388
	struct xfs_btree_cur	*cur,
	union xfs_btree_key	*k1,
	union xfs_btree_key	*k2)
{
389 390
	return be32_to_cpu(k1->alloc.ar_startblock) <
	       be32_to_cpu(k2->alloc.ar_startblock);
391 392 393
}

STATIC int
394
xfs_bnobt_recs_inorder(
395 396 397 398
	struct xfs_btree_cur	*cur,
	union xfs_btree_rec	*r1,
	union xfs_btree_rec	*r2)
{
399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414
	return be32_to_cpu(r1->alloc.ar_startblock) +
		be32_to_cpu(r1->alloc.ar_blockcount) <=
		be32_to_cpu(r2->alloc.ar_startblock);
}

STATIC int
xfs_cntbt_keys_inorder(
	struct xfs_btree_cur	*cur,
	union xfs_btree_key	*k1,
	union xfs_btree_key	*k2)
{
	return be32_to_cpu(k1->alloc.ar_blockcount) <
		be32_to_cpu(k2->alloc.ar_blockcount) ||
		(k1->alloc.ar_blockcount == k2->alloc.ar_blockcount &&
		 be32_to_cpu(k1->alloc.ar_startblock) <
		 be32_to_cpu(k2->alloc.ar_startblock));
415 416
}

417 418 419 420 421 422 423 424 425 426 427 428 429 430
STATIC int
xfs_cntbt_recs_inorder(
	struct xfs_btree_cur	*cur,
	union xfs_btree_rec	*r1,
	union xfs_btree_rec	*r2)
{
	return be32_to_cpu(r1->alloc.ar_blockcount) <
		be32_to_cpu(r2->alloc.ar_blockcount) ||
		(r1->alloc.ar_blockcount == r2->alloc.ar_blockcount &&
		 be32_to_cpu(r1->alloc.ar_startblock) <
		 be32_to_cpu(r2->alloc.ar_startblock));
}

static const struct xfs_btree_ops xfs_bnobt_ops = {
431 432 433
	.rec_len		= sizeof(xfs_alloc_rec_t),
	.key_len		= sizeof(xfs_alloc_key_t),

434
	.dup_cursor		= xfs_allocbt_dup_cursor,
435
	.set_root		= xfs_allocbt_set_root,
436
	.alloc_block		= xfs_allocbt_alloc_block,
437
	.free_block		= xfs_allocbt_free_block,
438
	.update_lastrec		= xfs_allocbt_update_lastrec,
439
	.get_minrecs		= xfs_allocbt_get_minrecs,
440
	.get_maxrecs		= xfs_allocbt_get_maxrecs,
441
	.init_key_from_rec	= xfs_allocbt_init_key_from_rec,
442
	.init_high_key_from_rec	= xfs_bnobt_init_high_key_from_rec,
443
	.init_rec_from_cur	= xfs_allocbt_init_rec_from_cur,
444
	.init_ptr_from_cur	= xfs_allocbt_init_ptr_from_cur,
445
	.key_diff		= xfs_bnobt_key_diff,
446
	.buf_ops		= &xfs_bnobt_buf_ops,
447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467
	.diff_two_keys		= xfs_bnobt_diff_two_keys,
	.keys_inorder		= xfs_bnobt_keys_inorder,
	.recs_inorder		= xfs_bnobt_recs_inorder,
};

static const struct xfs_btree_ops xfs_cntbt_ops = {
	.rec_len		= sizeof(xfs_alloc_rec_t),
	.key_len		= sizeof(xfs_alloc_key_t),

	.dup_cursor		= xfs_allocbt_dup_cursor,
	.set_root		= xfs_allocbt_set_root,
	.alloc_block		= xfs_allocbt_alloc_block,
	.free_block		= xfs_allocbt_free_block,
	.update_lastrec		= xfs_allocbt_update_lastrec,
	.get_minrecs		= xfs_allocbt_get_minrecs,
	.get_maxrecs		= xfs_allocbt_get_maxrecs,
	.init_key_from_rec	= xfs_allocbt_init_key_from_rec,
	.init_high_key_from_rec	= xfs_cntbt_init_high_key_from_rec,
	.init_rec_from_cur	= xfs_allocbt_init_rec_from_cur,
	.init_ptr_from_cur	= xfs_allocbt_init_ptr_from_cur,
	.key_diff		= xfs_cntbt_key_diff,
468
	.buf_ops		= &xfs_cntbt_buf_ops,
469 470 471
	.diff_two_keys		= xfs_cntbt_diff_two_keys,
	.keys_inorder		= xfs_cntbt_keys_inorder,
	.recs_inorder		= xfs_cntbt_recs_inorder,
472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489
};

/*
 * Allocate a new allocation btree cursor.
 */
struct xfs_btree_cur *			/* new alloc btree cursor */
xfs_allocbt_init_cursor(
	struct xfs_mount	*mp,		/* file system mount point */
	struct xfs_trans	*tp,		/* transaction pointer */
	struct xfs_buf		*agbp,		/* buffer for agf structure */
	xfs_agnumber_t		agno,		/* allocation group number */
	xfs_btnum_t		btnum)		/* btree identifier */
{
	struct xfs_agf		*agf = XFS_BUF_TO_AGF(agbp);
	struct xfs_btree_cur	*cur;

	ASSERT(btnum == XFS_BTNUM_BNO || btnum == XFS_BTNUM_CNT);

490
	cur = kmem_zone_zalloc(xfs_btree_cur_zone, KM_NOFS);
491 492 493 494 495

	cur->bc_tp = tp;
	cur->bc_mp = mp;
	cur->bc_btnum = btnum;
	cur->bc_blocklog = mp->m_sb.sb_blocklog;
496 497

	if (btnum == XFS_BTNUM_CNT) {
498 499
		cur->bc_statoff = XFS_STATS_CALC_INDEX(xs_abtc_2);
		cur->bc_ops = &xfs_cntbt_ops;
500
		cur->bc_nlevels = be32_to_cpu(agf->agf_levels[XFS_BTNUM_CNT]);
501
		cur->bc_flags = XFS_BTREE_LASTREC_UPDATE;
502
	} else {
503 504
		cur->bc_statoff = XFS_STATS_CALC_INDEX(xs_abtb_2);
		cur->bc_ops = &xfs_bnobt_ops;
505 506
		cur->bc_nlevels = be32_to_cpu(agf->agf_levels[XFS_BTNUM_BNO]);
	}
507 508 509

	cur->bc_private.a.agbp = agbp;
	cur->bc_private.a.agno = agno;
510
	cur->bc_private.a.priv.abt.active = false;
511

512 513 514
	if (xfs_sb_version_hascrc(&mp->m_sb))
		cur->bc_flags |= XFS_BTREE_CRC_BLOCKS;

515 516
	return cur;
}
517 518 519 520 521 522 523 524 525 526

/*
 * Calculate number of records in an alloc btree block.
 */
int
xfs_allocbt_maxrecs(
	struct xfs_mount	*mp,
	int			blocklen,
	int			leaf)
{
527
	blocklen -= XFS_ALLOC_BLOCK_LEN(mp);
528 529 530 531 532

	if (leaf)
		return blocklen / sizeof(xfs_alloc_rec_t);
	return blocklen / (sizeof(xfs_alloc_key_t) + sizeof(xfs_alloc_ptr_t));
}
533 534 535 536 537 538 539 540 541

/* Calculate the freespace btree size for some records. */
xfs_extlen_t
xfs_allocbt_calc_size(
	struct xfs_mount	*mp,
	unsigned long long	len)
{
	return xfs_btree_calc_size(mp->m_alloc_mnr, len);
}