xref: /openbmc/linux/fs/xfs/libxfs/xfs_alloc_btree.c (revision be709d48)
1 // SPDX-License-Identifier: GPL-2.0
2 /*
3  * Copyright (c) 2000-2001,2005 Silicon Graphics, Inc.
4  * All Rights Reserved.
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_sb.h"
13 #include "xfs_mount.h"
14 #include "xfs_btree.h"
15 #include "xfs_alloc_btree.h"
16 #include "xfs_alloc.h"
17 #include "xfs_extent_busy.h"
18 #include "xfs_error.h"
19 #include "xfs_trace.h"
20 #include "xfs_cksum.h"
21 #include "xfs_trans.h"
22 
23 
24 STATIC struct xfs_btree_cur *
25 xfs_allocbt_dup_cursor(
26 	struct xfs_btree_cur	*cur)
27 {
28 	return xfs_allocbt_init_cursor(cur->bc_mp, cur->bc_tp,
29 			cur->bc_private.a.agbp, cur->bc_private.a.agno,
30 			cur->bc_btnum);
31 }
32 
33 STATIC void
34 xfs_allocbt_set_root(
35 	struct xfs_btree_cur	*cur,
36 	union xfs_btree_ptr	*ptr,
37 	int			inc)
38 {
39 	struct xfs_buf		*agbp = cur->bc_private.a.agbp;
40 	struct xfs_agf		*agf = XFS_BUF_TO_AGF(agbp);
41 	xfs_agnumber_t		seqno = be32_to_cpu(agf->agf_seqno);
42 	int			btnum = cur->bc_btnum;
43 	struct xfs_perag	*pag = xfs_perag_get(cur->bc_mp, seqno);
44 
45 	ASSERT(ptr->s != 0);
46 
47 	agf->agf_roots[btnum] = ptr->s;
48 	be32_add_cpu(&agf->agf_levels[btnum], inc);
49 	pag->pagf_levels[btnum] += inc;
50 	xfs_perag_put(pag);
51 
52 	xfs_alloc_log_agf(cur->bc_tp, agbp, XFS_AGF_ROOTS | XFS_AGF_LEVELS);
53 }
54 
55 STATIC int
56 xfs_allocbt_alloc_block(
57 	struct xfs_btree_cur	*cur,
58 	union xfs_btree_ptr	*start,
59 	union xfs_btree_ptr	*new,
60 	int			*stat)
61 {
62 	int			error;
63 	xfs_agblock_t		bno;
64 
65 	/* Allocate the new block from the freelist. If we can't, give up.  */
66 	error = xfs_alloc_get_freelist(cur->bc_tp, cur->bc_private.a.agbp,
67 				       &bno, 1);
68 	if (error)
69 		return error;
70 
71 	if (bno == NULLAGBLOCK) {
72 		*stat = 0;
73 		return 0;
74 	}
75 
76 	xfs_extent_busy_reuse(cur->bc_mp, cur->bc_private.a.agno, bno, 1, false);
77 
78 	xfs_trans_agbtree_delta(cur->bc_tp, 1);
79 	new->s = cpu_to_be32(bno);
80 
81 	*stat = 1;
82 	return 0;
83 }
84 
85 STATIC int
86 xfs_allocbt_free_block(
87 	struct xfs_btree_cur	*cur,
88 	struct xfs_buf		*bp)
89 {
90 	struct xfs_buf		*agbp = cur->bc_private.a.agbp;
91 	struct xfs_agf		*agf = XFS_BUF_TO_AGF(agbp);
92 	xfs_agblock_t		bno;
93 	int			error;
94 
95 	bno = xfs_daddr_to_agbno(cur->bc_mp, XFS_BUF_ADDR(bp));
96 	error = xfs_alloc_put_freelist(cur->bc_tp, agbp, NULL, bno, 1);
97 	if (error)
98 		return error;
99 
100 	xfs_extent_busy_insert(cur->bc_tp, be32_to_cpu(agf->agf_seqno), bno, 1,
101 			      XFS_EXTENT_BUSY_SKIP_DISCARD);
102 	xfs_trans_agbtree_delta(cur->bc_tp, -1);
103 	return 0;
104 }
105 
106 /*
107  * Update the longest extent in the AGF
108  */
109 STATIC void
110 xfs_allocbt_update_lastrec(
111 	struct xfs_btree_cur	*cur,
112 	struct xfs_btree_block	*block,
113 	union xfs_btree_rec	*rec,
114 	int			ptr,
115 	int			reason)
116 {
117 	struct xfs_agf		*agf = XFS_BUF_TO_AGF(cur->bc_private.a.agbp);
118 	xfs_agnumber_t		seqno = be32_to_cpu(agf->agf_seqno);
119 	struct xfs_perag	*pag;
120 	__be32			len;
121 	int			numrecs;
122 
123 	ASSERT(cur->bc_btnum == XFS_BTNUM_CNT);
124 
125 	switch (reason) {
126 	case LASTREC_UPDATE:
127 		/*
128 		 * If this is the last leaf block and it's the last record,
129 		 * then update the size of the longest extent in the AG.
130 		 */
131 		if (ptr != xfs_btree_get_numrecs(block))
132 			return;
133 		len = rec->alloc.ar_blockcount;
134 		break;
135 	case LASTREC_INSREC:
136 		if (be32_to_cpu(rec->alloc.ar_blockcount) <=
137 		    be32_to_cpu(agf->agf_longest))
138 			return;
139 		len = rec->alloc.ar_blockcount;
140 		break;
141 	case LASTREC_DELREC:
142 		numrecs = xfs_btree_get_numrecs(block);
143 		if (ptr <= numrecs)
144 			return;
145 		ASSERT(ptr == numrecs + 1);
146 
147 		if (numrecs) {
148 			xfs_alloc_rec_t *rrp;
149 
150 			rrp = XFS_ALLOC_REC_ADDR(cur->bc_mp, block, numrecs);
151 			len = rrp->ar_blockcount;
152 		} else {
153 			len = 0;
154 		}
155 
156 		break;
157 	default:
158 		ASSERT(0);
159 		return;
160 	}
161 
162 	agf->agf_longest = len;
163 	pag = xfs_perag_get(cur->bc_mp, seqno);
164 	pag->pagf_longest = be32_to_cpu(len);
165 	xfs_perag_put(pag);
166 	xfs_alloc_log_agf(cur->bc_tp, cur->bc_private.a.agbp, XFS_AGF_LONGEST);
167 }
168 
169 STATIC int
170 xfs_allocbt_get_minrecs(
171 	struct xfs_btree_cur	*cur,
172 	int			level)
173 {
174 	return cur->bc_mp->m_alloc_mnr[level != 0];
175 }
176 
177 STATIC int
178 xfs_allocbt_get_maxrecs(
179 	struct xfs_btree_cur	*cur,
180 	int			level)
181 {
182 	return cur->bc_mp->m_alloc_mxr[level != 0];
183 }
184 
185 STATIC void
186 xfs_allocbt_init_key_from_rec(
187 	union xfs_btree_key	*key,
188 	union xfs_btree_rec	*rec)
189 {
190 	key->alloc.ar_startblock = rec->alloc.ar_startblock;
191 	key->alloc.ar_blockcount = rec->alloc.ar_blockcount;
192 }
193 
194 STATIC void
195 xfs_bnobt_init_high_key_from_rec(
196 	union xfs_btree_key	*key,
197 	union xfs_btree_rec	*rec)
198 {
199 	__u32			x;
200 
201 	x = be32_to_cpu(rec->alloc.ar_startblock);
202 	x += be32_to_cpu(rec->alloc.ar_blockcount) - 1;
203 	key->alloc.ar_startblock = cpu_to_be32(x);
204 	key->alloc.ar_blockcount = 0;
205 }
206 
207 STATIC void
208 xfs_cntbt_init_high_key_from_rec(
209 	union xfs_btree_key	*key,
210 	union xfs_btree_rec	*rec)
211 {
212 	key->alloc.ar_blockcount = rec->alloc.ar_blockcount;
213 	key->alloc.ar_startblock = 0;
214 }
215 
216 STATIC void
217 xfs_allocbt_init_rec_from_cur(
218 	struct xfs_btree_cur	*cur,
219 	union xfs_btree_rec	*rec)
220 {
221 	rec->alloc.ar_startblock = cpu_to_be32(cur->bc_rec.a.ar_startblock);
222 	rec->alloc.ar_blockcount = cpu_to_be32(cur->bc_rec.a.ar_blockcount);
223 }
224 
225 STATIC void
226 xfs_allocbt_init_ptr_from_cur(
227 	struct xfs_btree_cur	*cur,
228 	union xfs_btree_ptr	*ptr)
229 {
230 	struct xfs_agf		*agf = XFS_BUF_TO_AGF(cur->bc_private.a.agbp);
231 
232 	ASSERT(cur->bc_private.a.agno == be32_to_cpu(agf->agf_seqno));
233 
234 	ptr->s = agf->agf_roots[cur->bc_btnum];
235 }
236 
237 STATIC int64_t
238 xfs_bnobt_key_diff(
239 	struct xfs_btree_cur	*cur,
240 	union xfs_btree_key	*key)
241 {
242 	xfs_alloc_rec_incore_t	*rec = &cur->bc_rec.a;
243 	xfs_alloc_key_t		*kp = &key->alloc;
244 
245 	return (int64_t)be32_to_cpu(kp->ar_startblock) - rec->ar_startblock;
246 }
247 
248 STATIC int64_t
249 xfs_cntbt_key_diff(
250 	struct xfs_btree_cur	*cur,
251 	union xfs_btree_key	*key)
252 {
253 	xfs_alloc_rec_incore_t	*rec = &cur->bc_rec.a;
254 	xfs_alloc_key_t		*kp = &key->alloc;
255 	int64_t			diff;
256 
257 	diff = (int64_t)be32_to_cpu(kp->ar_blockcount) - rec->ar_blockcount;
258 	if (diff)
259 		return diff;
260 
261 	return (int64_t)be32_to_cpu(kp->ar_startblock) - rec->ar_startblock;
262 }
263 
264 STATIC int64_t
265 xfs_bnobt_diff_two_keys(
266 	struct xfs_btree_cur	*cur,
267 	union xfs_btree_key	*k1,
268 	union xfs_btree_key	*k2)
269 {
270 	return (int64_t)be32_to_cpu(k1->alloc.ar_startblock) -
271 			  be32_to_cpu(k2->alloc.ar_startblock);
272 }
273 
274 STATIC int64_t
275 xfs_cntbt_diff_two_keys(
276 	struct xfs_btree_cur	*cur,
277 	union xfs_btree_key	*k1,
278 	union xfs_btree_key	*k2)
279 {
280 	int64_t			diff;
281 
282 	diff =  be32_to_cpu(k1->alloc.ar_blockcount) -
283 		be32_to_cpu(k2->alloc.ar_blockcount);
284 	if (diff)
285 		return diff;
286 
287 	return  be32_to_cpu(k1->alloc.ar_startblock) -
288 		be32_to_cpu(k2->alloc.ar_startblock);
289 }
290 
291 static xfs_failaddr_t
292 xfs_allocbt_verify(
293 	struct xfs_buf		*bp)
294 {
295 	struct xfs_mount	*mp = bp->b_target->bt_mount;
296 	struct xfs_btree_block	*block = XFS_BUF_TO_BLOCK(bp);
297 	struct xfs_perag	*pag = bp->b_pag;
298 	xfs_failaddr_t		fa;
299 	unsigned int		level;
300 	xfs_btnum_t		btnum = XFS_BTNUM_BNOi;
301 
302 	if (!xfs_verify_magic(bp, block->bb_magic))
303 		return __this_address;
304 
305 	if (xfs_sb_version_hascrc(&mp->m_sb)) {
306 		fa = xfs_btree_sblock_v5hdr_verify(bp);
307 		if (fa)
308 			return fa;
309 	}
310 
311 	/*
312 	 * The perag may not be attached during grow operations or fully
313 	 * initialized from the AGF during log recovery. Therefore we can only
314 	 * check against maximum tree depth from those contexts.
315 	 *
316 	 * Otherwise check against the per-tree limit. Peek at one of the
317 	 * verifier magic values to determine the type of tree we're verifying
318 	 * against.
319 	 */
320 	level = be16_to_cpu(block->bb_level);
321 	if (bp->b_ops->magic[0] == cpu_to_be32(XFS_ABTC_MAGIC))
322 		btnum = XFS_BTNUM_CNTi;
323 	if (pag && pag->pagf_init) {
324 		if (level >= pag->pagf_levels[btnum])
325 			return __this_address;
326 	} else if (level >= mp->m_ag_maxlevels)
327 		return __this_address;
328 
329 	return xfs_btree_sblock_verify(bp, mp->m_alloc_mxr[level != 0]);
330 }
331 
332 static void
333 xfs_allocbt_read_verify(
334 	struct xfs_buf	*bp)
335 {
336 	xfs_failaddr_t	fa;
337 
338 	if (!xfs_btree_sblock_verify_crc(bp))
339 		xfs_verifier_error(bp, -EFSBADCRC, __this_address);
340 	else {
341 		fa = xfs_allocbt_verify(bp);
342 		if (fa)
343 			xfs_verifier_error(bp, -EFSCORRUPTED, fa);
344 	}
345 
346 	if (bp->b_error)
347 		trace_xfs_btree_corrupt(bp, _RET_IP_);
348 }
349 
350 static void
351 xfs_allocbt_write_verify(
352 	struct xfs_buf	*bp)
353 {
354 	xfs_failaddr_t	fa;
355 
356 	fa = xfs_allocbt_verify(bp);
357 	if (fa) {
358 		trace_xfs_btree_corrupt(bp, _RET_IP_);
359 		xfs_verifier_error(bp, -EFSCORRUPTED, fa);
360 		return;
361 	}
362 	xfs_btree_sblock_calc_crc(bp);
363 
364 }
365 
366 const struct xfs_buf_ops xfs_bnobt_buf_ops = {
367 	.name = "xfs_bnobt",
368 	.magic = { cpu_to_be32(XFS_ABTB_MAGIC),
369 		   cpu_to_be32(XFS_ABTB_CRC_MAGIC) },
370 	.verify_read = xfs_allocbt_read_verify,
371 	.verify_write = xfs_allocbt_write_verify,
372 	.verify_struct = xfs_allocbt_verify,
373 };
374 
375 const struct xfs_buf_ops xfs_cntbt_buf_ops = {
376 	.name = "xfs_cntbt",
377 	.magic = { cpu_to_be32(XFS_ABTC_MAGIC),
378 		   cpu_to_be32(XFS_ABTC_CRC_MAGIC) },
379 	.verify_read = xfs_allocbt_read_verify,
380 	.verify_write = xfs_allocbt_write_verify,
381 	.verify_struct = xfs_allocbt_verify,
382 };
383 
384 STATIC int
385 xfs_bnobt_keys_inorder(
386 	struct xfs_btree_cur	*cur,
387 	union xfs_btree_key	*k1,
388 	union xfs_btree_key	*k2)
389 {
390 	return be32_to_cpu(k1->alloc.ar_startblock) <
391 	       be32_to_cpu(k2->alloc.ar_startblock);
392 }
393 
394 STATIC int
395 xfs_bnobt_recs_inorder(
396 	struct xfs_btree_cur	*cur,
397 	union xfs_btree_rec	*r1,
398 	union xfs_btree_rec	*r2)
399 {
400 	return be32_to_cpu(r1->alloc.ar_startblock) +
401 		be32_to_cpu(r1->alloc.ar_blockcount) <=
402 		be32_to_cpu(r2->alloc.ar_startblock);
403 }
404 
405 STATIC int
406 xfs_cntbt_keys_inorder(
407 	struct xfs_btree_cur	*cur,
408 	union xfs_btree_key	*k1,
409 	union xfs_btree_key	*k2)
410 {
411 	return be32_to_cpu(k1->alloc.ar_blockcount) <
412 		be32_to_cpu(k2->alloc.ar_blockcount) ||
413 		(k1->alloc.ar_blockcount == k2->alloc.ar_blockcount &&
414 		 be32_to_cpu(k1->alloc.ar_startblock) <
415 		 be32_to_cpu(k2->alloc.ar_startblock));
416 }
417 
418 STATIC int
419 xfs_cntbt_recs_inorder(
420 	struct xfs_btree_cur	*cur,
421 	union xfs_btree_rec	*r1,
422 	union xfs_btree_rec	*r2)
423 {
424 	return be32_to_cpu(r1->alloc.ar_blockcount) <
425 		be32_to_cpu(r2->alloc.ar_blockcount) ||
426 		(r1->alloc.ar_blockcount == r2->alloc.ar_blockcount &&
427 		 be32_to_cpu(r1->alloc.ar_startblock) <
428 		 be32_to_cpu(r2->alloc.ar_startblock));
429 }
430 
431 static const struct xfs_btree_ops xfs_bnobt_ops = {
432 	.rec_len		= sizeof(xfs_alloc_rec_t),
433 	.key_len		= sizeof(xfs_alloc_key_t),
434 
435 	.dup_cursor		= xfs_allocbt_dup_cursor,
436 	.set_root		= xfs_allocbt_set_root,
437 	.alloc_block		= xfs_allocbt_alloc_block,
438 	.free_block		= xfs_allocbt_free_block,
439 	.update_lastrec		= xfs_allocbt_update_lastrec,
440 	.get_minrecs		= xfs_allocbt_get_minrecs,
441 	.get_maxrecs		= xfs_allocbt_get_maxrecs,
442 	.init_key_from_rec	= xfs_allocbt_init_key_from_rec,
443 	.init_high_key_from_rec	= xfs_bnobt_init_high_key_from_rec,
444 	.init_rec_from_cur	= xfs_allocbt_init_rec_from_cur,
445 	.init_ptr_from_cur	= xfs_allocbt_init_ptr_from_cur,
446 	.key_diff		= xfs_bnobt_key_diff,
447 	.buf_ops		= &xfs_bnobt_buf_ops,
448 	.diff_two_keys		= xfs_bnobt_diff_two_keys,
449 	.keys_inorder		= xfs_bnobt_keys_inorder,
450 	.recs_inorder		= xfs_bnobt_recs_inorder,
451 };
452 
453 static const struct xfs_btree_ops xfs_cntbt_ops = {
454 	.rec_len		= sizeof(xfs_alloc_rec_t),
455 	.key_len		= sizeof(xfs_alloc_key_t),
456 
457 	.dup_cursor		= xfs_allocbt_dup_cursor,
458 	.set_root		= xfs_allocbt_set_root,
459 	.alloc_block		= xfs_allocbt_alloc_block,
460 	.free_block		= xfs_allocbt_free_block,
461 	.update_lastrec		= xfs_allocbt_update_lastrec,
462 	.get_minrecs		= xfs_allocbt_get_minrecs,
463 	.get_maxrecs		= xfs_allocbt_get_maxrecs,
464 	.init_key_from_rec	= xfs_allocbt_init_key_from_rec,
465 	.init_high_key_from_rec	= xfs_cntbt_init_high_key_from_rec,
466 	.init_rec_from_cur	= xfs_allocbt_init_rec_from_cur,
467 	.init_ptr_from_cur	= xfs_allocbt_init_ptr_from_cur,
468 	.key_diff		= xfs_cntbt_key_diff,
469 	.buf_ops		= &xfs_cntbt_buf_ops,
470 	.diff_two_keys		= xfs_cntbt_diff_two_keys,
471 	.keys_inorder		= xfs_cntbt_keys_inorder,
472 	.recs_inorder		= xfs_cntbt_recs_inorder,
473 };
474 
475 /*
476  * Allocate a new allocation btree cursor.
477  */
478 struct xfs_btree_cur *			/* new alloc btree cursor */
479 xfs_allocbt_init_cursor(
480 	struct xfs_mount	*mp,		/* file system mount point */
481 	struct xfs_trans	*tp,		/* transaction pointer */
482 	struct xfs_buf		*agbp,		/* buffer for agf structure */
483 	xfs_agnumber_t		agno,		/* allocation group number */
484 	xfs_btnum_t		btnum)		/* btree identifier */
485 {
486 	struct xfs_agf		*agf = XFS_BUF_TO_AGF(agbp);
487 	struct xfs_btree_cur	*cur;
488 
489 	ASSERT(btnum == XFS_BTNUM_BNO || btnum == XFS_BTNUM_CNT);
490 
491 	cur = kmem_zone_zalloc(xfs_btree_cur_zone, KM_NOFS);
492 
493 	cur->bc_tp = tp;
494 	cur->bc_mp = mp;
495 	cur->bc_btnum = btnum;
496 	cur->bc_blocklog = mp->m_sb.sb_blocklog;
497 
498 	if (btnum == XFS_BTNUM_CNT) {
499 		cur->bc_statoff = XFS_STATS_CALC_INDEX(xs_abtc_2);
500 		cur->bc_ops = &xfs_cntbt_ops;
501 		cur->bc_nlevels = be32_to_cpu(agf->agf_levels[XFS_BTNUM_CNT]);
502 		cur->bc_flags = XFS_BTREE_LASTREC_UPDATE;
503 	} else {
504 		cur->bc_statoff = XFS_STATS_CALC_INDEX(xs_abtb_2);
505 		cur->bc_ops = &xfs_bnobt_ops;
506 		cur->bc_nlevels = be32_to_cpu(agf->agf_levels[XFS_BTNUM_BNO]);
507 	}
508 
509 	cur->bc_private.a.agbp = agbp;
510 	cur->bc_private.a.agno = agno;
511 
512 	if (xfs_sb_version_hascrc(&mp->m_sb))
513 		cur->bc_flags |= XFS_BTREE_CRC_BLOCKS;
514 
515 	return cur;
516 }
517 
518 /*
519  * Calculate number of records in an alloc btree block.
520  */
521 int
522 xfs_allocbt_maxrecs(
523 	struct xfs_mount	*mp,
524 	int			blocklen,
525 	int			leaf)
526 {
527 	blocklen -= XFS_ALLOC_BLOCK_LEN(mp);
528 
529 	if (leaf)
530 		return blocklen / sizeof(xfs_alloc_rec_t);
531 	return blocklen / (sizeof(xfs_alloc_key_t) + sizeof(xfs_alloc_ptr_t));
532 }
533 
534 /* Calculate the freespace btree size for some records. */
535 xfs_extlen_t
536 xfs_allocbt_calc_size(
537 	struct xfs_mount	*mp,
538 	unsigned long long	len)
539 {
540 	return xfs_btree_calc_size(mp->m_alloc_mnr, len);
541 }
542