1 // SPDX-License-Identifier: GPL-2.0+
2 /*
3 * Copyright (C) 2017 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_btree.h"
11 #include "xfs_rmap.h"
12 #include "xfs_refcount.h"
13 #include "scrub/scrub.h"
14 #include "scrub/common.h"
15 #include "scrub/btree.h"
16
17 /*
18 * Set us up to scrub reference count btrees.
19 */
20 int
xchk_setup_ag_refcountbt(struct xfs_scrub * sc,struct xfs_inode * ip)21 xchk_setup_ag_refcountbt(
22 struct xfs_scrub *sc,
23 struct xfs_inode *ip)
24 {
25 return xchk_setup_ag_btree(sc, ip, false);
26 }
27
28 /* Reference count btree scrubber. */
29
30 /*
31 * Confirming Reference Counts via Reverse Mappings
32 *
33 * We want to count the reverse mappings overlapping a refcount record
34 * (bno, len, refcount), allowing for the possibility that some of the
35 * overlap may come from smaller adjoining reverse mappings, while some
36 * comes from single extents which overlap the range entirely. The
37 * outer loop is as follows:
38 *
39 * 1. For all reverse mappings overlapping the refcount extent,
40 * a. If a given rmap completely overlaps, mark it as seen.
41 * b. Otherwise, record the fragment (in agbno order) for later
42 * processing.
43 *
44 * Once we've seen all the rmaps, we know that for all blocks in the
45 * refcount record we want to find $refcount owners and we've already
46 * visited $seen extents that overlap all the blocks. Therefore, we
47 * need to find ($refcount - $seen) owners for every block in the
48 * extent; call that quantity $target_nr. Proceed as follows:
49 *
50 * 2. Pull the first $target_nr fragments from the list; all of them
51 * should start at or before the start of the extent.
52 * Call this subset of fragments the working set.
53 * 3. Until there are no more unprocessed fragments,
54 * a. Find the shortest fragments in the set and remove them.
55 * b. Note the block number of the end of these fragments.
56 * c. Pull the same number of fragments from the list. All of these
57 * fragments should start at the block number recorded in the
58 * previous step.
59 * d. Put those fragments in the set.
60 * 4. Check that there are $target_nr fragments remaining in the list,
61 * and that they all end at or beyond the end of the refcount extent.
62 *
63 * If the refcount is correct, all the check conditions in the algorithm
64 * should always hold true. If not, the refcount is incorrect.
65 */
66 struct xchk_refcnt_frag {
67 struct list_head list;
68 struct xfs_rmap_irec rm;
69 };
70
71 struct xchk_refcnt_check {
72 struct xfs_scrub *sc;
73 struct list_head fragments;
74
75 /* refcount extent we're examining */
76 xfs_agblock_t bno;
77 xfs_extlen_t len;
78 xfs_nlink_t refcount;
79
80 /* number of owners seen */
81 xfs_nlink_t seen;
82 };
83
84 /*
85 * Decide if the given rmap is large enough that we can redeem it
86 * towards refcount verification now, or if it's a fragment, in
87 * which case we'll hang onto it in the hopes that we'll later
88 * discover that we've collected exactly the correct number of
89 * fragments as the refcountbt says we should have.
90 */
91 STATIC int
xchk_refcountbt_rmap_check(struct xfs_btree_cur * cur,struct xfs_rmap_irec * rec,void * priv)92 xchk_refcountbt_rmap_check(
93 struct xfs_btree_cur *cur,
94 struct xfs_rmap_irec *rec,
95 void *priv)
96 {
97 struct xchk_refcnt_check *refchk = priv;
98 struct xchk_refcnt_frag *frag;
99 xfs_agblock_t rm_last;
100 xfs_agblock_t rc_last;
101 int error = 0;
102
103 if (xchk_should_terminate(refchk->sc, &error))
104 return error;
105
106 rm_last = rec->rm_startblock + rec->rm_blockcount - 1;
107 rc_last = refchk->bno + refchk->len - 1;
108
109 /* Confirm that a single-owner refc extent is a CoW stage. */
110 if (refchk->refcount == 1 && rec->rm_owner != XFS_RMAP_OWN_COW) {
111 xchk_btree_xref_set_corrupt(refchk->sc, cur, 0);
112 return 0;
113 }
114
115 if (rec->rm_startblock <= refchk->bno && rm_last >= rc_last) {
116 /*
117 * The rmap overlaps the refcount record, so we can confirm
118 * one refcount owner seen.
119 */
120 refchk->seen++;
121 } else {
122 /*
123 * This rmap covers only part of the refcount record, so
124 * save the fragment for later processing. If the rmapbt
125 * is healthy each rmap_irec we see will be in agbno order
126 * so we don't need insertion sort here.
127 */
128 frag = kmem_alloc(sizeof(struct xchk_refcnt_frag),
129 KM_MAYFAIL);
130 if (!frag)
131 return -ENOMEM;
132 memcpy(&frag->rm, rec, sizeof(frag->rm));
133 list_add_tail(&frag->list, &refchk->fragments);
134 }
135
136 return 0;
137 }
138
139 /*
140 * Given a bunch of rmap fragments, iterate through them, keeping
141 * a running tally of the refcount. If this ever deviates from
142 * what we expect (which is the refcountbt's refcount minus the
143 * number of extents that totally covered the refcountbt extent),
144 * we have a refcountbt error.
145 */
146 STATIC void
xchk_refcountbt_process_rmap_fragments(struct xchk_refcnt_check * refchk)147 xchk_refcountbt_process_rmap_fragments(
148 struct xchk_refcnt_check *refchk)
149 {
150 struct list_head worklist;
151 struct xchk_refcnt_frag *frag;
152 struct xchk_refcnt_frag *n;
153 xfs_agblock_t bno;
154 xfs_agblock_t rbno;
155 xfs_agblock_t next_rbno;
156 xfs_nlink_t nr;
157 xfs_nlink_t target_nr;
158
159 target_nr = refchk->refcount - refchk->seen;
160 if (target_nr == 0)
161 return;
162
163 /*
164 * There are (refchk->rc.rc_refcount - refchk->nr refcount)
165 * references we haven't found yet. Pull that many off the
166 * fragment list and figure out where the smallest rmap ends
167 * (and therefore the next rmap should start). All the rmaps
168 * we pull off should start at or before the beginning of the
169 * refcount record's range.
170 */
171 INIT_LIST_HEAD(&worklist);
172 rbno = NULLAGBLOCK;
173 nr = 1;
174
175 /* Make sure the fragments actually /are/ in agbno order. */
176 bno = 0;
177 list_for_each_entry(frag, &refchk->fragments, list) {
178 if (frag->rm.rm_startblock < bno)
179 goto done;
180 bno = frag->rm.rm_startblock;
181 }
182
183 /*
184 * Find all the rmaps that start at or before the refc extent,
185 * and put them on the worklist.
186 */
187 list_for_each_entry_safe(frag, n, &refchk->fragments, list) {
188 if (frag->rm.rm_startblock > refchk->bno)
189 goto done;
190 bno = frag->rm.rm_startblock + frag->rm.rm_blockcount;
191 if (bno < rbno)
192 rbno = bno;
193 list_move_tail(&frag->list, &worklist);
194 if (nr == target_nr)
195 break;
196 nr++;
197 }
198
199 /*
200 * We should have found exactly $target_nr rmap fragments starting
201 * at or before the refcount extent.
202 */
203 if (nr != target_nr)
204 goto done;
205
206 while (!list_empty(&refchk->fragments)) {
207 /* Discard any fragments ending at rbno from the worklist. */
208 nr = 0;
209 next_rbno = NULLAGBLOCK;
210 list_for_each_entry_safe(frag, n, &worklist, list) {
211 bno = frag->rm.rm_startblock + frag->rm.rm_blockcount;
212 if (bno != rbno) {
213 if (bno < next_rbno)
214 next_rbno = bno;
215 continue;
216 }
217 list_del(&frag->list);
218 kmem_free(frag);
219 nr++;
220 }
221
222 /* Try to add nr rmaps starting at rbno to the worklist. */
223 list_for_each_entry_safe(frag, n, &refchk->fragments, list) {
224 bno = frag->rm.rm_startblock + frag->rm.rm_blockcount;
225 if (frag->rm.rm_startblock != rbno)
226 goto done;
227 list_move_tail(&frag->list, &worklist);
228 if (next_rbno > bno)
229 next_rbno = bno;
230 nr--;
231 if (nr == 0)
232 break;
233 }
234
235 /*
236 * If we get here and nr > 0, this means that we added fewer
237 * items to the worklist than we discarded because the fragment
238 * list ran out of items. Therefore, we cannot maintain the
239 * required refcount. Something is wrong, so we're done.
240 */
241 if (nr)
242 goto done;
243
244 rbno = next_rbno;
245 }
246
247 /*
248 * Make sure the last extent we processed ends at or beyond
249 * the end of the refcount extent.
250 */
251 if (rbno < refchk->bno + refchk->len)
252 goto done;
253
254 /* Actually record us having seen the remaining refcount. */
255 refchk->seen = refchk->refcount;
256 done:
257 /* Delete fragments and work list. */
258 list_for_each_entry_safe(frag, n, &worklist, list) {
259 list_del(&frag->list);
260 kmem_free(frag);
261 }
262 list_for_each_entry_safe(frag, n, &refchk->fragments, list) {
263 list_del(&frag->list);
264 kmem_free(frag);
265 }
266 }
267
268 /* Use the rmap entries covering this extent to verify the refcount. */
269 STATIC void
xchk_refcountbt_xref_rmap(struct xfs_scrub * sc,xfs_agblock_t bno,xfs_extlen_t len,xfs_nlink_t refcount)270 xchk_refcountbt_xref_rmap(
271 struct xfs_scrub *sc,
272 xfs_agblock_t bno,
273 xfs_extlen_t len,
274 xfs_nlink_t refcount)
275 {
276 struct xchk_refcnt_check refchk = {
277 .sc = sc,
278 .bno = bno,
279 .len = len,
280 .refcount = refcount,
281 .seen = 0,
282 };
283 struct xfs_rmap_irec low;
284 struct xfs_rmap_irec high;
285 struct xchk_refcnt_frag *frag;
286 struct xchk_refcnt_frag *n;
287 int error;
288
289 if (!sc->sa.rmap_cur || xchk_skip_xref(sc->sm))
290 return;
291
292 /* Cross-reference with the rmapbt to confirm the refcount. */
293 memset(&low, 0, sizeof(low));
294 low.rm_startblock = bno;
295 memset(&high, 0xFF, sizeof(high));
296 high.rm_startblock = bno + len - 1;
297
298 INIT_LIST_HEAD(&refchk.fragments);
299 error = xfs_rmap_query_range(sc->sa.rmap_cur, &low, &high,
300 &xchk_refcountbt_rmap_check, &refchk);
301 if (!xchk_should_check_xref(sc, &error, &sc->sa.rmap_cur))
302 goto out_free;
303
304 xchk_refcountbt_process_rmap_fragments(&refchk);
305 if (refcount != refchk.seen)
306 xchk_btree_xref_set_corrupt(sc, sc->sa.rmap_cur, 0);
307
308 out_free:
309 list_for_each_entry_safe(frag, n, &refchk.fragments, list) {
310 list_del(&frag->list);
311 kmem_free(frag);
312 }
313 }
314
315 /* Cross-reference with the other btrees. */
316 STATIC void
xchk_refcountbt_xref(struct xfs_scrub * sc,xfs_agblock_t agbno,xfs_extlen_t len,xfs_nlink_t refcount)317 xchk_refcountbt_xref(
318 struct xfs_scrub *sc,
319 xfs_agblock_t agbno,
320 xfs_extlen_t len,
321 xfs_nlink_t refcount)
322 {
323 if (sc->sm->sm_flags & XFS_SCRUB_OFLAG_CORRUPT)
324 return;
325
326 xchk_xref_is_used_space(sc, agbno, len);
327 xchk_xref_is_not_inode_chunk(sc, agbno, len);
328 xchk_refcountbt_xref_rmap(sc, agbno, len, refcount);
329 }
330
331 /* Scrub a refcountbt record. */
332 STATIC int
xchk_refcountbt_rec(struct xchk_btree * bs,union xfs_btree_rec * rec)333 xchk_refcountbt_rec(
334 struct xchk_btree *bs,
335 union xfs_btree_rec *rec)
336 {
337 struct xfs_mount *mp = bs->cur->bc_mp;
338 xfs_agblock_t *cow_blocks = bs->private;
339 xfs_agnumber_t agno = bs->cur->bc_private.a.agno;
340 xfs_agblock_t bno;
341 xfs_extlen_t len;
342 xfs_nlink_t refcount;
343 bool has_cowflag;
344
345 bno = be32_to_cpu(rec->refc.rc_startblock);
346 len = be32_to_cpu(rec->refc.rc_blockcount);
347 refcount = be32_to_cpu(rec->refc.rc_refcount);
348
349 /* Only CoW records can have refcount == 1. */
350 has_cowflag = (bno & XFS_REFC_COW_START);
351 if ((refcount == 1 && !has_cowflag) || (refcount != 1 && has_cowflag))
352 xchk_btree_set_corrupt(bs->sc, bs->cur, 0);
353 if (has_cowflag)
354 (*cow_blocks) += len;
355
356 /* Check the extent. */
357 bno &= ~XFS_REFC_COW_START;
358 if (bno + len <= bno ||
359 !xfs_verify_agbno(mp, agno, bno) ||
360 !xfs_verify_agbno(mp, agno, bno + len - 1))
361 xchk_btree_set_corrupt(bs->sc, bs->cur, 0);
362
363 if (refcount == 0)
364 xchk_btree_set_corrupt(bs->sc, bs->cur, 0);
365
366 xchk_refcountbt_xref(bs->sc, bno, len, refcount);
367
368 return 0;
369 }
370
371 /* Make sure we have as many refc blocks as the rmap says. */
372 STATIC void
xchk_refcount_xref_rmap(struct xfs_scrub * sc,xfs_filblks_t cow_blocks)373 xchk_refcount_xref_rmap(
374 struct xfs_scrub *sc,
375 xfs_filblks_t cow_blocks)
376 {
377 xfs_extlen_t refcbt_blocks = 0;
378 xfs_filblks_t blocks;
379 int error;
380
381 if (!sc->sa.rmap_cur || xchk_skip_xref(sc->sm))
382 return;
383
384 /* Check that we saw as many refcbt blocks as the rmap knows about. */
385 error = xfs_btree_count_blocks(sc->sa.refc_cur, &refcbt_blocks);
386 if (!xchk_btree_process_error(sc, sc->sa.refc_cur, 0, &error))
387 return;
388 error = xchk_count_rmap_ownedby_ag(sc, sc->sa.rmap_cur,
389 &XFS_RMAP_OINFO_REFC, &blocks);
390 if (!xchk_should_check_xref(sc, &error, &sc->sa.rmap_cur))
391 return;
392 if (blocks != refcbt_blocks)
393 xchk_btree_xref_set_corrupt(sc, sc->sa.rmap_cur, 0);
394
395 /* Check that we saw as many cow blocks as the rmap knows about. */
396 error = xchk_count_rmap_ownedby_ag(sc, sc->sa.rmap_cur,
397 &XFS_RMAP_OINFO_COW, &blocks);
398 if (!xchk_should_check_xref(sc, &error, &sc->sa.rmap_cur))
399 return;
400 if (blocks != cow_blocks)
401 xchk_btree_xref_set_corrupt(sc, sc->sa.rmap_cur, 0);
402 }
403
404 /* Scrub the refcount btree for some AG. */
405 int
xchk_refcountbt(struct xfs_scrub * sc)406 xchk_refcountbt(
407 struct xfs_scrub *sc)
408 {
409 xfs_agblock_t cow_blocks = 0;
410 int error;
411
412 error = xchk_btree(sc, sc->sa.refc_cur, xchk_refcountbt_rec,
413 &XFS_RMAP_OINFO_REFC, &cow_blocks);
414 if (error)
415 return error;
416
417 xchk_refcount_xref_rmap(sc, cow_blocks);
418
419 return 0;
420 }
421
422 /* xref check that a cow staging extent is marked in the refcountbt. */
423 void
xchk_xref_is_cow_staging(struct xfs_scrub * sc,xfs_agblock_t agbno,xfs_extlen_t len)424 xchk_xref_is_cow_staging(
425 struct xfs_scrub *sc,
426 xfs_agblock_t agbno,
427 xfs_extlen_t len)
428 {
429 struct xfs_refcount_irec rc;
430 bool has_cowflag;
431 int has_refcount;
432 int error;
433
434 if (!sc->sa.refc_cur || xchk_skip_xref(sc->sm))
435 return;
436
437 /* Find the CoW staging extent. */
438 error = xfs_refcount_lookup_le(sc->sa.refc_cur,
439 agbno + XFS_REFC_COW_START, &has_refcount);
440 if (!xchk_should_check_xref(sc, &error, &sc->sa.refc_cur))
441 return;
442 if (!has_refcount) {
443 xchk_btree_xref_set_corrupt(sc, sc->sa.refc_cur, 0);
444 return;
445 }
446
447 error = xfs_refcount_get_rec(sc->sa.refc_cur, &rc, &has_refcount);
448 if (!xchk_should_check_xref(sc, &error, &sc->sa.refc_cur))
449 return;
450 if (!has_refcount) {
451 xchk_btree_xref_set_corrupt(sc, sc->sa.refc_cur, 0);
452 return;
453 }
454
455 /* CoW flag must be set, refcount must be 1. */
456 has_cowflag = (rc.rc_startblock & XFS_REFC_COW_START);
457 if (!has_cowflag || rc.rc_refcount != 1)
458 xchk_btree_xref_set_corrupt(sc, sc->sa.refc_cur, 0);
459
460 /* Must be at least as long as what was passed in */
461 if (rc.rc_blockcount < len)
462 xchk_btree_xref_set_corrupt(sc, sc->sa.refc_cur, 0);
463 }
464
465 /*
466 * xref check that the extent is not shared. Only file data blocks
467 * can have multiple owners.
468 */
469 void
xchk_xref_is_not_shared(struct xfs_scrub * sc,xfs_agblock_t agbno,xfs_extlen_t len)470 xchk_xref_is_not_shared(
471 struct xfs_scrub *sc,
472 xfs_agblock_t agbno,
473 xfs_extlen_t len)
474 {
475 bool shared;
476 int error;
477
478 if (!sc->sa.refc_cur || xchk_skip_xref(sc->sm))
479 return;
480
481 error = xfs_refcount_has_record(sc->sa.refc_cur, agbno, len, &shared);
482 if (!xchk_should_check_xref(sc, &error, &sc->sa.refc_cur))
483 return;
484 if (shared)
485 xchk_btree_xref_set_corrupt(sc, sc->sa.refc_cur, 0);
486 }
487