]> git.karo-electronics.de Git - karo-tx-linux.git/blob - fs/xfs/xfs_alloc_btree.c
Merge remote-tracking branch 'net-next/master'
[karo-tx-linux.git] / fs / xfs / xfs_alloc_btree.c
1 /*
2  * Copyright (c) 2000-2001,2005 Silicon Graphics, Inc.
3  * All Rights Reserved.
4  *
5  * This program is free software; you can redistribute it and/or
6  * modify it under the terms of the GNU General Public License as
7  * published by the Free Software Foundation.
8  *
9  * This program is distributed in the hope that it would be useful,
10  * but WITHOUT ANY WARRANTY; without even the implied warranty of
11  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
12  * GNU General Public License for more details.
13  *
14  * You should have received a copy of the GNU General Public License
15  * along with this program; if not, write the Free Software Foundation,
16  * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
17  */
18 #include "xfs.h"
19 #include "xfs_fs.h"
20 #include "xfs_format.h"
21 #include "xfs_log_format.h"
22 #include "xfs_trans_resv.h"
23 #include "xfs_sb.h"
24 #include "xfs_ag.h"
25 #include "xfs_mount.h"
26 #include "xfs_btree.h"
27 #include "xfs_alloc_btree.h"
28 #include "xfs_alloc.h"
29 #include "xfs_extent_busy.h"
30 #include "xfs_error.h"
31 #include "xfs_trace.h"
32 #include "xfs_cksum.h"
33 #include "xfs_trans.h"
34
35
36 STATIC struct xfs_btree_cur *
37 xfs_allocbt_dup_cursor(
38         struct xfs_btree_cur    *cur)
39 {
40         return xfs_allocbt_init_cursor(cur->bc_mp, cur->bc_tp,
41                         cur->bc_private.a.agbp, cur->bc_private.a.agno,
42                         cur->bc_btnum);
43 }
44
45 STATIC void
46 xfs_allocbt_set_root(
47         struct xfs_btree_cur    *cur,
48         union xfs_btree_ptr     *ptr,
49         int                     inc)
50 {
51         struct xfs_buf          *agbp = cur->bc_private.a.agbp;
52         struct xfs_agf          *agf = XFS_BUF_TO_AGF(agbp);
53         xfs_agnumber_t          seqno = be32_to_cpu(agf->agf_seqno);
54         int                     btnum = cur->bc_btnum;
55         struct xfs_perag        *pag = xfs_perag_get(cur->bc_mp, seqno);
56
57         ASSERT(ptr->s != 0);
58
59         agf->agf_roots[btnum] = ptr->s;
60         be32_add_cpu(&agf->agf_levels[btnum], inc);
61         pag->pagf_levels[btnum] += inc;
62         xfs_perag_put(pag);
63
64         xfs_alloc_log_agf(cur->bc_tp, agbp, XFS_AGF_ROOTS | XFS_AGF_LEVELS);
65 }
66
67 STATIC int
68 xfs_allocbt_alloc_block(
69         struct xfs_btree_cur    *cur,
70         union xfs_btree_ptr     *start,
71         union xfs_btree_ptr     *new,
72         int                     length,
73         int                     *stat)
74 {
75         int                     error;
76         xfs_agblock_t           bno;
77
78         XFS_BTREE_TRACE_CURSOR(cur, XBT_ENTRY);
79
80         /* Allocate the new block from the freelist. If we can't, give up.  */
81         error = xfs_alloc_get_freelist(cur->bc_tp, cur->bc_private.a.agbp,
82                                        &bno, 1);
83         if (error) {
84                 XFS_BTREE_TRACE_CURSOR(cur, XBT_ERROR);
85                 return error;
86         }
87
88         if (bno == NULLAGBLOCK) {
89                 XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
90                 *stat = 0;
91                 return 0;
92         }
93
94         xfs_extent_busy_reuse(cur->bc_mp, cur->bc_private.a.agno, bno, 1, false);
95
96         xfs_trans_agbtree_delta(cur->bc_tp, 1);
97         new->s = cpu_to_be32(bno);
98
99         XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
100         *stat = 1;
101         return 0;
102 }
103
104 STATIC int
105 xfs_allocbt_free_block(
106         struct xfs_btree_cur    *cur,
107         struct xfs_buf          *bp)
108 {
109         struct xfs_buf          *agbp = cur->bc_private.a.agbp;
110         struct xfs_agf          *agf = XFS_BUF_TO_AGF(agbp);
111         xfs_agblock_t           bno;
112         int                     error;
113
114         bno = xfs_daddr_to_agbno(cur->bc_mp, XFS_BUF_ADDR(bp));
115         error = xfs_alloc_put_freelist(cur->bc_tp, agbp, NULL, bno, 1);
116         if (error)
117                 return error;
118
119         xfs_extent_busy_insert(cur->bc_tp, be32_to_cpu(agf->agf_seqno), bno, 1,
120                               XFS_EXTENT_BUSY_SKIP_DISCARD);
121         xfs_trans_agbtree_delta(cur->bc_tp, -1);
122
123         xfs_trans_binval(cur->bc_tp, bp);
124         return 0;
125 }
126
127 /*
128  * Update the longest extent in the AGF
129  */
130 STATIC void
131 xfs_allocbt_update_lastrec(
132         struct xfs_btree_cur    *cur,
133         struct xfs_btree_block  *block,
134         union xfs_btree_rec     *rec,
135         int                     ptr,
136         int                     reason)
137 {
138         struct xfs_agf          *agf = XFS_BUF_TO_AGF(cur->bc_private.a.agbp);
139         xfs_agnumber_t          seqno = be32_to_cpu(agf->agf_seqno);
140         struct xfs_perag        *pag;
141         __be32                  len;
142         int                     numrecs;
143
144         ASSERT(cur->bc_btnum == XFS_BTNUM_CNT);
145
146         switch (reason) {
147         case LASTREC_UPDATE:
148                 /*
149                  * If this is the last leaf block and it's the last record,
150                  * then update the size of the longest extent in the AG.
151                  */
152                 if (ptr != xfs_btree_get_numrecs(block))
153                         return;
154                 len = rec->alloc.ar_blockcount;
155                 break;
156         case LASTREC_INSREC:
157                 if (be32_to_cpu(rec->alloc.ar_blockcount) <=
158                     be32_to_cpu(agf->agf_longest))
159                         return;
160                 len = rec->alloc.ar_blockcount;
161                 break;
162         case LASTREC_DELREC:
163                 numrecs = xfs_btree_get_numrecs(block);
164                 if (ptr <= numrecs)
165                         return;
166                 ASSERT(ptr == numrecs + 1);
167
168                 if (numrecs) {
169                         xfs_alloc_rec_t *rrp;
170
171                         rrp = XFS_ALLOC_REC_ADDR(cur->bc_mp, block, numrecs);
172                         len = rrp->ar_blockcount;
173                 } else {
174                         len = 0;
175                 }
176
177                 break;
178         default:
179                 ASSERT(0);
180                 return;
181         }
182
183         agf->agf_longest = len;
184         pag = xfs_perag_get(cur->bc_mp, seqno);
185         pag->pagf_longest = be32_to_cpu(len);
186         xfs_perag_put(pag);
187         xfs_alloc_log_agf(cur->bc_tp, cur->bc_private.a.agbp, XFS_AGF_LONGEST);
188 }
189
190 STATIC int
191 xfs_allocbt_get_minrecs(
192         struct xfs_btree_cur    *cur,
193         int                     level)
194 {
195         return cur->bc_mp->m_alloc_mnr[level != 0];
196 }
197
198 STATIC int
199 xfs_allocbt_get_maxrecs(
200         struct xfs_btree_cur    *cur,
201         int                     level)
202 {
203         return cur->bc_mp->m_alloc_mxr[level != 0];
204 }
205
206 STATIC void
207 xfs_allocbt_init_key_from_rec(
208         union xfs_btree_key     *key,
209         union xfs_btree_rec     *rec)
210 {
211         ASSERT(rec->alloc.ar_startblock != 0);
212
213         key->alloc.ar_startblock = rec->alloc.ar_startblock;
214         key->alloc.ar_blockcount = rec->alloc.ar_blockcount;
215 }
216
217 STATIC void
218 xfs_allocbt_init_rec_from_key(
219         union xfs_btree_key     *key,
220         union xfs_btree_rec     *rec)
221 {
222         ASSERT(key->alloc.ar_startblock != 0);
223
224         rec->alloc.ar_startblock = key->alloc.ar_startblock;
225         rec->alloc.ar_blockcount = key->alloc.ar_blockcount;
226 }
227
228 STATIC void
229 xfs_allocbt_init_rec_from_cur(
230         struct xfs_btree_cur    *cur,
231         union xfs_btree_rec     *rec)
232 {
233         ASSERT(cur->bc_rec.a.ar_startblock != 0);
234
235         rec->alloc.ar_startblock = cpu_to_be32(cur->bc_rec.a.ar_startblock);
236         rec->alloc.ar_blockcount = cpu_to_be32(cur->bc_rec.a.ar_blockcount);
237 }
238
239 STATIC void
240 xfs_allocbt_init_ptr_from_cur(
241         struct xfs_btree_cur    *cur,
242         union xfs_btree_ptr     *ptr)
243 {
244         struct xfs_agf          *agf = XFS_BUF_TO_AGF(cur->bc_private.a.agbp);
245
246         ASSERT(cur->bc_private.a.agno == be32_to_cpu(agf->agf_seqno));
247         ASSERT(agf->agf_roots[cur->bc_btnum] != 0);
248
249         ptr->s = agf->agf_roots[cur->bc_btnum];
250 }
251
252 STATIC __int64_t
253 xfs_allocbt_key_diff(
254         struct xfs_btree_cur    *cur,
255         union xfs_btree_key     *key)
256 {
257         xfs_alloc_rec_incore_t  *rec = &cur->bc_rec.a;
258         xfs_alloc_key_t         *kp = &key->alloc;
259         __int64_t               diff;
260
261         if (cur->bc_btnum == XFS_BTNUM_BNO) {
262                 return (__int64_t)be32_to_cpu(kp->ar_startblock) -
263                                 rec->ar_startblock;
264         }
265
266         diff = (__int64_t)be32_to_cpu(kp->ar_blockcount) - rec->ar_blockcount;
267         if (diff)
268                 return diff;
269
270         return (__int64_t)be32_to_cpu(kp->ar_startblock) - rec->ar_startblock;
271 }
272
273 static bool
274 xfs_allocbt_verify(
275         struct xfs_buf          *bp)
276 {
277         struct xfs_mount        *mp = bp->b_target->bt_mount;
278         struct xfs_btree_block  *block = XFS_BUF_TO_BLOCK(bp);
279         struct xfs_perag        *pag = bp->b_pag;
280         unsigned int            level;
281
282         /*
283          * magic number and level verification
284          *
285          * During growfs operations, we can't verify the exact level or owner as
286          * the perag is not fully initialised and hence not attached to the
287          * buffer.  In this case, check against the maximum tree depth.
288          *
289          * Similarly, during log recovery we will have a perag structure
290          * attached, but the agf information will not yet have been initialised
291          * from the on disk AGF. Again, we can only check against maximum limits
292          * in this case.
293          */
294         level = be16_to_cpu(block->bb_level);
295         switch (block->bb_magic) {
296         case cpu_to_be32(XFS_ABTB_CRC_MAGIC):
297                 if (!xfs_sb_version_hascrc(&mp->m_sb))
298                         return false;
299                 if (!uuid_equal(&block->bb_u.s.bb_uuid, &mp->m_sb.sb_uuid))
300                         return false;
301                 if (block->bb_u.s.bb_blkno != cpu_to_be64(bp->b_bn))
302                         return false;
303                 if (pag &&
304                     be32_to_cpu(block->bb_u.s.bb_owner) != pag->pag_agno)
305                         return false;
306                 /* fall through */
307         case cpu_to_be32(XFS_ABTB_MAGIC):
308                 if (pag && pag->pagf_init) {
309                         if (level >= pag->pagf_levels[XFS_BTNUM_BNOi])
310                                 return false;
311                 } else if (level >= mp->m_ag_maxlevels)
312                         return false;
313                 break;
314         case cpu_to_be32(XFS_ABTC_CRC_MAGIC):
315                 if (!xfs_sb_version_hascrc(&mp->m_sb))
316                         return false;
317                 if (!uuid_equal(&block->bb_u.s.bb_uuid, &mp->m_sb.sb_uuid))
318                         return false;
319                 if (block->bb_u.s.bb_blkno != cpu_to_be64(bp->b_bn))
320                         return false;
321                 if (pag &&
322                     be32_to_cpu(block->bb_u.s.bb_owner) != pag->pag_agno)
323                         return false;
324                 /* fall through */
325         case cpu_to_be32(XFS_ABTC_MAGIC):
326                 if (pag && pag->pagf_init) {
327                         if (level >= pag->pagf_levels[XFS_BTNUM_CNTi])
328                                 return false;
329                 } else if (level >= mp->m_ag_maxlevels)
330                         return false;
331                 break;
332         default:
333                 return false;
334         }
335
336         /* numrecs verification */
337         if (be16_to_cpu(block->bb_numrecs) > mp->m_alloc_mxr[level != 0])
338                 return false;
339
340         /* sibling pointer verification */
341         if (!block->bb_u.s.bb_leftsib ||
342             (be32_to_cpu(block->bb_u.s.bb_leftsib) >= mp->m_sb.sb_agblocks &&
343              block->bb_u.s.bb_leftsib != cpu_to_be32(NULLAGBLOCK)))
344                 return false;
345         if (!block->bb_u.s.bb_rightsib ||
346             (be32_to_cpu(block->bb_u.s.bb_rightsib) >= mp->m_sb.sb_agblocks &&
347              block->bb_u.s.bb_rightsib != cpu_to_be32(NULLAGBLOCK)))
348                 return false;
349
350         return true;
351 }
352
353 static void
354 xfs_allocbt_read_verify(
355         struct xfs_buf  *bp)
356 {
357         if (!(xfs_btree_sblock_verify_crc(bp) &&
358               xfs_allocbt_verify(bp))) {
359                 trace_xfs_btree_corrupt(bp, _RET_IP_);
360                 XFS_CORRUPTION_ERROR(__func__, XFS_ERRLEVEL_LOW,
361                                      bp->b_target->bt_mount, bp->b_addr);
362                 xfs_buf_ioerror(bp, EFSCORRUPTED);
363         }
364 }
365
366 static void
367 xfs_allocbt_write_verify(
368         struct xfs_buf  *bp)
369 {
370         if (!xfs_allocbt_verify(bp)) {
371                 trace_xfs_btree_corrupt(bp, _RET_IP_);
372                 XFS_CORRUPTION_ERROR(__func__, XFS_ERRLEVEL_LOW,
373                                      bp->b_target->bt_mount, bp->b_addr);
374                 xfs_buf_ioerror(bp, EFSCORRUPTED);
375         }
376         xfs_btree_sblock_calc_crc(bp);
377
378 }
379
380 const struct xfs_buf_ops xfs_allocbt_buf_ops = {
381         .verify_read = xfs_allocbt_read_verify,
382         .verify_write = xfs_allocbt_write_verify,
383 };
384
385
386 #if defined(DEBUG) || defined(XFS_WARN)
387 STATIC int
388 xfs_allocbt_keys_inorder(
389         struct xfs_btree_cur    *cur,
390         union xfs_btree_key     *k1,
391         union xfs_btree_key     *k2)
392 {
393         if (cur->bc_btnum == XFS_BTNUM_BNO) {
394                 return be32_to_cpu(k1->alloc.ar_startblock) <
395                        be32_to_cpu(k2->alloc.ar_startblock);
396         } else {
397                 return be32_to_cpu(k1->alloc.ar_blockcount) <
398                         be32_to_cpu(k2->alloc.ar_blockcount) ||
399                         (k1->alloc.ar_blockcount == k2->alloc.ar_blockcount &&
400                          be32_to_cpu(k1->alloc.ar_startblock) <
401                          be32_to_cpu(k2->alloc.ar_startblock));
402         }
403 }
404
405 STATIC int
406 xfs_allocbt_recs_inorder(
407         struct xfs_btree_cur    *cur,
408         union xfs_btree_rec     *r1,
409         union xfs_btree_rec     *r2)
410 {
411         if (cur->bc_btnum == XFS_BTNUM_BNO) {
412                 return be32_to_cpu(r1->alloc.ar_startblock) +
413                         be32_to_cpu(r1->alloc.ar_blockcount) <=
414                         be32_to_cpu(r2->alloc.ar_startblock);
415         } else {
416                 return be32_to_cpu(r1->alloc.ar_blockcount) <
417                         be32_to_cpu(r2->alloc.ar_blockcount) ||
418                         (r1->alloc.ar_blockcount == r2->alloc.ar_blockcount &&
419                          be32_to_cpu(r1->alloc.ar_startblock) <
420                          be32_to_cpu(r2->alloc.ar_startblock));
421         }
422 }
423 #endif  /* DEBUG */
424
425 static const struct xfs_btree_ops xfs_allocbt_ops = {
426         .rec_len                = sizeof(xfs_alloc_rec_t),
427         .key_len                = sizeof(xfs_alloc_key_t),
428
429         .dup_cursor             = xfs_allocbt_dup_cursor,
430         .set_root               = xfs_allocbt_set_root,
431         .alloc_block            = xfs_allocbt_alloc_block,
432         .free_block             = xfs_allocbt_free_block,
433         .update_lastrec         = xfs_allocbt_update_lastrec,
434         .get_minrecs            = xfs_allocbt_get_minrecs,
435         .get_maxrecs            = xfs_allocbt_get_maxrecs,
436         .init_key_from_rec      = xfs_allocbt_init_key_from_rec,
437         .init_rec_from_key      = xfs_allocbt_init_rec_from_key,
438         .init_rec_from_cur      = xfs_allocbt_init_rec_from_cur,
439         .init_ptr_from_cur      = xfs_allocbt_init_ptr_from_cur,
440         .key_diff               = xfs_allocbt_key_diff,
441         .buf_ops                = &xfs_allocbt_buf_ops,
442 #if defined(DEBUG) || defined(XFS_WARN)
443         .keys_inorder           = xfs_allocbt_keys_inorder,
444         .recs_inorder           = xfs_allocbt_recs_inorder,
445 #endif
446 };
447
448 /*
449  * Allocate a new allocation btree cursor.
450  */
451 struct xfs_btree_cur *                  /* new alloc btree cursor */
452 xfs_allocbt_init_cursor(
453         struct xfs_mount        *mp,            /* file system mount point */
454         struct xfs_trans        *tp,            /* transaction pointer */
455         struct xfs_buf          *agbp,          /* buffer for agf structure */
456         xfs_agnumber_t          agno,           /* allocation group number */
457         xfs_btnum_t             btnum)          /* btree identifier */
458 {
459         struct xfs_agf          *agf = XFS_BUF_TO_AGF(agbp);
460         struct xfs_btree_cur    *cur;
461
462         ASSERT(btnum == XFS_BTNUM_BNO || btnum == XFS_BTNUM_CNT);
463
464         cur = kmem_zone_zalloc(xfs_btree_cur_zone, KM_SLEEP);
465
466         cur->bc_tp = tp;
467         cur->bc_mp = mp;
468         cur->bc_btnum = btnum;
469         cur->bc_blocklog = mp->m_sb.sb_blocklog;
470         cur->bc_ops = &xfs_allocbt_ops;
471
472         if (btnum == XFS_BTNUM_CNT) {
473                 cur->bc_nlevels = be32_to_cpu(agf->agf_levels[XFS_BTNUM_CNT]);
474                 cur->bc_flags = XFS_BTREE_LASTREC_UPDATE;
475         } else {
476                 cur->bc_nlevels = be32_to_cpu(agf->agf_levels[XFS_BTNUM_BNO]);
477         }
478
479         cur->bc_private.a.agbp = agbp;
480         cur->bc_private.a.agno = agno;
481
482         if (xfs_sb_version_hascrc(&mp->m_sb))
483                 cur->bc_flags |= XFS_BTREE_CRC_BLOCKS;
484
485         return cur;
486 }
487
488 /*
489  * Calculate number of records in an alloc btree block.
490  */
491 int
492 xfs_allocbt_maxrecs(
493         struct xfs_mount        *mp,
494         int                     blocklen,
495         int                     leaf)
496 {
497         blocklen -= XFS_ALLOC_BLOCK_LEN(mp);
498
499         if (leaf)
500                 return blocklen / sizeof(xfs_alloc_rec_t);
501         return blocklen / (sizeof(xfs_alloc_key_t) + sizeof(xfs_alloc_ptr_t));
502 }