2 * Copyright (c) 2000-2001,2005 Silicon Graphics, Inc.
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.
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.
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
18 #include "libxfs_priv.h"
20 #include "xfs_shared.h"
21 #include "xfs_format.h"
22 #include "xfs_log_format.h"
23 #include "xfs_trans_resv.h"
25 #include "xfs_mount.h"
26 #include "xfs_inode.h"
27 #include "xfs_btree.h"
28 #include "xfs_ialloc.h"
29 #include "xfs_ialloc_btree.h"
30 #include "xfs_alloc.h"
31 #include "xfs_trace.h"
32 #include "xfs_cksum.h"
33 #include "xfs_trans.h"
37 xfs_inobt_get_minrecs(
38 struct xfs_btree_cur
*cur
,
41 return cur
->bc_mp
->m_inobt_mnr
[level
!= 0];
44 STATIC
struct xfs_btree_cur
*
46 struct xfs_btree_cur
*cur
)
48 return xfs_inobt_init_cursor(cur
->bc_mp
, cur
->bc_tp
,
49 cur
->bc_private
.a
.agbp
, cur
->bc_private
.a
.agno
,
55 struct xfs_btree_cur
*cur
,
56 union xfs_btree_ptr
*nptr
,
57 int inc
) /* level change */
59 struct xfs_buf
*agbp
= cur
->bc_private
.a
.agbp
;
60 struct xfs_agi
*agi
= XFS_BUF_TO_AGI(agbp
);
62 agi
->agi_root
= nptr
->s
;
63 be32_add_cpu(&agi
->agi_level
, inc
);
64 xfs_ialloc_log_agi(cur
->bc_tp
, agbp
, XFS_AGI_ROOT
| XFS_AGI_LEVEL
);
69 struct xfs_btree_cur
*cur
,
70 union xfs_btree_ptr
*nptr
,
71 int inc
) /* level change */
73 struct xfs_buf
*agbp
= cur
->bc_private
.a
.agbp
;
74 struct xfs_agi
*agi
= XFS_BUF_TO_AGI(agbp
);
76 agi
->agi_free_root
= nptr
->s
;
77 be32_add_cpu(&agi
->agi_free_level
, inc
);
78 xfs_ialloc_log_agi(cur
->bc_tp
, agbp
,
79 XFS_AGI_FREE_ROOT
| XFS_AGI_FREE_LEVEL
);
83 xfs_inobt_alloc_block(
84 struct xfs_btree_cur
*cur
,
85 union xfs_btree_ptr
*start
,
86 union xfs_btree_ptr
*new,
89 xfs_alloc_arg_t args
; /* block allocation args */
90 int error
; /* error return value */
91 xfs_agblock_t sbno
= be32_to_cpu(start
->s
);
93 XFS_BTREE_TRACE_CURSOR(cur
, XBT_ENTRY
);
95 memset(&args
, 0, sizeof(args
));
98 args
.fsbno
= XFS_AGB_TO_FSB(args
.mp
, cur
->bc_private
.a
.agno
, sbno
);
102 args
.type
= XFS_ALLOCTYPE_NEAR_BNO
;
104 error
= xfs_alloc_vextent(&args
);
106 XFS_BTREE_TRACE_CURSOR(cur
, XBT_ERROR
);
109 if (args
.fsbno
== NULLFSBLOCK
) {
110 XFS_BTREE_TRACE_CURSOR(cur
, XBT_EXIT
);
114 ASSERT(args
.len
== 1);
115 XFS_BTREE_TRACE_CURSOR(cur
, XBT_EXIT
);
117 new->s
= cpu_to_be32(XFS_FSB_TO_AGBNO(args
.mp
, args
.fsbno
));
123 xfs_inobt_free_block(
124 struct xfs_btree_cur
*cur
,
130 fsbno
= XFS_DADDR_TO_FSB(cur
->bc_mp
, XFS_BUF_ADDR(bp
));
131 error
= xfs_free_extent(cur
->bc_tp
, fsbno
, 1);
135 xfs_trans_binval(cur
->bc_tp
, bp
);
140 xfs_inobt_get_maxrecs(
141 struct xfs_btree_cur
*cur
,
144 return cur
->bc_mp
->m_inobt_mxr
[level
!= 0];
148 xfs_inobt_init_key_from_rec(
149 union xfs_btree_key
*key
,
150 union xfs_btree_rec
*rec
)
152 key
->inobt
.ir_startino
= rec
->inobt
.ir_startino
;
156 xfs_inobt_init_rec_from_key(
157 union xfs_btree_key
*key
,
158 union xfs_btree_rec
*rec
)
160 rec
->inobt
.ir_startino
= key
->inobt
.ir_startino
;
164 xfs_inobt_init_rec_from_cur(
165 struct xfs_btree_cur
*cur
,
166 union xfs_btree_rec
*rec
)
168 rec
->inobt
.ir_startino
= cpu_to_be32(cur
->bc_rec
.i
.ir_startino
);
169 if (xfs_sb_version_hassparseinodes(&cur
->bc_mp
->m_sb
)) {
170 rec
->inobt
.ir_u
.sp
.ir_holemask
=
171 cpu_to_be16(cur
->bc_rec
.i
.ir_holemask
);
172 rec
->inobt
.ir_u
.sp
.ir_count
= cur
->bc_rec
.i
.ir_count
;
173 rec
->inobt
.ir_u
.sp
.ir_freecount
= cur
->bc_rec
.i
.ir_freecount
;
175 /* ir_holemask/ir_count not supported on-disk */
176 rec
->inobt
.ir_u
.f
.ir_freecount
=
177 cpu_to_be32(cur
->bc_rec
.i
.ir_freecount
);
179 rec
->inobt
.ir_free
= cpu_to_be64(cur
->bc_rec
.i
.ir_free
);
183 * initial value of ptr for lookup
186 xfs_inobt_init_ptr_from_cur(
187 struct xfs_btree_cur
*cur
,
188 union xfs_btree_ptr
*ptr
)
190 struct xfs_agi
*agi
= XFS_BUF_TO_AGI(cur
->bc_private
.a
.agbp
);
192 ASSERT(cur
->bc_private
.a
.agno
== be32_to_cpu(agi
->agi_seqno
));
194 ptr
->s
= agi
->agi_root
;
198 xfs_finobt_init_ptr_from_cur(
199 struct xfs_btree_cur
*cur
,
200 union xfs_btree_ptr
*ptr
)
202 struct xfs_agi
*agi
= XFS_BUF_TO_AGI(cur
->bc_private
.a
.agbp
);
204 ASSERT(cur
->bc_private
.a
.agno
== be32_to_cpu(agi
->agi_seqno
));
205 ptr
->s
= agi
->agi_free_root
;
210 struct xfs_btree_cur
*cur
,
211 union xfs_btree_key
*key
)
213 return (__int64_t
)be32_to_cpu(key
->inobt
.ir_startino
) -
214 cur
->bc_rec
.i
.ir_startino
;
221 struct xfs_mount
*mp
= bp
->b_target
->bt_mount
;
222 struct xfs_btree_block
*block
= XFS_BUF_TO_BLOCK(bp
);
226 * During growfs operations, we can't verify the exact owner as the
227 * perag is not fully initialised and hence not attached to the buffer.
229 * Similarly, during log recovery we will have a perag structure
230 * attached, but the agi information will not yet have been initialised
231 * from the on disk AGI. We don't currently use any of this information,
232 * but beware of the landmine (i.e. need to check pag->pagi_init) if we
235 switch (block
->bb_magic
) {
236 case cpu_to_be32(XFS_IBT_CRC_MAGIC
):
237 case cpu_to_be32(XFS_FIBT_CRC_MAGIC
):
238 if (!xfs_sb_version_hascrc(&mp
->m_sb
))
240 if (!xfs_btree_sblock_v5hdr_verify(bp
))
243 case cpu_to_be32(XFS_IBT_MAGIC
):
244 case cpu_to_be32(XFS_FIBT_MAGIC
):
250 /* level verification */
251 level
= be16_to_cpu(block
->bb_level
);
252 if (level
>= mp
->m_in_maxlevels
)
255 return xfs_btree_sblock_verify(bp
, mp
->m_inobt_mxr
[level
!= 0]);
259 xfs_inobt_read_verify(
262 if (!xfs_btree_sblock_verify_crc(bp
))
263 xfs_buf_ioerror(bp
, -EFSBADCRC
);
264 else if (!xfs_inobt_verify(bp
))
265 xfs_buf_ioerror(bp
, -EFSCORRUPTED
);
268 trace_xfs_btree_corrupt(bp
, _RET_IP_
);
269 xfs_verifier_error(bp
);
274 xfs_inobt_write_verify(
277 if (!xfs_inobt_verify(bp
)) {
278 trace_xfs_btree_corrupt(bp
, _RET_IP_
);
279 xfs_buf_ioerror(bp
, -EFSCORRUPTED
);
280 xfs_verifier_error(bp
);
283 xfs_btree_sblock_calc_crc(bp
);
287 const struct xfs_buf_ops xfs_inobt_buf_ops
= {
289 .verify_read
= xfs_inobt_read_verify
,
290 .verify_write
= xfs_inobt_write_verify
,
293 #if defined(DEBUG) || defined(XFS_WARN)
295 xfs_inobt_keys_inorder(
296 struct xfs_btree_cur
*cur
,
297 union xfs_btree_key
*k1
,
298 union xfs_btree_key
*k2
)
300 return be32_to_cpu(k1
->inobt
.ir_startino
) <
301 be32_to_cpu(k2
->inobt
.ir_startino
);
305 xfs_inobt_recs_inorder(
306 struct xfs_btree_cur
*cur
,
307 union xfs_btree_rec
*r1
,
308 union xfs_btree_rec
*r2
)
310 return be32_to_cpu(r1
->inobt
.ir_startino
) + XFS_INODES_PER_CHUNK
<=
311 be32_to_cpu(r2
->inobt
.ir_startino
);
315 static const struct xfs_btree_ops xfs_inobt_ops
= {
316 .rec_len
= sizeof(xfs_inobt_rec_t
),
317 .key_len
= sizeof(xfs_inobt_key_t
),
319 .dup_cursor
= xfs_inobt_dup_cursor
,
320 .set_root
= xfs_inobt_set_root
,
321 .alloc_block
= xfs_inobt_alloc_block
,
322 .free_block
= xfs_inobt_free_block
,
323 .get_minrecs
= xfs_inobt_get_minrecs
,
324 .get_maxrecs
= xfs_inobt_get_maxrecs
,
325 .init_key_from_rec
= xfs_inobt_init_key_from_rec
,
326 .init_rec_from_key
= xfs_inobt_init_rec_from_key
,
327 .init_rec_from_cur
= xfs_inobt_init_rec_from_cur
,
328 .init_ptr_from_cur
= xfs_inobt_init_ptr_from_cur
,
329 .key_diff
= xfs_inobt_key_diff
,
330 .buf_ops
= &xfs_inobt_buf_ops
,
331 #if defined(DEBUG) || defined(XFS_WARN)
332 .keys_inorder
= xfs_inobt_keys_inorder
,
333 .recs_inorder
= xfs_inobt_recs_inorder
,
337 static const struct xfs_btree_ops xfs_finobt_ops
= {
338 .rec_len
= sizeof(xfs_inobt_rec_t
),
339 .key_len
= sizeof(xfs_inobt_key_t
),
341 .dup_cursor
= xfs_inobt_dup_cursor
,
342 .set_root
= xfs_finobt_set_root
,
343 .alloc_block
= xfs_inobt_alloc_block
,
344 .free_block
= xfs_inobt_free_block
,
345 .get_minrecs
= xfs_inobt_get_minrecs
,
346 .get_maxrecs
= xfs_inobt_get_maxrecs
,
347 .init_key_from_rec
= xfs_inobt_init_key_from_rec
,
348 .init_rec_from_key
= xfs_inobt_init_rec_from_key
,
349 .init_rec_from_cur
= xfs_inobt_init_rec_from_cur
,
350 .init_ptr_from_cur
= xfs_finobt_init_ptr_from_cur
,
351 .key_diff
= xfs_inobt_key_diff
,
352 .buf_ops
= &xfs_inobt_buf_ops
,
353 #if defined(DEBUG) || defined(XFS_WARN)
354 .keys_inorder
= xfs_inobt_keys_inorder
,
355 .recs_inorder
= xfs_inobt_recs_inorder
,
360 * Allocate a new inode btree cursor.
362 struct xfs_btree_cur
* /* new inode btree cursor */
363 xfs_inobt_init_cursor(
364 struct xfs_mount
*mp
, /* file system mount point */
365 struct xfs_trans
*tp
, /* transaction pointer */
366 struct xfs_buf
*agbp
, /* buffer for agi structure */
367 xfs_agnumber_t agno
, /* allocation group number */
368 xfs_btnum_t btnum
) /* ialloc or free ino btree */
370 struct xfs_agi
*agi
= XFS_BUF_TO_AGI(agbp
);
371 struct xfs_btree_cur
*cur
;
373 cur
= kmem_zone_zalloc(xfs_btree_cur_zone
, KM_SLEEP
);
377 cur
->bc_btnum
= btnum
;
378 if (btnum
== XFS_BTNUM_INO
) {
379 cur
->bc_nlevels
= be32_to_cpu(agi
->agi_level
);
380 cur
->bc_ops
= &xfs_inobt_ops
;
382 cur
->bc_nlevels
= be32_to_cpu(agi
->agi_free_level
);
383 cur
->bc_ops
= &xfs_finobt_ops
;
386 cur
->bc_blocklog
= mp
->m_sb
.sb_blocklog
;
388 if (xfs_sb_version_hascrc(&mp
->m_sb
))
389 cur
->bc_flags
|= XFS_BTREE_CRC_BLOCKS
;
391 cur
->bc_private
.a
.agbp
= agbp
;
392 cur
->bc_private
.a
.agno
= agno
;
398 * Calculate number of records in an inobt btree block.
402 struct xfs_mount
*mp
,
406 blocklen
-= XFS_INOBT_BLOCK_LEN(mp
);
409 return blocklen
/ sizeof(xfs_inobt_rec_t
);
410 return blocklen
/ (sizeof(xfs_inobt_key_t
) + sizeof(xfs_inobt_ptr_t
));
414 * Convert the inode record holemask to an inode allocation bitmap. The inode
415 * allocation bitmap is inode granularity and specifies whether an inode is
416 * physically allocated on disk (not whether the inode is considered allocated
417 * or free by the fs).
419 * A bit value of 1 means the inode is allocated, a value of 0 means it is free.
422 xfs_inobt_irec_to_allocmask(
423 struct xfs_inobt_rec_incore
*rec
)
431 * The holemask has 16-bits for a 64 inode record. Therefore each
432 * holemask bit represents multiple inodes. Create a mask of bits to set
433 * in the allocmask for each holemask bit.
435 inodespbit
= (1 << XFS_INODES_PER_HOLEMASK_BIT
) - 1;
438 * Allocated inodes are represented by 0 bits in holemask. Invert the 0
439 * bits to 1 and convert to a uint so we can use xfs_next_bit(). Mask
440 * anything beyond the 16 holemask bits since this casts to a larger
443 allocbitmap
= ~rec
->ir_holemask
& ((1 << XFS_INOBT_HOLEMASK_BITS
) - 1);
446 * allocbitmap is the inverted holemask so every set bit represents
447 * allocated inodes. To expand from 16-bit holemask granularity to
448 * 64-bit (e.g., bit-per-inode), set inodespbit bits in the target
449 * bitmap for every holemask bit.
451 nextbit
= xfs_next_bit(&allocbitmap
, 1, 0);
452 while (nextbit
!= -1) {
453 ASSERT(nextbit
< (sizeof(rec
->ir_holemask
) * NBBY
));
455 bitmap
|= (inodespbit
<<
456 (nextbit
* XFS_INODES_PER_HOLEMASK_BIT
));
458 nextbit
= xfs_next_bit(&allocbitmap
, 1, nextbit
+ 1);
464 #if defined(DEBUG) || defined(XFS_WARN)
466 * Verify that an in-core inode record has a valid inode count.
469 xfs_inobt_rec_check_count(
470 struct xfs_mount
*mp
,
471 struct xfs_inobt_rec_incore
*rec
)
478 wordsz
= sizeof(allocbmap
) / sizeof(unsigned int);
479 allocbmap
= xfs_inobt_irec_to_allocmask(rec
);
481 nextbit
= xfs_next_bit((uint
*) &allocbmap
, wordsz
, nextbit
);
482 while (nextbit
!= -1) {
484 nextbit
= xfs_next_bit((uint
*) &allocbmap
, wordsz
,
488 if (inocount
!= rec
->ir_count
)
489 return -EFSCORRUPTED
;