2 * Copyright (c) 2000-2006 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
20 #include "xfs_shared.h"
21 #include "xfs_format.h"
22 #include "xfs_log_format.h"
23 #include "xfs_trans_resv.h"
24 #include "xfs_mount.h"
25 #include "xfs_inode.h"
26 #include "xfs_error.h"
27 #include "xfs_cksum.h"
28 #include "xfs_icache.h"
29 #include "xfs_trans.h"
30 #include "xfs_ialloc.h"
33 * Check that none of the inode's in the buffer have a next
34 * unlinked field of 0.
46 j
= mp
->m_inode_cluster_size
>> mp
->m_sb
.sb_inodelog
;
48 for (i
= 0; i
< j
; i
++) {
49 dip
= xfs_buf_offset(bp
, i
* mp
->m_sb
.sb_inodesize
);
50 if (!dip
->di_next_unlinked
) {
52 "Detected bogus zero next_unlinked field in inode %d buffer 0x%llx.",
53 i
, (long long)bp
->b_bn
);
60 * If we are doing readahead on an inode buffer, we might be in log recovery
61 * reading an inode allocation buffer that hasn't yet been replayed, and hence
62 * has not had the inode cores stamped into it. Hence for readahead, the buffer
63 * may be potentially invalid.
65 * If the readahead buffer is invalid, we don't want to mark it with an error,
66 * but we do want to clear the DONE status of the buffer so that a followup read
67 * will re-read it from disk. This will ensure that we don't get an unnecessary
68 * warnings during log recovery and we don't get unnecssary panics on debug
76 struct xfs_mount
*mp
= bp
->b_target
->bt_mount
;
81 * Validate the magic number and version of every inode in the buffer
83 ni
= XFS_BB_TO_FSB(mp
, bp
->b_length
) * mp
->m_sb
.sb_inopblock
;
84 for (i
= 0; i
< ni
; i
++) {
88 dip
= xfs_buf_offset(bp
, (i
<< mp
->m_sb
.sb_inodelog
));
89 di_ok
= dip
->di_magic
== cpu_to_be16(XFS_DINODE_MAGIC
) &&
90 XFS_DINODE_GOOD_VERSION(dip
->di_version
);
91 if (unlikely(XFS_TEST_ERROR(!di_ok
, mp
,
92 XFS_ERRTAG_ITOBP_INOTOBP
,
93 XFS_RANDOM_ITOBP_INOTOBP
))) {
95 bp
->b_flags
&= ~XBF_DONE
;
99 xfs_buf_ioerror(bp
, -EFSCORRUPTED
);
100 xfs_verifier_error(bp
);
103 "bad inode magic/vsn daddr %lld #%d (magic=%x)",
104 (unsigned long long)bp
->b_bn
, i
,
105 be16_to_cpu(dip
->di_magic
));
109 xfs_inobp_check(mp
, bp
);
114 xfs_inode_buf_read_verify(
117 xfs_inode_buf_verify(bp
, false);
121 xfs_inode_buf_readahead_verify(
124 xfs_inode_buf_verify(bp
, true);
128 xfs_inode_buf_write_verify(
131 xfs_inode_buf_verify(bp
, false);
134 const struct xfs_buf_ops xfs_inode_buf_ops
= {
135 .verify_read
= xfs_inode_buf_read_verify
,
136 .verify_write
= xfs_inode_buf_write_verify
,
139 const struct xfs_buf_ops xfs_inode_buf_ra_ops
= {
140 .verify_read
= xfs_inode_buf_readahead_verify
,
141 .verify_write
= xfs_inode_buf_write_verify
,
146 * This routine is called to map an inode to the buffer containing the on-disk
147 * version of the inode. It returns a pointer to the buffer containing the
148 * on-disk inode in the bpp parameter, and in the dipp parameter it returns a
149 * pointer to the on-disk inode within that buffer.
151 * If a non-zero error is returned, then the contents of bpp and dipp are
156 struct xfs_mount
*mp
,
157 struct xfs_trans
*tp
,
158 struct xfs_imap
*imap
,
159 struct xfs_dinode
**dipp
,
160 struct xfs_buf
**bpp
,
167 buf_flags
|= XBF_UNMAPPED
;
168 error
= xfs_trans_read_buf(mp
, tp
, mp
->m_ddev_targp
, imap
->im_blkno
,
169 (int)imap
->im_len
, buf_flags
, &bp
,
172 if (error
== -EAGAIN
) {
173 ASSERT(buf_flags
& XBF_TRYLOCK
);
177 if (error
== -EFSCORRUPTED
&&
178 (iget_flags
& XFS_IGET_UNTRUSTED
))
181 xfs_warn(mp
, "%s: xfs_trans_read_buf() returned error %d.",
187 *dipp
= xfs_buf_offset(bp
, imap
->im_boffset
);
192 xfs_dinode_from_disk(
196 to
->di_magic
= be16_to_cpu(from
->di_magic
);
197 to
->di_mode
= be16_to_cpu(from
->di_mode
);
198 to
->di_version
= from
->di_version
;
199 to
->di_format
= from
->di_format
;
200 to
->di_onlink
= be16_to_cpu(from
->di_onlink
);
201 to
->di_uid
= be32_to_cpu(from
->di_uid
);
202 to
->di_gid
= be32_to_cpu(from
->di_gid
);
203 to
->di_nlink
= be32_to_cpu(from
->di_nlink
);
204 to
->di_projid_lo
= be16_to_cpu(from
->di_projid_lo
);
205 to
->di_projid_hi
= be16_to_cpu(from
->di_projid_hi
);
206 memcpy(to
->di_pad
, from
->di_pad
, sizeof(to
->di_pad
));
207 to
->di_flushiter
= be16_to_cpu(from
->di_flushiter
);
208 to
->di_atime
.t_sec
= be32_to_cpu(from
->di_atime
.t_sec
);
209 to
->di_atime
.t_nsec
= be32_to_cpu(from
->di_atime
.t_nsec
);
210 to
->di_mtime
.t_sec
= be32_to_cpu(from
->di_mtime
.t_sec
);
211 to
->di_mtime
.t_nsec
= be32_to_cpu(from
->di_mtime
.t_nsec
);
212 to
->di_ctime
.t_sec
= be32_to_cpu(from
->di_ctime
.t_sec
);
213 to
->di_ctime
.t_nsec
= be32_to_cpu(from
->di_ctime
.t_nsec
);
214 to
->di_size
= be64_to_cpu(from
->di_size
);
215 to
->di_nblocks
= be64_to_cpu(from
->di_nblocks
);
216 to
->di_extsize
= be32_to_cpu(from
->di_extsize
);
217 to
->di_nextents
= be32_to_cpu(from
->di_nextents
);
218 to
->di_anextents
= be16_to_cpu(from
->di_anextents
);
219 to
->di_forkoff
= from
->di_forkoff
;
220 to
->di_aformat
= from
->di_aformat
;
221 to
->di_dmevmask
= be32_to_cpu(from
->di_dmevmask
);
222 to
->di_dmstate
= be16_to_cpu(from
->di_dmstate
);
223 to
->di_flags
= be16_to_cpu(from
->di_flags
);
224 to
->di_gen
= be32_to_cpu(from
->di_gen
);
226 if (to
->di_version
== 3) {
227 to
->di_changecount
= be64_to_cpu(from
->di_changecount
);
228 to
->di_crtime
.t_sec
= be32_to_cpu(from
->di_crtime
.t_sec
);
229 to
->di_crtime
.t_nsec
= be32_to_cpu(from
->di_crtime
.t_nsec
);
230 to
->di_flags2
= be64_to_cpu(from
->di_flags2
);
231 to
->di_ino
= be64_to_cpu(from
->di_ino
);
232 to
->di_lsn
= be64_to_cpu(from
->di_lsn
);
233 memcpy(to
->di_pad2
, from
->di_pad2
, sizeof(to
->di_pad2
));
234 uuid_copy(&to
->di_uuid
, &from
->di_uuid
);
241 xfs_icdinode_t
*from
)
243 to
->di_magic
= cpu_to_be16(from
->di_magic
);
244 to
->di_mode
= cpu_to_be16(from
->di_mode
);
245 to
->di_version
= from
->di_version
;
246 to
->di_format
= from
->di_format
;
247 to
->di_onlink
= cpu_to_be16(from
->di_onlink
);
248 to
->di_uid
= cpu_to_be32(from
->di_uid
);
249 to
->di_gid
= cpu_to_be32(from
->di_gid
);
250 to
->di_nlink
= cpu_to_be32(from
->di_nlink
);
251 to
->di_projid_lo
= cpu_to_be16(from
->di_projid_lo
);
252 to
->di_projid_hi
= cpu_to_be16(from
->di_projid_hi
);
253 memcpy(to
->di_pad
, from
->di_pad
, sizeof(to
->di_pad
));
254 to
->di_atime
.t_sec
= cpu_to_be32(from
->di_atime
.t_sec
);
255 to
->di_atime
.t_nsec
= cpu_to_be32(from
->di_atime
.t_nsec
);
256 to
->di_mtime
.t_sec
= cpu_to_be32(from
->di_mtime
.t_sec
);
257 to
->di_mtime
.t_nsec
= cpu_to_be32(from
->di_mtime
.t_nsec
);
258 to
->di_ctime
.t_sec
= cpu_to_be32(from
->di_ctime
.t_sec
);
259 to
->di_ctime
.t_nsec
= cpu_to_be32(from
->di_ctime
.t_nsec
);
260 to
->di_size
= cpu_to_be64(from
->di_size
);
261 to
->di_nblocks
= cpu_to_be64(from
->di_nblocks
);
262 to
->di_extsize
= cpu_to_be32(from
->di_extsize
);
263 to
->di_nextents
= cpu_to_be32(from
->di_nextents
);
264 to
->di_anextents
= cpu_to_be16(from
->di_anextents
);
265 to
->di_forkoff
= from
->di_forkoff
;
266 to
->di_aformat
= from
->di_aformat
;
267 to
->di_dmevmask
= cpu_to_be32(from
->di_dmevmask
);
268 to
->di_dmstate
= cpu_to_be16(from
->di_dmstate
);
269 to
->di_flags
= cpu_to_be16(from
->di_flags
);
270 to
->di_gen
= cpu_to_be32(from
->di_gen
);
272 if (from
->di_version
== 3) {
273 to
->di_changecount
= cpu_to_be64(from
->di_changecount
);
274 to
->di_crtime
.t_sec
= cpu_to_be32(from
->di_crtime
.t_sec
);
275 to
->di_crtime
.t_nsec
= cpu_to_be32(from
->di_crtime
.t_nsec
);
276 to
->di_flags2
= cpu_to_be64(from
->di_flags2
);
277 to
->di_ino
= cpu_to_be64(from
->di_ino
);
278 to
->di_lsn
= cpu_to_be64(from
->di_lsn
);
279 memcpy(to
->di_pad2
, from
->di_pad2
, sizeof(to
->di_pad2
));
280 uuid_copy(&to
->di_uuid
, &from
->di_uuid
);
281 to
->di_flushiter
= 0;
283 to
->di_flushiter
= cpu_to_be16(from
->di_flushiter
);
289 struct xfs_mount
*mp
,
290 struct xfs_inode
*ip
,
291 struct xfs_dinode
*dip
)
293 if (dip
->di_magic
!= cpu_to_be16(XFS_DINODE_MAGIC
))
296 /* only version 3 or greater inodes are extensively verified here */
297 if (dip
->di_version
< 3)
300 if (!xfs_sb_version_hascrc(&mp
->m_sb
))
302 if (!xfs_verify_cksum((char *)dip
, mp
->m_sb
.sb_inodesize
,
305 if (be64_to_cpu(dip
->di_ino
) != ip
->i_ino
)
307 if (!uuid_equal(&dip
->di_uuid
, &mp
->m_sb
.sb_uuid
))
314 struct xfs_mount
*mp
,
315 struct xfs_dinode
*dip
)
319 if (dip
->di_version
< 3)
322 ASSERT(xfs_sb_version_hascrc(&mp
->m_sb
));
323 crc
= xfs_start_cksum((char *)dip
, mp
->m_sb
.sb_inodesize
,
325 dip
->di_crc
= xfs_end_cksum(crc
);
329 * Read the disk inode attributes into the in-core inode structure.
331 * For version 5 superblocks, if we are initialising a new inode and we are not
332 * utilising the XFS_MOUNT_IKEEP inode cluster mode, we can simple build the new
333 * inode core with a random generation number. If we are keeping inodes around,
334 * we need to read the inode cluster to get the existing generation number off
335 * disk. Further, if we are using version 4 superblocks (i.e. v1/v2 inode
336 * format) then log recovery is dependent on the di_flushiter field being
337 * initialised from the current on-disk value and hence we must also read the
352 * Fill in the location information in the in-core inode.
354 error
= xfs_imap(mp
, tp
, ip
->i_ino
, &ip
->i_imap
, iget_flags
);
358 /* shortcut IO on inode allocation if possible */
359 if ((iget_flags
& XFS_IGET_CREATE
) &&
360 xfs_sb_version_hascrc(&mp
->m_sb
) &&
361 !(mp
->m_flags
& XFS_MOUNT_IKEEP
)) {
362 /* initialise the on-disk inode core */
363 memset(&ip
->i_d
, 0, sizeof(ip
->i_d
));
364 ip
->i_d
.di_magic
= XFS_DINODE_MAGIC
;
365 ip
->i_d
.di_gen
= prandom_u32();
366 if (xfs_sb_version_hascrc(&mp
->m_sb
)) {
367 ip
->i_d
.di_version
= 3;
368 ip
->i_d
.di_ino
= ip
->i_ino
;
369 uuid_copy(&ip
->i_d
.di_uuid
, &mp
->m_sb
.sb_uuid
);
371 ip
->i_d
.di_version
= 2;
376 * Get pointers to the on-disk inode and the buffer containing it.
378 error
= xfs_imap_to_bp(mp
, tp
, &ip
->i_imap
, &dip
, &bp
, 0, iget_flags
);
382 /* even unallocated inodes are verified */
383 if (!xfs_dinode_verify(mp
, ip
, dip
)) {
384 xfs_alert(mp
, "%s: validation failed for inode %lld failed",
385 __func__
, ip
->i_ino
);
387 XFS_CORRUPTION_ERROR(__func__
, XFS_ERRLEVEL_LOW
, mp
, dip
);
388 error
= -EFSCORRUPTED
;
393 * If the on-disk inode is already linked to a directory
394 * entry, copy all of the inode into the in-core inode.
395 * xfs_iformat_fork() handles copying in the inode format
396 * specific information.
397 * Otherwise, just get the truly permanent information.
400 xfs_dinode_from_disk(&ip
->i_d
, dip
);
401 error
= xfs_iformat_fork(ip
, dip
);
404 xfs_alert(mp
, "%s: xfs_iformat() returned error %d",
411 * Partial initialisation of the in-core inode. Just the bits
412 * that xfs_ialloc won't overwrite or relies on being correct.
414 ip
->i_d
.di_magic
= be16_to_cpu(dip
->di_magic
);
415 ip
->i_d
.di_version
= dip
->di_version
;
416 ip
->i_d
.di_gen
= be32_to_cpu(dip
->di_gen
);
417 ip
->i_d
.di_flushiter
= be16_to_cpu(dip
->di_flushiter
);
419 if (dip
->di_version
== 3) {
420 ip
->i_d
.di_ino
= be64_to_cpu(dip
->di_ino
);
421 uuid_copy(&ip
->i_d
.di_uuid
, &dip
->di_uuid
);
425 * Make sure to pull in the mode here as well in
426 * case the inode is released without being used.
427 * This ensures that xfs_inactive() will see that
428 * the inode is already free and not try to mess
429 * with the uninitialized part of it.
435 * Automatically convert version 1 inode formats in memory to version 2
436 * inode format. If the inode is modified, it will get logged and
437 * rewritten as a version 2 inode. We can do this because we set the
438 * superblock feature bit for v2 inodes unconditionally during mount
439 * and it means the reast of the code can assume the inode version is 2
442 if (ip
->i_d
.di_version
== 1) {
443 ip
->i_d
.di_version
= 2;
444 memset(&(ip
->i_d
.di_pad
[0]), 0, sizeof(ip
->i_d
.di_pad
));
445 ip
->i_d
.di_nlink
= ip
->i_d
.di_onlink
;
446 ip
->i_d
.di_onlink
= 0;
447 xfs_set_projid(ip
, 0);
450 ip
->i_delayed_blks
= 0;
453 * Mark the buffer containing the inode as something to keep
454 * around for a while. This helps to keep recently accessed
455 * meta-data in-core longer.
457 xfs_buf_set_ref(bp
, XFS_INO_REF
);
460 * Use xfs_trans_brelse() to release the buffer containing the on-disk
461 * inode, because it was acquired with xfs_trans_read_buf() in
462 * xfs_imap_to_bp() above. If tp is NULL, this is just a normal
463 * brelse(). If we're within a transaction, then xfs_trans_brelse()
464 * will only release the buffer if it is not dirty within the
465 * transaction. It will be OK to release the buffer in this case,
466 * because inodes on disk are never destroyed and we will be locking the
467 * new in-core inode before putting it in the cache where other
468 * processes can find it. Thus we don't have to worry about the inode
469 * being changed just because we released the buffer.
472 xfs_trans_brelse(tp
, bp
);