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
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"
24 #include "xfs_mount.h"
25 #include "xfs_defer.h"
26 #include "xfs_inode.h"
27 #include "xfs_cksum.h"
28 #include "xfs_trans.h"
29 #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 xfs_dinode_good_version(
64 if (xfs_sb_version_hascrc(&mp
->m_sb
))
67 return version
== 1 || version
== 2;
71 * If we are doing readahead on an inode buffer, we might be in log recovery
72 * reading an inode allocation buffer that hasn't yet been replayed, and hence
73 * has not had the inode cores stamped into it. Hence for readahead, the buffer
74 * may be potentially invalid.
76 * If the readahead buffer is invalid, we need to mark it with an error and
77 * clear the DONE status of the buffer so that a followup read will re-read it
78 * from disk. We don't report the error otherwise to avoid warnings during log
79 * recovery and we don't get unnecssary panics on debug kernels. We use EIO here
80 * because all we want to do is say readahead failed; there is no-one to report
81 * the error to, so this will distinguish it from a non-ra verifier failure.
82 * Changes to this readahead error behavour also need to be reflected in
83 * xfs_dquot_buf_readahead_verify().
90 struct xfs_mount
*mp
= bp
->b_target
->bt_mount
;
95 * Validate the magic number and version of every inode in the buffer
97 ni
= XFS_BB_TO_FSB(mp
, bp
->b_length
) * mp
->m_sb
.sb_inopblock
;
98 for (i
= 0; i
< ni
; i
++) {
102 dip
= xfs_buf_offset(bp
, (i
<< mp
->m_sb
.sb_inodelog
));
103 di_ok
= dip
->di_magic
== cpu_to_be16(XFS_DINODE_MAGIC
) &&
104 xfs_dinode_good_version(mp
, dip
->di_version
);
105 if (unlikely(XFS_TEST_ERROR(!di_ok
, mp
,
106 XFS_ERRTAG_ITOBP_INOTOBP
,
107 XFS_RANDOM_ITOBP_INOTOBP
))) {
109 bp
->b_flags
&= ~XBF_DONE
;
110 xfs_buf_ioerror(bp
, -EIO
);
114 xfs_buf_ioerror(bp
, -EFSCORRUPTED
);
115 xfs_verifier_error(bp
);
118 "bad inode magic/vsn daddr %lld #%d (magic=%x)",
119 (unsigned long long)bp
->b_bn
, i
,
120 be16_to_cpu(dip
->di_magic
));
124 xfs_inobp_check(mp
, bp
);
129 xfs_inode_buf_read_verify(
132 xfs_inode_buf_verify(bp
, false);
136 xfs_inode_buf_readahead_verify(
139 xfs_inode_buf_verify(bp
, true);
143 xfs_inode_buf_write_verify(
146 xfs_inode_buf_verify(bp
, false);
149 const struct xfs_buf_ops xfs_inode_buf_ops
= {
151 .verify_read
= xfs_inode_buf_read_verify
,
152 .verify_write
= xfs_inode_buf_write_verify
,
155 const struct xfs_buf_ops xfs_inode_buf_ra_ops
= {
156 .name
= "xxfs_inode_ra",
157 .verify_read
= xfs_inode_buf_readahead_verify
,
158 .verify_write
= xfs_inode_buf_write_verify
,
163 * This routine is called to map an inode to the buffer containing the on-disk
164 * version of the inode. It returns a pointer to the buffer containing the
165 * on-disk inode in the bpp parameter, and in the dipp parameter it returns a
166 * pointer to the on-disk inode within that buffer.
168 * If a non-zero error is returned, then the contents of bpp and dipp are
173 struct xfs_mount
*mp
,
174 struct xfs_trans
*tp
,
175 struct xfs_imap
*imap
,
176 struct xfs_dinode
**dipp
,
177 struct xfs_buf
**bpp
,
184 buf_flags
|= XBF_UNMAPPED
;
185 error
= xfs_trans_read_buf(mp
, tp
, mp
->m_ddev_targp
, imap
->im_blkno
,
186 (int)imap
->im_len
, buf_flags
, &bp
,
189 if (error
== -EAGAIN
) {
190 ASSERT(buf_flags
& XBF_TRYLOCK
);
194 if (error
== -EFSCORRUPTED
&&
195 (iget_flags
& XFS_IGET_UNTRUSTED
))
198 xfs_warn(mp
, "%s: xfs_trans_read_buf() returned error %d.",
204 *dipp
= xfs_buf_offset(bp
, imap
->im_boffset
);
210 struct xfs_inode
*ip
,
211 struct xfs_dinode
*from
)
213 struct xfs_icdinode
*to
= &ip
->i_d
;
214 struct inode
*inode
= VFS_I(ip
);
218 * Convert v1 inodes immediately to v2 inode format as this is the
219 * minimum inode version format we support in the rest of the code.
221 to
->di_version
= from
->di_version
;
222 if (to
->di_version
== 1) {
223 set_nlink(inode
, be16_to_cpu(from
->di_onlink
));
224 to
->di_projid_lo
= 0;
225 to
->di_projid_hi
= 0;
228 set_nlink(inode
, be32_to_cpu(from
->di_nlink
));
229 to
->di_projid_lo
= be16_to_cpu(from
->di_projid_lo
);
230 to
->di_projid_hi
= be16_to_cpu(from
->di_projid_hi
);
233 to
->di_format
= from
->di_format
;
234 to
->di_uid
= be32_to_cpu(from
->di_uid
);
235 to
->di_gid
= be32_to_cpu(from
->di_gid
);
236 to
->di_flushiter
= be16_to_cpu(from
->di_flushiter
);
239 * Time is signed, so need to convert to signed 32 bit before
240 * storing in inode timestamp which may be 64 bit. Otherwise
241 * a time before epoch is converted to a time long after epoch
244 inode
->i_atime
.tv_sec
= (int)be32_to_cpu(from
->di_atime
.t_sec
);
245 inode
->i_atime
.tv_nsec
= (int)be32_to_cpu(from
->di_atime
.t_nsec
);
246 inode
->i_mtime
.tv_sec
= (int)be32_to_cpu(from
->di_mtime
.t_sec
);
247 inode
->i_mtime
.tv_nsec
= (int)be32_to_cpu(from
->di_mtime
.t_nsec
);
248 inode
->i_ctime
.tv_sec
= (int)be32_to_cpu(from
->di_ctime
.t_sec
);
249 inode
->i_ctime
.tv_nsec
= (int)be32_to_cpu(from
->di_ctime
.t_nsec
);
250 inode
->i_generation
= be32_to_cpu(from
->di_gen
);
251 inode
->i_mode
= be16_to_cpu(from
->di_mode
);
253 to
->di_size
= be64_to_cpu(from
->di_size
);
254 to
->di_nblocks
= be64_to_cpu(from
->di_nblocks
);
255 to
->di_extsize
= be32_to_cpu(from
->di_extsize
);
256 to
->di_nextents
= be32_to_cpu(from
->di_nextents
);
257 to
->di_anextents
= be16_to_cpu(from
->di_anextents
);
258 to
->di_forkoff
= from
->di_forkoff
;
259 to
->di_aformat
= from
->di_aformat
;
260 to
->di_dmevmask
= be32_to_cpu(from
->di_dmevmask
);
261 to
->di_dmstate
= be16_to_cpu(from
->di_dmstate
);
262 to
->di_flags
= be16_to_cpu(from
->di_flags
);
264 if (to
->di_version
== 3) {
265 inode
->i_version
= be64_to_cpu(from
->di_changecount
);
266 to
->di_crtime
.t_sec
= be32_to_cpu(from
->di_crtime
.t_sec
);
267 to
->di_crtime
.t_nsec
= be32_to_cpu(from
->di_crtime
.t_nsec
);
268 to
->di_flags2
= be64_to_cpu(from
->di_flags2
);
269 to
->di_cowextsize
= be32_to_cpu(from
->di_cowextsize
);
275 struct xfs_inode
*ip
,
276 struct xfs_dinode
*to
,
279 struct xfs_icdinode
*from
= &ip
->i_d
;
280 struct inode
*inode
= VFS_I(ip
);
282 to
->di_magic
= cpu_to_be16(XFS_DINODE_MAGIC
);
285 to
->di_version
= from
->di_version
;
286 to
->di_format
= from
->di_format
;
287 to
->di_uid
= cpu_to_be32(from
->di_uid
);
288 to
->di_gid
= cpu_to_be32(from
->di_gid
);
289 to
->di_projid_lo
= cpu_to_be16(from
->di_projid_lo
);
290 to
->di_projid_hi
= cpu_to_be16(from
->di_projid_hi
);
292 memset(to
->di_pad
, 0, sizeof(to
->di_pad
));
293 to
->di_atime
.t_sec
= cpu_to_be32(inode
->i_atime
.tv_sec
);
294 to
->di_atime
.t_nsec
= cpu_to_be32(inode
->i_atime
.tv_nsec
);
295 to
->di_mtime
.t_sec
= cpu_to_be32(inode
->i_mtime
.tv_sec
);
296 to
->di_mtime
.t_nsec
= cpu_to_be32(inode
->i_mtime
.tv_nsec
);
297 to
->di_ctime
.t_sec
= cpu_to_be32(inode
->i_ctime
.tv_sec
);
298 to
->di_ctime
.t_nsec
= cpu_to_be32(inode
->i_ctime
.tv_nsec
);
299 to
->di_nlink
= cpu_to_be32(inode
->i_nlink
);
300 to
->di_gen
= cpu_to_be32(inode
->i_generation
);
301 to
->di_mode
= cpu_to_be16(inode
->i_mode
);
303 to
->di_size
= cpu_to_be64(from
->di_size
);
304 to
->di_nblocks
= cpu_to_be64(from
->di_nblocks
);
305 to
->di_extsize
= cpu_to_be32(from
->di_extsize
);
306 to
->di_nextents
= cpu_to_be32(from
->di_nextents
);
307 to
->di_anextents
= cpu_to_be16(from
->di_anextents
);
308 to
->di_forkoff
= from
->di_forkoff
;
309 to
->di_aformat
= from
->di_aformat
;
310 to
->di_dmevmask
= cpu_to_be32(from
->di_dmevmask
);
311 to
->di_dmstate
= cpu_to_be16(from
->di_dmstate
);
312 to
->di_flags
= cpu_to_be16(from
->di_flags
);
314 if (from
->di_version
== 3) {
315 to
->di_changecount
= cpu_to_be64(inode
->i_version
);
316 to
->di_crtime
.t_sec
= cpu_to_be32(from
->di_crtime
.t_sec
);
317 to
->di_crtime
.t_nsec
= cpu_to_be32(from
->di_crtime
.t_nsec
);
318 to
->di_flags2
= cpu_to_be64(from
->di_flags2
);
319 to
->di_cowextsize
= cpu_to_be32(from
->di_cowextsize
);
320 to
->di_ino
= cpu_to_be64(ip
->i_ino
);
321 to
->di_lsn
= cpu_to_be64(lsn
);
322 memset(to
->di_pad2
, 0, sizeof(to
->di_pad2
));
323 uuid_copy(&to
->di_uuid
, &ip
->i_mount
->m_sb
.sb_meta_uuid
);
324 to
->di_flushiter
= 0;
326 to
->di_flushiter
= cpu_to_be16(from
->di_flushiter
);
331 xfs_log_dinode_to_disk(
332 struct xfs_log_dinode
*from
,
333 struct xfs_dinode
*to
)
335 to
->di_magic
= cpu_to_be16(from
->di_magic
);
336 to
->di_mode
= cpu_to_be16(from
->di_mode
);
337 to
->di_version
= from
->di_version
;
338 to
->di_format
= from
->di_format
;
340 to
->di_uid
= cpu_to_be32(from
->di_uid
);
341 to
->di_gid
= cpu_to_be32(from
->di_gid
);
342 to
->di_nlink
= cpu_to_be32(from
->di_nlink
);
343 to
->di_projid_lo
= cpu_to_be16(from
->di_projid_lo
);
344 to
->di_projid_hi
= cpu_to_be16(from
->di_projid_hi
);
345 memcpy(to
->di_pad
, from
->di_pad
, sizeof(to
->di_pad
));
347 to
->di_atime
.t_sec
= cpu_to_be32(from
->di_atime
.t_sec
);
348 to
->di_atime
.t_nsec
= cpu_to_be32(from
->di_atime
.t_nsec
);
349 to
->di_mtime
.t_sec
= cpu_to_be32(from
->di_mtime
.t_sec
);
350 to
->di_mtime
.t_nsec
= cpu_to_be32(from
->di_mtime
.t_nsec
);
351 to
->di_ctime
.t_sec
= cpu_to_be32(from
->di_ctime
.t_sec
);
352 to
->di_ctime
.t_nsec
= cpu_to_be32(from
->di_ctime
.t_nsec
);
354 to
->di_size
= cpu_to_be64(from
->di_size
);
355 to
->di_nblocks
= cpu_to_be64(from
->di_nblocks
);
356 to
->di_extsize
= cpu_to_be32(from
->di_extsize
);
357 to
->di_nextents
= cpu_to_be32(from
->di_nextents
);
358 to
->di_anextents
= cpu_to_be16(from
->di_anextents
);
359 to
->di_forkoff
= from
->di_forkoff
;
360 to
->di_aformat
= from
->di_aformat
;
361 to
->di_dmevmask
= cpu_to_be32(from
->di_dmevmask
);
362 to
->di_dmstate
= cpu_to_be16(from
->di_dmstate
);
363 to
->di_flags
= cpu_to_be16(from
->di_flags
);
364 to
->di_gen
= cpu_to_be32(from
->di_gen
);
366 if (from
->di_version
== 3) {
367 to
->di_changecount
= cpu_to_be64(from
->di_changecount
);
368 to
->di_crtime
.t_sec
= cpu_to_be32(from
->di_crtime
.t_sec
);
369 to
->di_crtime
.t_nsec
= cpu_to_be32(from
->di_crtime
.t_nsec
);
370 to
->di_flags2
= cpu_to_be64(from
->di_flags2
);
371 to
->di_cowextsize
= cpu_to_be32(from
->di_cowextsize
);
372 to
->di_ino
= cpu_to_be64(from
->di_ino
);
373 to
->di_lsn
= cpu_to_be64(from
->di_lsn
);
374 memcpy(to
->di_pad2
, from
->di_pad2
, sizeof(to
->di_pad2
));
375 uuid_copy(&to
->di_uuid
, &from
->di_uuid
);
376 to
->di_flushiter
= 0;
378 to
->di_flushiter
= cpu_to_be16(from
->di_flushiter
);
384 struct xfs_mount
*mp
,
386 struct xfs_dinode
*dip
)
392 if (dip
->di_magic
!= cpu_to_be16(XFS_DINODE_MAGIC
))
395 /* don't allow invalid i_size */
396 if (be64_to_cpu(dip
->di_size
) & (1ULL << 63))
399 mode
= be16_to_cpu(dip
->di_mode
);
400 if (mode
&& xfs_mode_to_ftype(mode
) == XFS_DIR3_FT_UNKNOWN
)
403 /* No zero-length symlinks/dirs. */
404 if ((S_ISLNK(mode
) || S_ISDIR(mode
)) && dip
->di_size
== 0)
407 /* only version 3 or greater inodes are extensively verified here */
408 if (dip
->di_version
< 3)
411 if (!xfs_sb_version_hascrc(&mp
->m_sb
))
413 if (!xfs_verify_cksum((char *)dip
, mp
->m_sb
.sb_inodesize
,
416 if (be64_to_cpu(dip
->di_ino
) != ino
)
418 if (!uuid_equal(&dip
->di_uuid
, &mp
->m_sb
.sb_meta_uuid
))
421 flags
= be16_to_cpu(dip
->di_flags
);
422 flags2
= be64_to_cpu(dip
->di_flags2
);
424 /* don't allow reflink/cowextsize if we don't have reflink */
425 if ((flags2
& (XFS_DIFLAG2_REFLINK
| XFS_DIFLAG2_COWEXTSIZE
)) &&
426 !xfs_sb_version_hasreflink(&mp
->m_sb
))
429 /* don't let reflink and realtime mix */
430 if ((flags2
& XFS_DIFLAG2_REFLINK
) && (flags
& XFS_DIFLAG_REALTIME
))
433 /* don't let reflink and dax mix */
434 if ((flags2
& XFS_DIFLAG2_REFLINK
) && (flags2
& XFS_DIFLAG2_DAX
))
442 struct xfs_mount
*mp
,
443 struct xfs_dinode
*dip
)
447 if (dip
->di_version
< 3)
450 ASSERT(xfs_sb_version_hascrc(&mp
->m_sb
));
451 crc
= xfs_start_cksum_update((char *)dip
, mp
->m_sb
.sb_inodesize
,
453 dip
->di_crc
= xfs_end_cksum(crc
);
457 * Read the disk inode attributes into the in-core inode structure.
459 * For version 5 superblocks, if we are initialising a new inode and we are not
460 * utilising the XFS_MOUNT_IKEEP inode cluster mode, we can simple build the new
461 * inode core with a random generation number. If we are keeping inodes around,
462 * we need to read the inode cluster to get the existing generation number off
463 * disk. Further, if we are using version 4 superblocks (i.e. v1/v2 inode
464 * format) then log recovery is dependent on the di_flushiter field being
465 * initialised from the current on-disk value and hence we must also read the
480 * Fill in the location information in the in-core inode.
482 error
= xfs_imap(mp
, tp
, ip
->i_ino
, &ip
->i_imap
, iget_flags
);
486 /* shortcut IO on inode allocation if possible */
487 if ((iget_flags
& XFS_IGET_CREATE
) &&
488 xfs_sb_version_hascrc(&mp
->m_sb
) &&
489 !(mp
->m_flags
& XFS_MOUNT_IKEEP
)) {
490 /* initialise the on-disk inode core */
491 memset(&ip
->i_d
, 0, sizeof(ip
->i_d
));
492 VFS_I(ip
)->i_generation
= prandom_u32();
493 if (xfs_sb_version_hascrc(&mp
->m_sb
))
494 ip
->i_d
.di_version
= 3;
496 ip
->i_d
.di_version
= 2;
501 * Get pointers to the on-disk inode and the buffer containing it.
503 error
= xfs_imap_to_bp(mp
, tp
, &ip
->i_imap
, &dip
, &bp
, 0, iget_flags
);
507 /* even unallocated inodes are verified */
508 if (!xfs_dinode_verify(mp
, ip
->i_ino
, dip
)) {
509 xfs_alert(mp
, "%s: validation failed for inode %lld",
510 __func__
, ip
->i_ino
);
512 XFS_CORRUPTION_ERROR(__func__
, XFS_ERRLEVEL_LOW
, mp
, dip
);
513 error
= -EFSCORRUPTED
;
518 * If the on-disk inode is already linked to a directory
519 * entry, copy all of the inode into the in-core inode.
520 * xfs_iformat_fork() handles copying in the inode format
521 * specific information.
522 * Otherwise, just get the truly permanent information.
525 xfs_inode_from_disk(ip
, dip
);
526 error
= xfs_iformat_fork(ip
, dip
);
529 xfs_alert(mp
, "%s: xfs_iformat() returned error %d",
536 * Partial initialisation of the in-core inode. Just the bits
537 * that xfs_ialloc won't overwrite or relies on being correct.
539 ip
->i_d
.di_version
= dip
->di_version
;
540 VFS_I(ip
)->i_generation
= be32_to_cpu(dip
->di_gen
);
541 ip
->i_d
.di_flushiter
= be16_to_cpu(dip
->di_flushiter
);
544 * Make sure to pull in the mode here as well in
545 * case the inode is released without being used.
546 * This ensures that xfs_inactive() will see that
547 * the inode is already free and not try to mess
548 * with the uninitialized part of it.
550 VFS_I(ip
)->i_mode
= 0;
553 ASSERT(ip
->i_d
.di_version
>= 2);
554 ip
->i_delayed_blks
= 0;
557 * Mark the buffer containing the inode as something to keep
558 * around for a while. This helps to keep recently accessed
559 * meta-data in-core longer.
561 xfs_buf_set_ref(bp
, XFS_INO_REF
);
564 * Use xfs_trans_brelse() to release the buffer containing the on-disk
565 * inode, because it was acquired with xfs_trans_read_buf() in
566 * xfs_imap_to_bp() above. If tp is NULL, this is just a normal
567 * brelse(). If we're within a transaction, then xfs_trans_brelse()
568 * will only release the buffer if it is not dirty within the
569 * transaction. It will be OK to release the buffer in this case,
570 * because inodes on disk are never destroyed and we will be locking the
571 * new in-core inode before putting it in the cache where other
572 * processes can find it. Thus we don't have to worry about the inode
573 * being changed just because we released the buffer.
576 xfs_trans_brelse(tp
, bp
);