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_defer.h"
26 #include "xfs_inode.h"
27 #include "xfs_errortag.h"
28 #include "xfs_error.h"
29 #include "xfs_cksum.h"
30 #include "xfs_icache.h"
31 #include "xfs_trans.h"
32 #include "xfs_ialloc.h"
35 #include <linux/iversion.h>
38 * Check that none of the inode's in the buffer have a next
39 * unlinked field of 0.
51 j = mp->m_inode_cluster_size >> mp->m_sb.sb_inodelog;
53 for (i = 0; i < j; i++) {
54 dip = xfs_buf_offset(bp, i * mp->m_sb.sb_inodesize);
55 if (!dip->di_next_unlinked) {
57 "Detected bogus zero next_unlinked field in inode %d buffer 0x%llx.",
58 i, (long long)bp->b_bn);
65 xfs_dinode_good_version(
69 if (xfs_sb_version_hascrc(&mp->m_sb))
72 return version == 1 || version == 2;
76 * If we are doing readahead on an inode buffer, we might be in log recovery
77 * reading an inode allocation buffer that hasn't yet been replayed, and hence
78 * has not had the inode cores stamped into it. Hence for readahead, the buffer
79 * may be potentially invalid.
81 * If the readahead buffer is invalid, we need to mark it with an error and
82 * clear the DONE status of the buffer so that a followup read will re-read it
83 * from disk. We don't report the error otherwise to avoid warnings during log
84 * recovery and we don't get unnecssary panics on debug kernels. We use EIO here
85 * because all we want to do is say readahead failed; there is no-one to report
86 * the error to, so this will distinguish it from a non-ra verifier failure.
87 * Changes to this readahead error behavour also need to be reflected in
88 * xfs_dquot_buf_readahead_verify().
95 struct xfs_mount *mp = bp->b_target->bt_mount;
100 * Validate the magic number and version of every inode in the buffer
102 ni = XFS_BB_TO_FSB(mp, bp->b_length) * mp->m_sb.sb_inopblock;
103 for (i = 0; i < ni; i++) {
107 dip = xfs_buf_offset(bp, (i << mp->m_sb.sb_inodelog));
108 di_ok = dip->di_magic == cpu_to_be16(XFS_DINODE_MAGIC) &&
109 xfs_dinode_good_version(mp, dip->di_version);
110 if (unlikely(XFS_TEST_ERROR(!di_ok, mp,
111 XFS_ERRTAG_ITOBP_INOTOBP))) {
113 bp->b_flags &= ~XBF_DONE;
114 xfs_buf_ioerror(bp, -EIO);
118 xfs_buf_ioerror(bp, -EFSCORRUPTED);
119 xfs_verifier_error(bp);
122 "bad inode magic/vsn daddr %lld #%d (magic=%x)",
123 (unsigned long long)bp->b_bn, i,
124 be16_to_cpu(dip->di_magic));
128 xfs_inobp_check(mp, bp);
133 xfs_inode_buf_read_verify(
136 xfs_inode_buf_verify(bp, false);
140 xfs_inode_buf_readahead_verify(
143 xfs_inode_buf_verify(bp, true);
147 xfs_inode_buf_write_verify(
150 xfs_inode_buf_verify(bp, false);
153 const struct xfs_buf_ops xfs_inode_buf_ops = {
155 .verify_read = xfs_inode_buf_read_verify,
156 .verify_write = xfs_inode_buf_write_verify,
159 const struct xfs_buf_ops xfs_inode_buf_ra_ops = {
160 .name = "xxfs_inode_ra",
161 .verify_read = xfs_inode_buf_readahead_verify,
162 .verify_write = xfs_inode_buf_write_verify,
167 * This routine is called to map an inode to the buffer containing the on-disk
168 * version of the inode. It returns a pointer to the buffer containing the
169 * on-disk inode in the bpp parameter, and in the dipp parameter it returns a
170 * pointer to the on-disk inode within that buffer.
172 * If a non-zero error is returned, then the contents of bpp and dipp are
177 struct xfs_mount *mp,
178 struct xfs_trans *tp,
179 struct xfs_imap *imap,
180 struct xfs_dinode **dipp,
181 struct xfs_buf **bpp,
188 buf_flags |= XBF_UNMAPPED;
189 error = xfs_trans_read_buf(mp, tp, mp->m_ddev_targp, imap->im_blkno,
190 (int)imap->im_len, buf_flags, &bp,
193 if (error == -EAGAIN) {
194 ASSERT(buf_flags & XBF_TRYLOCK);
198 if (error == -EFSCORRUPTED &&
199 (iget_flags & XFS_IGET_UNTRUSTED))
202 xfs_warn(mp, "%s: xfs_trans_read_buf() returned error %d.",
208 *dipp = xfs_buf_offset(bp, imap->im_boffset);
214 struct xfs_inode *ip,
215 struct xfs_dinode *from)
217 struct xfs_icdinode *to = &ip->i_d;
218 struct inode *inode = VFS_I(ip);
222 * Convert v1 inodes immediately to v2 inode format as this is the
223 * minimum inode version format we support in the rest of the code.
225 to->di_version = from->di_version;
226 if (to->di_version == 1) {
227 set_nlink(inode, be16_to_cpu(from->di_onlink));
228 to->di_projid_lo = 0;
229 to->di_projid_hi = 0;
232 set_nlink(inode, be32_to_cpu(from->di_nlink));
233 to->di_projid_lo = be16_to_cpu(from->di_projid_lo);
234 to->di_projid_hi = be16_to_cpu(from->di_projid_hi);
237 to->di_format = from->di_format;
238 to->di_uid = be32_to_cpu(from->di_uid);
239 to->di_gid = be32_to_cpu(from->di_gid);
240 to->di_flushiter = be16_to_cpu(from->di_flushiter);
243 * Time is signed, so need to convert to signed 32 bit before
244 * storing in inode timestamp which may be 64 bit. Otherwise
245 * a time before epoch is converted to a time long after epoch
248 inode->i_atime.tv_sec = (int)be32_to_cpu(from->di_atime.t_sec);
249 inode->i_atime.tv_nsec = (int)be32_to_cpu(from->di_atime.t_nsec);
250 inode->i_mtime.tv_sec = (int)be32_to_cpu(from->di_mtime.t_sec);
251 inode->i_mtime.tv_nsec = (int)be32_to_cpu(from->di_mtime.t_nsec);
252 inode->i_ctime.tv_sec = (int)be32_to_cpu(from->di_ctime.t_sec);
253 inode->i_ctime.tv_nsec = (int)be32_to_cpu(from->di_ctime.t_nsec);
254 inode->i_generation = be32_to_cpu(from->di_gen);
255 inode->i_mode = be16_to_cpu(from->di_mode);
257 to->di_size = be64_to_cpu(from->di_size);
258 to->di_nblocks = be64_to_cpu(from->di_nblocks);
259 to->di_extsize = be32_to_cpu(from->di_extsize);
260 to->di_nextents = be32_to_cpu(from->di_nextents);
261 to->di_anextents = be16_to_cpu(from->di_anextents);
262 to->di_forkoff = from->di_forkoff;
263 to->di_aformat = from->di_aformat;
264 to->di_dmevmask = be32_to_cpu(from->di_dmevmask);
265 to->di_dmstate = be16_to_cpu(from->di_dmstate);
266 to->di_flags = be16_to_cpu(from->di_flags);
268 if (to->di_version == 3) {
269 inode_set_iversion_queried(inode,
270 be64_to_cpu(from->di_changecount));
271 to->di_crtime.t_sec = be32_to_cpu(from->di_crtime.t_sec);
272 to->di_crtime.t_nsec = be32_to_cpu(from->di_crtime.t_nsec);
273 to->di_flags2 = be64_to_cpu(from->di_flags2);
274 to->di_cowextsize = be32_to_cpu(from->di_cowextsize);
280 struct xfs_inode *ip,
281 struct xfs_dinode *to,
284 struct xfs_icdinode *from = &ip->i_d;
285 struct inode *inode = VFS_I(ip);
287 to->di_magic = cpu_to_be16(XFS_DINODE_MAGIC);
290 to->di_version = from->di_version;
291 to->di_format = from->di_format;
292 to->di_uid = cpu_to_be32(from->di_uid);
293 to->di_gid = cpu_to_be32(from->di_gid);
294 to->di_projid_lo = cpu_to_be16(from->di_projid_lo);
295 to->di_projid_hi = cpu_to_be16(from->di_projid_hi);
297 memset(to->di_pad, 0, sizeof(to->di_pad));
298 to->di_atime.t_sec = cpu_to_be32(inode->i_atime.tv_sec);
299 to->di_atime.t_nsec = cpu_to_be32(inode->i_atime.tv_nsec);
300 to->di_mtime.t_sec = cpu_to_be32(inode->i_mtime.tv_sec);
301 to->di_mtime.t_nsec = cpu_to_be32(inode->i_mtime.tv_nsec);
302 to->di_ctime.t_sec = cpu_to_be32(inode->i_ctime.tv_sec);
303 to->di_ctime.t_nsec = cpu_to_be32(inode->i_ctime.tv_nsec);
304 to->di_nlink = cpu_to_be32(inode->i_nlink);
305 to->di_gen = cpu_to_be32(inode->i_generation);
306 to->di_mode = cpu_to_be16(inode->i_mode);
308 to->di_size = cpu_to_be64(from->di_size);
309 to->di_nblocks = cpu_to_be64(from->di_nblocks);
310 to->di_extsize = cpu_to_be32(from->di_extsize);
311 to->di_nextents = cpu_to_be32(from->di_nextents);
312 to->di_anextents = cpu_to_be16(from->di_anextents);
313 to->di_forkoff = from->di_forkoff;
314 to->di_aformat = from->di_aformat;
315 to->di_dmevmask = cpu_to_be32(from->di_dmevmask);
316 to->di_dmstate = cpu_to_be16(from->di_dmstate);
317 to->di_flags = cpu_to_be16(from->di_flags);
319 if (from->di_version == 3) {
320 to->di_changecount = cpu_to_be64(inode_peek_iversion(inode));
321 to->di_crtime.t_sec = cpu_to_be32(from->di_crtime.t_sec);
322 to->di_crtime.t_nsec = cpu_to_be32(from->di_crtime.t_nsec);
323 to->di_flags2 = cpu_to_be64(from->di_flags2);
324 to->di_cowextsize = cpu_to_be32(from->di_cowextsize);
325 to->di_ino = cpu_to_be64(ip->i_ino);
326 to->di_lsn = cpu_to_be64(lsn);
327 memset(to->di_pad2, 0, sizeof(to->di_pad2));
328 uuid_copy(&to->di_uuid, &ip->i_mount->m_sb.sb_meta_uuid);
329 to->di_flushiter = 0;
331 to->di_flushiter = cpu_to_be16(from->di_flushiter);
336 xfs_log_dinode_to_disk(
337 struct xfs_log_dinode *from,
338 struct xfs_dinode *to)
340 to->di_magic = cpu_to_be16(from->di_magic);
341 to->di_mode = cpu_to_be16(from->di_mode);
342 to->di_version = from->di_version;
343 to->di_format = from->di_format;
345 to->di_uid = cpu_to_be32(from->di_uid);
346 to->di_gid = cpu_to_be32(from->di_gid);
347 to->di_nlink = cpu_to_be32(from->di_nlink);
348 to->di_projid_lo = cpu_to_be16(from->di_projid_lo);
349 to->di_projid_hi = cpu_to_be16(from->di_projid_hi);
350 memcpy(to->di_pad, from->di_pad, sizeof(to->di_pad));
352 to->di_atime.t_sec = cpu_to_be32(from->di_atime.t_sec);
353 to->di_atime.t_nsec = cpu_to_be32(from->di_atime.t_nsec);
354 to->di_mtime.t_sec = cpu_to_be32(from->di_mtime.t_sec);
355 to->di_mtime.t_nsec = cpu_to_be32(from->di_mtime.t_nsec);
356 to->di_ctime.t_sec = cpu_to_be32(from->di_ctime.t_sec);
357 to->di_ctime.t_nsec = cpu_to_be32(from->di_ctime.t_nsec);
359 to->di_size = cpu_to_be64(from->di_size);
360 to->di_nblocks = cpu_to_be64(from->di_nblocks);
361 to->di_extsize = cpu_to_be32(from->di_extsize);
362 to->di_nextents = cpu_to_be32(from->di_nextents);
363 to->di_anextents = cpu_to_be16(from->di_anextents);
364 to->di_forkoff = from->di_forkoff;
365 to->di_aformat = from->di_aformat;
366 to->di_dmevmask = cpu_to_be32(from->di_dmevmask);
367 to->di_dmstate = cpu_to_be16(from->di_dmstate);
368 to->di_flags = cpu_to_be16(from->di_flags);
369 to->di_gen = cpu_to_be32(from->di_gen);
371 if (from->di_version == 3) {
372 to->di_changecount = cpu_to_be64(from->di_changecount);
373 to->di_crtime.t_sec = cpu_to_be32(from->di_crtime.t_sec);
374 to->di_crtime.t_nsec = cpu_to_be32(from->di_crtime.t_nsec);
375 to->di_flags2 = cpu_to_be64(from->di_flags2);
376 to->di_cowextsize = cpu_to_be32(from->di_cowextsize);
377 to->di_ino = cpu_to_be64(from->di_ino);
378 to->di_lsn = cpu_to_be64(from->di_lsn);
379 memcpy(to->di_pad2, from->di_pad2, sizeof(to->di_pad2));
380 uuid_copy(&to->di_uuid, &from->di_uuid);
381 to->di_flushiter = 0;
383 to->di_flushiter = cpu_to_be16(from->di_flushiter);
389 struct xfs_mount *mp,
391 struct xfs_dinode *dip)
397 if (dip->di_magic != cpu_to_be16(XFS_DINODE_MAGIC))
400 /* don't allow invalid i_size */
401 if (be64_to_cpu(dip->di_size) & (1ULL << 63))
404 mode = be16_to_cpu(dip->di_mode);
405 if (mode && xfs_mode_to_ftype(mode) == XFS_DIR3_FT_UNKNOWN)
408 /* No zero-length symlinks/dirs. */
409 if ((S_ISLNK(mode) || S_ISDIR(mode)) && dip->di_size == 0)
412 /* only version 3 or greater inodes are extensively verified here */
413 if (dip->di_version < 3)
416 if (!xfs_sb_version_hascrc(&mp->m_sb))
418 if (!xfs_verify_cksum((char *)dip, mp->m_sb.sb_inodesize,
421 if (be64_to_cpu(dip->di_ino) != ino)
423 if (!uuid_equal(&dip->di_uuid, &mp->m_sb.sb_meta_uuid))
426 flags = be16_to_cpu(dip->di_flags);
427 flags2 = be64_to_cpu(dip->di_flags2);
429 /* don't allow reflink/cowextsize if we don't have reflink */
430 if ((flags2 & (XFS_DIFLAG2_REFLINK | XFS_DIFLAG2_COWEXTSIZE)) &&
431 !xfs_sb_version_hasreflink(&mp->m_sb))
434 /* don't let reflink and realtime mix */
435 if ((flags2 & XFS_DIFLAG2_REFLINK) && (flags & XFS_DIFLAG_REALTIME))
438 /* don't let reflink and dax mix */
439 if ((flags2 & XFS_DIFLAG2_REFLINK) && (flags2 & XFS_DIFLAG2_DAX))
447 struct xfs_mount *mp,
448 struct xfs_dinode *dip)
452 if (dip->di_version < 3)
455 ASSERT(xfs_sb_version_hascrc(&mp->m_sb));
456 crc = xfs_start_cksum_update((char *)dip, mp->m_sb.sb_inodesize,
458 dip->di_crc = xfs_end_cksum(crc);
462 * Read the disk inode attributes into the in-core inode structure.
464 * For version 5 superblocks, if we are initialising a new inode and we are not
465 * utilising the XFS_MOUNT_IKEEP inode cluster mode, we can simple build the new
466 * inode core with a random generation number. If we are keeping inodes around,
467 * we need to read the inode cluster to get the existing generation number off
468 * disk. Further, if we are using version 4 superblocks (i.e. v1/v2 inode
469 * format) then log recovery is dependent on the di_flushiter field being
470 * initialised from the current on-disk value and hence we must also read the
485 * Fill in the location information in the in-core inode.
487 error = xfs_imap(mp, tp, ip->i_ino, &ip->i_imap, iget_flags);
491 /* shortcut IO on inode allocation if possible */
492 if ((iget_flags & XFS_IGET_CREATE) &&
493 xfs_sb_version_hascrc(&mp->m_sb) &&
494 !(mp->m_flags & XFS_MOUNT_IKEEP)) {
495 /* initialise the on-disk inode core */
496 memset(&ip->i_d, 0, sizeof(ip->i_d));
497 VFS_I(ip)->i_generation = prandom_u32();
498 if (xfs_sb_version_hascrc(&mp->m_sb))
499 ip->i_d.di_version = 3;
501 ip->i_d.di_version = 2;
506 * Get pointers to the on-disk inode and the buffer containing it.
508 error = xfs_imap_to_bp(mp, tp, &ip->i_imap, &dip, &bp, 0, iget_flags);
512 /* even unallocated inodes are verified */
513 if (!xfs_dinode_verify(mp, ip->i_ino, dip)) {
514 xfs_alert(mp, "%s: validation failed for inode %lld",
515 __func__, ip->i_ino);
517 XFS_CORRUPTION_ERROR(__func__, XFS_ERRLEVEL_LOW, mp, dip);
518 error = -EFSCORRUPTED;
523 * If the on-disk inode is already linked to a directory
524 * entry, copy all of the inode into the in-core inode.
525 * xfs_iformat_fork() handles copying in the inode format
526 * specific information.
527 * Otherwise, just get the truly permanent information.
530 xfs_inode_from_disk(ip, dip);
531 error = xfs_iformat_fork(ip, dip);
534 xfs_alert(mp, "%s: xfs_iformat() returned error %d",
541 * Partial initialisation of the in-core inode. Just the bits
542 * that xfs_ialloc won't overwrite or relies on being correct.
544 ip->i_d.di_version = dip->di_version;
545 VFS_I(ip)->i_generation = be32_to_cpu(dip->di_gen);
546 ip->i_d.di_flushiter = be16_to_cpu(dip->di_flushiter);
549 * Make sure to pull in the mode here as well in
550 * case the inode is released without being used.
551 * This ensures that xfs_inactive() will see that
552 * the inode is already free and not try to mess
553 * with the uninitialized part of it.
555 VFS_I(ip)->i_mode = 0;
558 ASSERT(ip->i_d.di_version >= 2);
559 ip->i_delayed_blks = 0;
562 * Mark the buffer containing the inode as something to keep
563 * around for a while. This helps to keep recently accessed
564 * meta-data in-core longer.
566 xfs_buf_set_ref(bp, XFS_INO_REF);
569 * Use xfs_trans_brelse() to release the buffer containing the on-disk
570 * inode, because it was acquired with xfs_trans_read_buf() in
571 * xfs_imap_to_bp() above. If tp is NULL, this is just a normal
572 * brelse(). If we're within a transaction, then xfs_trans_brelse()
573 * will only release the buffer if it is not dirty within the
574 * transaction. It will be OK to release the buffer in this case,
575 * because inodes on disk are never destroyed and we will be locking the
576 * new in-core inode before putting it in the cache where other
577 * processes can find it. Thus we don't have to worry about the inode
578 * being changed just because we released the buffer.
581 xfs_trans_brelse(tp, bp);