]>
Commit | Line | Data |
---|---|---|
2bd0ea18 | 1 | /* |
5e656dbb | 2 | * Copyright (c) 2000-2006 Silicon Graphics, Inc. |
da23017d | 3 | * All Rights Reserved. |
5000d01d | 4 | * |
da23017d NS |
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 | |
2bd0ea18 | 7 | * published by the Free Software Foundation. |
5000d01d | 8 | * |
da23017d NS |
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. | |
5000d01d | 13 | * |
da23017d NS |
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 | |
2bd0ea18 NS |
17 | */ |
18 | ||
19 | #include <xfs.h> | |
20 | ||
5e656dbb BN |
21 | /* |
22 | * Check that none of the inode's in the buffer have a next | |
23 | * unlinked field of 0. | |
24 | */ | |
25 | #if defined(DEBUG) | |
2bd0ea18 NS |
26 | void |
27 | xfs_inobp_check( | |
28 | xfs_mount_t *mp, | |
29 | xfs_buf_t *bp) | |
30 | { | |
31 | int i; | |
32 | int j; | |
33 | xfs_dinode_t *dip; | |
34 | ||
35 | j = mp->m_inode_cluster_size >> mp->m_sb.sb_inodelog; | |
36 | ||
37 | for (i = 0; i < j; i++) { | |
38 | dip = (xfs_dinode_t *)xfs_buf_offset(bp, | |
39 | i * mp->m_sb.sb_inodesize); | |
46eca962 | 40 | if (!dip->di_next_unlinked) { |
a2ceac1f | 41 | xfs_alert(mp, |
ea6a00d4 DC |
42 | "Detected bogus zero next_unlinked field in inode %d buffer 0x%llx.", |
43 | i, (long long)bp->b_bn); | |
2bd0ea18 NS |
44 | } |
45 | } | |
46 | } | |
47 | #endif | |
48 | ||
f85fc622 DC |
49 | /* |
50 | * If we are doing readahead on an inode buffer, we might be in log recovery | |
51 | * reading an inode allocation buffer that hasn't yet been replayed, and hence | |
52 | * has not had the inode cores stamped into it. Hence for readahead, the buffer | |
53 | * may be potentially invalid. | |
54 | * | |
55 | * If the readahead buffer is invalid, we don't want to mark it with an error, | |
56 | * but we do want to clear the DONE status of the buffer so that a followup read | |
57 | * will re-read it from disk. This will ensure that we don't get an unnecessary | |
58 | * warnings during log recovery and we don't get unnecessary panics on debug | |
59 | * kernels. | |
60 | */ | |
a2ceac1f DC |
61 | static void |
62 | xfs_inode_buf_verify( | |
f85fc622 DC |
63 | struct xfs_buf *bp, |
64 | bool readahead) | |
5e656dbb | 65 | { |
a2ceac1f | 66 | struct xfs_mount *mp = bp->b_target->bt_mount; |
5e656dbb BN |
67 | int i; |
68 | int ni; | |
5e656dbb BN |
69 | |
70 | /* | |
71 | * Validate the magic number and version of every inode in the buffer | |
5e656dbb | 72 | */ |
a2ceac1f | 73 | ni = XFS_BB_TO_FSB(mp, bp->b_length) * mp->m_sb.sb_inopblock; |
5e656dbb BN |
74 | for (i = 0; i < ni; i++) { |
75 | int di_ok; | |
76 | xfs_dinode_t *dip; | |
77 | ||
a2ceac1f | 78 | dip = (struct xfs_dinode *)xfs_buf_offset(bp, |
5e656dbb | 79 | (i << mp->m_sb.sb_inodelog)); |
a2ceac1f | 80 | di_ok = dip->di_magic == cpu_to_be16(XFS_DINODE_MAGIC) && |
56b2de80 | 81 | XFS_DINODE_GOOD_VERSION(dip->di_version); |
5e656dbb BN |
82 | if (unlikely(XFS_TEST_ERROR(!di_ok, mp, |
83 | XFS_ERRTAG_ITOBP_INOTOBP, | |
84 | XFS_RANDOM_ITOBP_INOTOBP))) { | |
f85fc622 DC |
85 | if (readahead) { |
86 | bp->b_flags &= ~XBF_DONE; | |
87 | return; | |
88 | } | |
89 | ||
a2ceac1f DC |
90 | xfs_buf_ioerror(bp, EFSCORRUPTED); |
91 | XFS_CORRUPTION_ERROR(__func__, XFS_ERRLEVEL_HIGH, | |
92 | mp, dip); | |
5e656dbb | 93 | #ifdef DEBUG |
ea6a00d4 | 94 | xfs_alert(mp, |
a2ceac1f DC |
95 | "bad inode magic/vsn daddr %lld #%d (magic=%x)", |
96 | (unsigned long long)bp->b_bn, i, | |
56b2de80 | 97 | be16_to_cpu(dip->di_magic)); |
5e656dbb | 98 | #endif |
5e656dbb BN |
99 | } |
100 | } | |
5e656dbb | 101 | xfs_inobp_check(mp, bp); |
a2ceac1f | 102 | } |
5e656dbb | 103 | |
e6d77a21 | 104 | |
a2ceac1f DC |
105 | static void |
106 | xfs_inode_buf_read_verify( | |
107 | struct xfs_buf *bp) | |
108 | { | |
f85fc622 DC |
109 | xfs_inode_buf_verify(bp, false); |
110 | } | |
111 | ||
112 | static void | |
113 | xfs_inode_buf_readahead_verify( | |
114 | struct xfs_buf *bp) | |
115 | { | |
116 | xfs_inode_buf_verify(bp, true); | |
5e656dbb | 117 | } |
2bd0ea18 | 118 | |
a2ceac1f DC |
119 | static void |
120 | xfs_inode_buf_write_verify( | |
121 | struct xfs_buf *bp) | |
56b2de80 | 122 | { |
f85fc622 | 123 | xfs_inode_buf_verify(bp, false); |
56b2de80 DC |
124 | } |
125 | ||
a2ceac1f DC |
126 | const struct xfs_buf_ops xfs_inode_buf_ops = { |
127 | .verify_read = xfs_inode_buf_read_verify, | |
128 | .verify_write = xfs_inode_buf_write_verify, | |
129 | }; | |
130 | ||
f85fc622 DC |
131 | const struct xfs_buf_ops xfs_inode_buf_ra_ops = { |
132 | .verify_read = xfs_inode_buf_readahead_verify, | |
133 | .verify_write = xfs_inode_buf_write_verify, | |
134 | }; | |
135 | ||
136 | ||
2bd0ea18 | 137 | /* |
a2ceac1f DC |
138 | * This routine is called to map an inode to the buffer containing the on-disk |
139 | * version of the inode. It returns a pointer to the buffer containing the | |
140 | * on-disk inode in the bpp parameter, and in the dipp parameter it returns a | |
141 | * pointer to the on-disk inode within that buffer. | |
2bd0ea18 | 142 | * |
a2ceac1f DC |
143 | * If a non-zero error is returned, then the contents of bpp and dipp are |
144 | * undefined. | |
2bd0ea18 NS |
145 | */ |
146 | int | |
a2ceac1f DC |
147 | xfs_imap_to_bp( |
148 | struct xfs_mount *mp, | |
149 | struct xfs_trans *tp, | |
150 | struct xfs_imap *imap, | |
151 | struct xfs_dinode **dipp, | |
152 | struct xfs_buf **bpp, | |
153 | uint buf_flags, | |
154 | uint iget_flags) | |
2bd0ea18 | 155 | { |
a2ceac1f DC |
156 | struct xfs_buf *bp; |
157 | int error; | |
2bd0ea18 | 158 | |
a2ceac1f DC |
159 | buf_flags |= XBF_UNMAPPED; |
160 | error = xfs_trans_read_buf(mp, tp, mp->m_ddev_targp, imap->im_blkno, | |
161 | (int)imap->im_len, buf_flags, &bp, | |
162 | &xfs_inode_buf_ops); | |
163 | if (error) { | |
164 | if (error == EAGAIN) { | |
165 | ASSERT(buf_flags & XBF_TRYLOCK); | |
166 | return error; | |
167 | } | |
2bd0ea18 | 168 | |
a2ceac1f DC |
169 | if (error == EFSCORRUPTED && |
170 | (iget_flags & XFS_IGET_UNTRUSTED)) | |
171 | return XFS_ERROR(EINVAL); | |
2bd0ea18 | 172 | |
a2ceac1f DC |
173 | xfs_warn(mp, "%s: xfs_trans_read_buf() returned error %d.", |
174 | __func__, error); | |
175 | return error; | |
2bd0ea18 | 176 | } |
2bd0ea18 | 177 | |
2bd0ea18 | 178 | *bpp = bp; |
a2ceac1f | 179 | *dipp = (struct xfs_dinode *)xfs_buf_offset(bp, imap->im_boffset); |
2bd0ea18 NS |
180 | return 0; |
181 | } | |
182 | ||
5000d01d | 183 | void |
5e656dbb BN |
184 | xfs_dinode_from_disk( |
185 | xfs_icdinode_t *to, | |
56b2de80 | 186 | xfs_dinode_t *from) |
2bd0ea18 | 187 | { |
5e656dbb BN |
188 | to->di_magic = be16_to_cpu(from->di_magic); |
189 | to->di_mode = be16_to_cpu(from->di_mode); | |
190 | to->di_version = from ->di_version; | |
191 | to->di_format = from->di_format; | |
192 | to->di_onlink = be16_to_cpu(from->di_onlink); | |
193 | to->di_uid = be32_to_cpu(from->di_uid); | |
194 | to->di_gid = be32_to_cpu(from->di_gid); | |
195 | to->di_nlink = be32_to_cpu(from->di_nlink); | |
22bc10ed AM |
196 | to->di_projid_lo = be16_to_cpu(from->di_projid_lo); |
197 | to->di_projid_hi = be16_to_cpu(from->di_projid_hi); | |
5e656dbb BN |
198 | memcpy(to->di_pad, from->di_pad, sizeof(to->di_pad)); |
199 | to->di_flushiter = be16_to_cpu(from->di_flushiter); | |
200 | to->di_atime.t_sec = be32_to_cpu(from->di_atime.t_sec); | |
201 | to->di_atime.t_nsec = be32_to_cpu(from->di_atime.t_nsec); | |
202 | to->di_mtime.t_sec = be32_to_cpu(from->di_mtime.t_sec); | |
203 | to->di_mtime.t_nsec = be32_to_cpu(from->di_mtime.t_nsec); | |
204 | to->di_ctime.t_sec = be32_to_cpu(from->di_ctime.t_sec); | |
205 | to->di_ctime.t_nsec = be32_to_cpu(from->di_ctime.t_nsec); | |
206 | to->di_size = be64_to_cpu(from->di_size); | |
207 | to->di_nblocks = be64_to_cpu(from->di_nblocks); | |
208 | to->di_extsize = be32_to_cpu(from->di_extsize); | |
209 | to->di_nextents = be32_to_cpu(from->di_nextents); | |
210 | to->di_anextents = be16_to_cpu(from->di_anextents); | |
211 | to->di_forkoff = from->di_forkoff; | |
212 | to->di_aformat = from->di_aformat; | |
213 | to->di_dmevmask = be32_to_cpu(from->di_dmevmask); | |
214 | to->di_dmstate = be16_to_cpu(from->di_dmstate); | |
215 | to->di_flags = be16_to_cpu(from->di_flags); | |
216 | to->di_gen = be32_to_cpu(from->di_gen); | |
41ce5f36 DC |
217 | |
218 | if (to->di_version == 3) { | |
219 | to->di_changecount = be64_to_cpu(from->di_changecount); | |
220 | to->di_crtime.t_sec = be32_to_cpu(from->di_crtime.t_sec); | |
221 | to->di_crtime.t_nsec = be32_to_cpu(from->di_crtime.t_nsec); | |
222 | to->di_flags2 = be64_to_cpu(from->di_flags2); | |
223 | to->di_ino = be64_to_cpu(from->di_ino); | |
224 | to->di_lsn = be64_to_cpu(from->di_lsn); | |
225 | memcpy(to->di_pad2, from->di_pad2, sizeof(to->di_pad2)); | |
f85fc622 | 226 | uuid_copy(&to->di_uuid, &from->di_uuid); |
41ce5f36 | 227 | } |
5e656dbb BN |
228 | } |
229 | ||
230 | void | |
231 | xfs_dinode_to_disk( | |
56b2de80 | 232 | xfs_dinode_t *to, |
5e656dbb BN |
233 | xfs_icdinode_t *from) |
234 | { | |
235 | to->di_magic = cpu_to_be16(from->di_magic); | |
236 | to->di_mode = cpu_to_be16(from->di_mode); | |
237 | to->di_version = from ->di_version; | |
238 | to->di_format = from->di_format; | |
239 | to->di_onlink = cpu_to_be16(from->di_onlink); | |
240 | to->di_uid = cpu_to_be32(from->di_uid); | |
241 | to->di_gid = cpu_to_be32(from->di_gid); | |
242 | to->di_nlink = cpu_to_be32(from->di_nlink); | |
22bc10ed AM |
243 | to->di_projid_lo = cpu_to_be16(from->di_projid_lo); |
244 | to->di_projid_hi = cpu_to_be16(from->di_projid_hi); | |
5e656dbb | 245 | memcpy(to->di_pad, from->di_pad, sizeof(to->di_pad)); |
5e656dbb BN |
246 | to->di_atime.t_sec = cpu_to_be32(from->di_atime.t_sec); |
247 | to->di_atime.t_nsec = cpu_to_be32(from->di_atime.t_nsec); | |
248 | to->di_mtime.t_sec = cpu_to_be32(from->di_mtime.t_sec); | |
249 | to->di_mtime.t_nsec = cpu_to_be32(from->di_mtime.t_nsec); | |
250 | to->di_ctime.t_sec = cpu_to_be32(from->di_ctime.t_sec); | |
251 | to->di_ctime.t_nsec = cpu_to_be32(from->di_ctime.t_nsec); | |
252 | to->di_size = cpu_to_be64(from->di_size); | |
253 | to->di_nblocks = cpu_to_be64(from->di_nblocks); | |
254 | to->di_extsize = cpu_to_be32(from->di_extsize); | |
255 | to->di_nextents = cpu_to_be32(from->di_nextents); | |
256 | to->di_anextents = cpu_to_be16(from->di_anextents); | |
257 | to->di_forkoff = from->di_forkoff; | |
258 | to->di_aformat = from->di_aformat; | |
259 | to->di_dmevmask = cpu_to_be32(from->di_dmevmask); | |
260 | to->di_dmstate = cpu_to_be16(from->di_dmstate); | |
261 | to->di_flags = cpu_to_be16(from->di_flags); | |
262 | to->di_gen = cpu_to_be32(from->di_gen); | |
41ce5f36 DC |
263 | |
264 | if (from->di_version == 3) { | |
265 | to->di_changecount = cpu_to_be64(from->di_changecount); | |
266 | to->di_crtime.t_sec = cpu_to_be32(from->di_crtime.t_sec); | |
267 | to->di_crtime.t_nsec = cpu_to_be32(from->di_crtime.t_nsec); | |
268 | to->di_flags2 = cpu_to_be64(from->di_flags2); | |
269 | to->di_ino = cpu_to_be64(from->di_ino); | |
270 | to->di_lsn = cpu_to_be64(from->di_lsn); | |
271 | memcpy(to->di_pad2, from->di_pad2, sizeof(to->di_pad2)); | |
f85fc622 | 272 | uuid_copy(&to->di_uuid, &from->di_uuid); |
fde71651 DC |
273 | to->di_flushiter = 0; |
274 | } else { | |
275 | to->di_flushiter = cpu_to_be16(from->di_flushiter); | |
41ce5f36 DC |
276 | } |
277 | } | |
278 | ||
a73b88f2 | 279 | bool |
41ce5f36 DC |
280 | xfs_dinode_verify( |
281 | struct xfs_mount *mp, | |
a73b88f2 | 282 | xfs_ino_t ino, |
41ce5f36 DC |
283 | struct xfs_dinode *dip) |
284 | { | |
285 | if (dip->di_magic != cpu_to_be16(XFS_DINODE_MAGIC)) | |
286 | return false; | |
287 | ||
288 | /* only version 3 or greater inodes are extensively verified here */ | |
289 | if (dip->di_version < 3) | |
290 | return true; | |
291 | ||
292 | if (!xfs_sb_version_hascrc(&mp->m_sb)) | |
293 | return false; | |
294 | if (!xfs_verify_cksum((char *)dip, mp->m_sb.sb_inodesize, | |
295 | offsetof(struct xfs_dinode, di_crc))) | |
296 | return false; | |
a73b88f2 | 297 | if (be64_to_cpu(dip->di_ino) != ino) |
41ce5f36 DC |
298 | return false; |
299 | if (!uuid_equal(&dip->di_uuid, &mp->m_sb.sb_uuid)) | |
300 | return false; | |
301 | return true; | |
302 | } | |
e6d77a21 | 303 | |
41ce5f36 DC |
304 | void |
305 | xfs_dinode_calc_crc( | |
306 | struct xfs_mount *mp, | |
307 | struct xfs_dinode *dip) | |
308 | { | |
309 | __uint32_t crc; | |
310 | ||
311 | if (dip->di_version < 3) | |
312 | return; | |
313 | ||
314 | ASSERT(xfs_sb_version_hascrc(&mp->m_sb)); | |
315 | crc = xfs_start_cksum((char *)dip, mp->m_sb.sb_inodesize, | |
316 | offsetof(struct xfs_dinode, di_crc)); | |
317 | dip->di_crc = xfs_end_cksum(crc); | |
2bd0ea18 NS |
318 | } |
319 | ||
a2ceac1f DC |
320 | /* |
321 | * Read the disk inode attributes into the in-core inode structure. | |
322 | */ | |
323 | int | |
324 | xfs_iread( | |
325 | xfs_mount_t *mp, | |
326 | xfs_trans_t *tp, | |
327 | xfs_inode_t *ip, | |
328 | uint iget_flags) | |
329 | { | |
330 | xfs_buf_t *bp; | |
331 | xfs_dinode_t *dip; | |
332 | int error; | |
333 | ||
334 | /* | |
335 | * Fill in the location information in the in-core inode. | |
336 | */ | |
337 | error = xfs_imap(mp, tp, ip->i_ino, &ip->i_imap, iget_flags); | |
338 | if (error) | |
339 | return error; | |
340 | ||
341 | /* | |
342 | * Get pointers to the on-disk inode and the buffer containing it. | |
343 | */ | |
344 | error = xfs_imap_to_bp(mp, tp, &ip->i_imap, &dip, &bp, 0, iget_flags); | |
345 | if (error) | |
346 | return error; | |
347 | ||
41ce5f36 | 348 | /* even unallocated inodes are verified */ |
a73b88f2 | 349 | if (!xfs_dinode_verify(mp, ip->i_ino, dip)) { |
41ce5f36 DC |
350 | xfs_alert(mp, "%s: validation failed for inode %lld failed", |
351 | __func__, ip->i_ino); | |
352 | ||
353 | XFS_CORRUPTION_ERROR(__func__, XFS_ERRLEVEL_LOW, mp, dip); | |
354 | error = XFS_ERROR(EFSCORRUPTED); | |
a2ceac1f DC |
355 | goto out_brelse; |
356 | } | |
357 | ||
358 | /* | |
359 | * If the on-disk inode is already linked to a directory | |
360 | * entry, copy all of the inode into the in-core inode. | |
5d90ab5a | 361 | * xfs_iformat_fork() handles copying in the inode format |
a2ceac1f DC |
362 | * specific information. |
363 | * Otherwise, just get the truly permanent information. | |
364 | */ | |
365 | if (dip->di_mode) { | |
366 | xfs_dinode_from_disk(&ip->i_d, dip); | |
5d90ab5a | 367 | error = xfs_iformat_fork(ip, dip); |
a2ceac1f DC |
368 | if (error) { |
369 | #ifdef DEBUG | |
370 | xfs_alert(mp, "%s: xfs_iformat() returned error %d", | |
371 | __func__, error); | |
372 | #endif /* DEBUG */ | |
373 | goto out_brelse; | |
374 | } | |
375 | } else { | |
41ce5f36 DC |
376 | /* |
377 | * Partial initialisation of the in-core inode. Just the bits | |
378 | * that xfs_ialloc won't overwrite or relies on being correct. | |
379 | */ | |
a2ceac1f DC |
380 | ip->i_d.di_magic = be16_to_cpu(dip->di_magic); |
381 | ip->i_d.di_version = dip->di_version; | |
382 | ip->i_d.di_gen = be32_to_cpu(dip->di_gen); | |
383 | ip->i_d.di_flushiter = be16_to_cpu(dip->di_flushiter); | |
41ce5f36 DC |
384 | |
385 | if (dip->di_version == 3) { | |
386 | ip->i_d.di_ino = be64_to_cpu(dip->di_ino); | |
387 | uuid_copy(&ip->i_d.di_uuid, &dip->di_uuid); | |
388 | } | |
389 | ||
a2ceac1f DC |
390 | /* |
391 | * Make sure to pull in the mode here as well in | |
392 | * case the inode is released without being used. | |
393 | * This ensures that xfs_inactive() will see that | |
394 | * the inode is already free and not try to mess | |
395 | * with the uninitialized part of it. | |
396 | */ | |
397 | ip->i_d.di_mode = 0; | |
398 | } | |
399 | ||
400 | /* | |
401 | * The inode format changed when we moved the link count and | |
402 | * made it 32 bits long. If this is an old format inode, | |
403 | * convert it in memory to look like a new one. If it gets | |
404 | * flushed to disk we will convert back before flushing or | |
405 | * logging it. We zero out the new projid field and the old link | |
406 | * count field. We'll handle clearing the pad field (the remains | |
407 | * of the old uuid field) when we actually convert the inode to | |
408 | * the new format. We don't change the version number so that we | |
409 | * can distinguish this from a real new format inode. | |
410 | */ | |
411 | if (ip->i_d.di_version == 1) { | |
412 | ip->i_d.di_nlink = ip->i_d.di_onlink; | |
413 | ip->i_d.di_onlink = 0; | |
414 | xfs_set_projid(&ip->i_d, 0); | |
415 | } | |
416 | ||
417 | ip->i_delayed_blks = 0; | |
418 | ||
419 | /* | |
420 | * Mark the buffer containing the inode as something to keep | |
421 | * around for a while. This helps to keep recently accessed | |
422 | * meta-data in-core longer. | |
423 | */ | |
424 | xfs_buf_set_ref(bp, XFS_INO_REF); | |
425 | ||
426 | /* | |
1d519883 DC |
427 | * Use xfs_trans_brelse() to release the buffer containing the on-disk |
428 | * inode, because it was acquired with xfs_trans_read_buf() in | |
429 | * xfs_imap_to_bp() above. If tp is NULL, this is just a normal | |
a2ceac1f DC |
430 | * brelse(). If we're within a transaction, then xfs_trans_brelse() |
431 | * will only release the buffer if it is not dirty within the | |
432 | * transaction. It will be OK to release the buffer in this case, | |
1d519883 DC |
433 | * because inodes on disk are never destroyed and we will be locking the |
434 | * new in-core inode before putting it in the cache where other | |
435 | * processes can find it. Thus we don't have to worry about the inode | |
436 | * being changed just because we released the buffer. | |
a2ceac1f DC |
437 | */ |
438 | out_brelse: | |
439 | xfs_trans_brelse(tp, bp); | |
440 | return error; | |
441 | } |