Commit | Line | Data |
---|---|---|
1fd7115e DC |
1 | /* |
2 | * Copyright (c) 2000-2006 Silicon Graphics, Inc. | |
3 | * All Rights Reserved. | |
4 | * | |
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. | |
8 | * | |
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. | |
13 | * | |
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 | |
17 | */ | |
18 | #include "xfs.h" | |
19 | #include "xfs_fs.h" | |
70a9883c | 20 | #include "xfs_shared.h" |
239880ef DC |
21 | #include "xfs_format.h" |
22 | #include "xfs_log_format.h" | |
23 | #include "xfs_trans_resv.h" | |
1fd7115e | 24 | #include "xfs_mount.h" |
1fd7115e DC |
25 | #include "xfs_inode.h" |
26 | #include "xfs_error.h" | |
27 | #include "xfs_cksum.h" | |
28 | #include "xfs_icache.h" | |
239880ef | 29 | #include "xfs_trans.h" |
a4fbe6ab | 30 | #include "xfs_ialloc.h" |
1fd7115e DC |
31 | |
32 | /* | |
33 | * Check that none of the inode's in the buffer have a next | |
34 | * unlinked field of 0. | |
35 | */ | |
36 | #if defined(DEBUG) | |
37 | void | |
38 | xfs_inobp_check( | |
39 | xfs_mount_t *mp, | |
40 | xfs_buf_t *bp) | |
41 | { | |
42 | int i; | |
43 | int j; | |
44 | xfs_dinode_t *dip; | |
45 | ||
46 | j = mp->m_inode_cluster_size >> mp->m_sb.sb_inodelog; | |
47 | ||
48 | for (i = 0; i < j; i++) { | |
49 | dip = (xfs_dinode_t *)xfs_buf_offset(bp, | |
50 | i * mp->m_sb.sb_inodesize); | |
51 | if (!dip->di_next_unlinked) { | |
52 | xfs_alert(mp, | |
74ffa796 DC |
53 | "Detected bogus zero next_unlinked field in inode %d buffer 0x%llx.", |
54 | i, (long long)bp->b_bn); | |
1fd7115e DC |
55 | } |
56 | } | |
57 | } | |
58 | #endif | |
59 | ||
d8914002 DC |
60 | /* |
61 | * If we are doing readahead on an inode buffer, we might be in log recovery | |
62 | * reading an inode allocation buffer that hasn't yet been replayed, and hence | |
63 | * has not had the inode cores stamped into it. Hence for readahead, the buffer | |
64 | * may be potentially invalid. | |
65 | * | |
66 | * If the readahead buffer is invalid, we don't want to mark it with an error, | |
67 | * but we do want to clear the DONE status of the buffer so that a followup read | |
68 | * will re-read it from disk. This will ensure that we don't get an unnecessary | |
69 | * warnings during log recovery and we don't get unnecssary panics on debug | |
70 | * kernels. | |
71 | */ | |
1fd7115e DC |
72 | static void |
73 | xfs_inode_buf_verify( | |
d8914002 DC |
74 | struct xfs_buf *bp, |
75 | bool readahead) | |
1fd7115e DC |
76 | { |
77 | struct xfs_mount *mp = bp->b_target->bt_mount; | |
78 | int i; | |
79 | int ni; | |
80 | ||
81 | /* | |
82 | * Validate the magic number and version of every inode in the buffer | |
83 | */ | |
84 | ni = XFS_BB_TO_FSB(mp, bp->b_length) * mp->m_sb.sb_inopblock; | |
85 | for (i = 0; i < ni; i++) { | |
86 | int di_ok; | |
87 | xfs_dinode_t *dip; | |
88 | ||
89 | dip = (struct xfs_dinode *)xfs_buf_offset(bp, | |
90 | (i << mp->m_sb.sb_inodelog)); | |
91 | di_ok = dip->di_magic == cpu_to_be16(XFS_DINODE_MAGIC) && | |
92 | XFS_DINODE_GOOD_VERSION(dip->di_version); | |
93 | if (unlikely(XFS_TEST_ERROR(!di_ok, mp, | |
94 | XFS_ERRTAG_ITOBP_INOTOBP, | |
95 | XFS_RANDOM_ITOBP_INOTOBP))) { | |
d8914002 DC |
96 | if (readahead) { |
97 | bp->b_flags &= ~XBF_DONE; | |
98 | return; | |
99 | } | |
100 | ||
2451337d | 101 | xfs_buf_ioerror(bp, -EFSCORRUPTED); |
ce5028cf | 102 | xfs_verifier_error(bp); |
1fd7115e | 103 | #ifdef DEBUG |
74ffa796 | 104 | xfs_alert(mp, |
1fd7115e DC |
105 | "bad inode magic/vsn daddr %lld #%d (magic=%x)", |
106 | (unsigned long long)bp->b_bn, i, | |
107 | be16_to_cpu(dip->di_magic)); | |
1fd7115e DC |
108 | #endif |
109 | } | |
110 | } | |
111 | xfs_inobp_check(mp, bp); | |
112 | } | |
113 | ||
114 | ||
115 | static void | |
116 | xfs_inode_buf_read_verify( | |
117 | struct xfs_buf *bp) | |
118 | { | |
d8914002 DC |
119 | xfs_inode_buf_verify(bp, false); |
120 | } | |
121 | ||
122 | static void | |
123 | xfs_inode_buf_readahead_verify( | |
124 | struct xfs_buf *bp) | |
125 | { | |
126 | xfs_inode_buf_verify(bp, true); | |
1fd7115e DC |
127 | } |
128 | ||
129 | static void | |
130 | xfs_inode_buf_write_verify( | |
131 | struct xfs_buf *bp) | |
132 | { | |
d8914002 | 133 | xfs_inode_buf_verify(bp, false); |
1fd7115e DC |
134 | } |
135 | ||
136 | const struct xfs_buf_ops xfs_inode_buf_ops = { | |
137 | .verify_read = xfs_inode_buf_read_verify, | |
138 | .verify_write = xfs_inode_buf_write_verify, | |
139 | }; | |
140 | ||
d8914002 DC |
141 | const struct xfs_buf_ops xfs_inode_buf_ra_ops = { |
142 | .verify_read = xfs_inode_buf_readahead_verify, | |
143 | .verify_write = xfs_inode_buf_write_verify, | |
144 | }; | |
145 | ||
1fd7115e DC |
146 | |
147 | /* | |
148 | * This routine is called to map an inode to the buffer containing the on-disk | |
149 | * version of the inode. It returns a pointer to the buffer containing the | |
150 | * on-disk inode in the bpp parameter, and in the dipp parameter it returns a | |
151 | * pointer to the on-disk inode within that buffer. | |
152 | * | |
153 | * If a non-zero error is returned, then the contents of bpp and dipp are | |
154 | * undefined. | |
155 | */ | |
156 | int | |
157 | xfs_imap_to_bp( | |
158 | struct xfs_mount *mp, | |
159 | struct xfs_trans *tp, | |
160 | struct xfs_imap *imap, | |
161 | struct xfs_dinode **dipp, | |
162 | struct xfs_buf **bpp, | |
163 | uint buf_flags, | |
164 | uint iget_flags) | |
165 | { | |
166 | struct xfs_buf *bp; | |
167 | int error; | |
168 | ||
169 | buf_flags |= XBF_UNMAPPED; | |
170 | error = xfs_trans_read_buf(mp, tp, mp->m_ddev_targp, imap->im_blkno, | |
171 | (int)imap->im_len, buf_flags, &bp, | |
172 | &xfs_inode_buf_ops); | |
173 | if (error) { | |
2451337d | 174 | if (error == -EAGAIN) { |
1fd7115e DC |
175 | ASSERT(buf_flags & XBF_TRYLOCK); |
176 | return error; | |
177 | } | |
178 | ||
2451337d | 179 | if (error == -EFSCORRUPTED && |
1fd7115e | 180 | (iget_flags & XFS_IGET_UNTRUSTED)) |
2451337d | 181 | return -EINVAL; |
1fd7115e DC |
182 | |
183 | xfs_warn(mp, "%s: xfs_trans_read_buf() returned error %d.", | |
184 | __func__, error); | |
185 | return error; | |
186 | } | |
187 | ||
188 | *bpp = bp; | |
189 | *dipp = (struct xfs_dinode *)xfs_buf_offset(bp, imap->im_boffset); | |
190 | return 0; | |
191 | } | |
192 | ||
638f4416 | 193 | void |
1fd7115e DC |
194 | xfs_dinode_from_disk( |
195 | xfs_icdinode_t *to, | |
196 | xfs_dinode_t *from) | |
197 | { | |
198 | to->di_magic = be16_to_cpu(from->di_magic); | |
199 | to->di_mode = be16_to_cpu(from->di_mode); | |
200 | to->di_version = from ->di_version; | |
201 | to->di_format = from->di_format; | |
202 | to->di_onlink = be16_to_cpu(from->di_onlink); | |
203 | to->di_uid = be32_to_cpu(from->di_uid); | |
204 | to->di_gid = be32_to_cpu(from->di_gid); | |
205 | to->di_nlink = be32_to_cpu(from->di_nlink); | |
206 | to->di_projid_lo = be16_to_cpu(from->di_projid_lo); | |
207 | to->di_projid_hi = be16_to_cpu(from->di_projid_hi); | |
208 | memcpy(to->di_pad, from->di_pad, sizeof(to->di_pad)); | |
209 | to->di_flushiter = be16_to_cpu(from->di_flushiter); | |
210 | to->di_atime.t_sec = be32_to_cpu(from->di_atime.t_sec); | |
211 | to->di_atime.t_nsec = be32_to_cpu(from->di_atime.t_nsec); | |
212 | to->di_mtime.t_sec = be32_to_cpu(from->di_mtime.t_sec); | |
213 | to->di_mtime.t_nsec = be32_to_cpu(from->di_mtime.t_nsec); | |
214 | to->di_ctime.t_sec = be32_to_cpu(from->di_ctime.t_sec); | |
215 | to->di_ctime.t_nsec = be32_to_cpu(from->di_ctime.t_nsec); | |
216 | to->di_size = be64_to_cpu(from->di_size); | |
217 | to->di_nblocks = be64_to_cpu(from->di_nblocks); | |
218 | to->di_extsize = be32_to_cpu(from->di_extsize); | |
219 | to->di_nextents = be32_to_cpu(from->di_nextents); | |
220 | to->di_anextents = be16_to_cpu(from->di_anextents); | |
221 | to->di_forkoff = from->di_forkoff; | |
222 | to->di_aformat = from->di_aformat; | |
223 | to->di_dmevmask = be32_to_cpu(from->di_dmevmask); | |
224 | to->di_dmstate = be16_to_cpu(from->di_dmstate); | |
225 | to->di_flags = be16_to_cpu(from->di_flags); | |
226 | to->di_gen = be32_to_cpu(from->di_gen); | |
227 | ||
228 | if (to->di_version == 3) { | |
229 | to->di_changecount = be64_to_cpu(from->di_changecount); | |
230 | to->di_crtime.t_sec = be32_to_cpu(from->di_crtime.t_sec); | |
231 | to->di_crtime.t_nsec = be32_to_cpu(from->di_crtime.t_nsec); | |
232 | to->di_flags2 = be64_to_cpu(from->di_flags2); | |
233 | to->di_ino = be64_to_cpu(from->di_ino); | |
234 | to->di_lsn = be64_to_cpu(from->di_lsn); | |
235 | memcpy(to->di_pad2, from->di_pad2, sizeof(to->di_pad2)); | |
236 | uuid_copy(&to->di_uuid, &from->di_uuid); | |
237 | } | |
238 | } | |
239 | ||
240 | void | |
241 | xfs_dinode_to_disk( | |
242 | xfs_dinode_t *to, | |
243 | xfs_icdinode_t *from) | |
244 | { | |
245 | to->di_magic = cpu_to_be16(from->di_magic); | |
246 | to->di_mode = cpu_to_be16(from->di_mode); | |
247 | to->di_version = from ->di_version; | |
248 | to->di_format = from->di_format; | |
249 | to->di_onlink = cpu_to_be16(from->di_onlink); | |
250 | to->di_uid = cpu_to_be32(from->di_uid); | |
251 | to->di_gid = cpu_to_be32(from->di_gid); | |
252 | to->di_nlink = cpu_to_be32(from->di_nlink); | |
253 | to->di_projid_lo = cpu_to_be16(from->di_projid_lo); | |
254 | to->di_projid_hi = cpu_to_be16(from->di_projid_hi); | |
255 | memcpy(to->di_pad, from->di_pad, sizeof(to->di_pad)); | |
256 | to->di_atime.t_sec = cpu_to_be32(from->di_atime.t_sec); | |
257 | to->di_atime.t_nsec = cpu_to_be32(from->di_atime.t_nsec); | |
258 | to->di_mtime.t_sec = cpu_to_be32(from->di_mtime.t_sec); | |
259 | to->di_mtime.t_nsec = cpu_to_be32(from->di_mtime.t_nsec); | |
260 | to->di_ctime.t_sec = cpu_to_be32(from->di_ctime.t_sec); | |
261 | to->di_ctime.t_nsec = cpu_to_be32(from->di_ctime.t_nsec); | |
262 | to->di_size = cpu_to_be64(from->di_size); | |
263 | to->di_nblocks = cpu_to_be64(from->di_nblocks); | |
264 | to->di_extsize = cpu_to_be32(from->di_extsize); | |
265 | to->di_nextents = cpu_to_be32(from->di_nextents); | |
266 | to->di_anextents = cpu_to_be16(from->di_anextents); | |
267 | to->di_forkoff = from->di_forkoff; | |
268 | to->di_aformat = from->di_aformat; | |
269 | to->di_dmevmask = cpu_to_be32(from->di_dmevmask); | |
270 | to->di_dmstate = cpu_to_be16(from->di_dmstate); | |
271 | to->di_flags = cpu_to_be16(from->di_flags); | |
272 | to->di_gen = cpu_to_be32(from->di_gen); | |
273 | ||
274 | if (from->di_version == 3) { | |
275 | to->di_changecount = cpu_to_be64(from->di_changecount); | |
276 | to->di_crtime.t_sec = cpu_to_be32(from->di_crtime.t_sec); | |
277 | to->di_crtime.t_nsec = cpu_to_be32(from->di_crtime.t_nsec); | |
278 | to->di_flags2 = cpu_to_be64(from->di_flags2); | |
279 | to->di_ino = cpu_to_be64(from->di_ino); | |
280 | to->di_lsn = cpu_to_be64(from->di_lsn); | |
281 | memcpy(to->di_pad2, from->di_pad2, sizeof(to->di_pad2)); | |
282 | uuid_copy(&to->di_uuid, &from->di_uuid); | |
283 | to->di_flushiter = 0; | |
284 | } else { | |
285 | to->di_flushiter = cpu_to_be16(from->di_flushiter); | |
286 | } | |
287 | } | |
288 | ||
289 | static bool | |
290 | xfs_dinode_verify( | |
291 | struct xfs_mount *mp, | |
292 | struct xfs_inode *ip, | |
293 | struct xfs_dinode *dip) | |
294 | { | |
295 | if (dip->di_magic != cpu_to_be16(XFS_DINODE_MAGIC)) | |
296 | return false; | |
297 | ||
298 | /* only version 3 or greater inodes are extensively verified here */ | |
299 | if (dip->di_version < 3) | |
300 | return true; | |
301 | ||
302 | if (!xfs_sb_version_hascrc(&mp->m_sb)) | |
303 | return false; | |
304 | if (!xfs_verify_cksum((char *)dip, mp->m_sb.sb_inodesize, | |
533b81c8 | 305 | XFS_DINODE_CRC_OFF)) |
1fd7115e DC |
306 | return false; |
307 | if (be64_to_cpu(dip->di_ino) != ip->i_ino) | |
308 | return false; | |
309 | if (!uuid_equal(&dip->di_uuid, &mp->m_sb.sb_uuid)) | |
310 | return false; | |
311 | return true; | |
312 | } | |
313 | ||
314 | void | |
315 | xfs_dinode_calc_crc( | |
316 | struct xfs_mount *mp, | |
317 | struct xfs_dinode *dip) | |
318 | { | |
319 | __uint32_t crc; | |
320 | ||
321 | if (dip->di_version < 3) | |
322 | return; | |
323 | ||
324 | ASSERT(xfs_sb_version_hascrc(&mp->m_sb)); | |
325 | crc = xfs_start_cksum((char *)dip, mp->m_sb.sb_inodesize, | |
533b81c8 | 326 | XFS_DINODE_CRC_OFF); |
1fd7115e DC |
327 | dip->di_crc = xfs_end_cksum(crc); |
328 | } | |
329 | ||
330 | /* | |
331 | * Read the disk inode attributes into the in-core inode structure. | |
332 | * | |
333 | * For version 5 superblocks, if we are initialising a new inode and we are not | |
334 | * utilising the XFS_MOUNT_IKEEP inode cluster mode, we can simple build the new | |
335 | * inode core with a random generation number. If we are keeping inodes around, | |
336 | * we need to read the inode cluster to get the existing generation number off | |
337 | * disk. Further, if we are using version 4 superblocks (i.e. v1/v2 inode | |
338 | * format) then log recovery is dependent on the di_flushiter field being | |
339 | * initialised from the current on-disk value and hence we must also read the | |
340 | * inode off disk. | |
341 | */ | |
342 | int | |
343 | xfs_iread( | |
344 | xfs_mount_t *mp, | |
345 | xfs_trans_t *tp, | |
346 | xfs_inode_t *ip, | |
347 | uint iget_flags) | |
348 | { | |
349 | xfs_buf_t *bp; | |
350 | xfs_dinode_t *dip; | |
351 | int error; | |
352 | ||
353 | /* | |
354 | * Fill in the location information in the in-core inode. | |
355 | */ | |
356 | error = xfs_imap(mp, tp, ip->i_ino, &ip->i_imap, iget_flags); | |
357 | if (error) | |
358 | return error; | |
359 | ||
360 | /* shortcut IO on inode allocation if possible */ | |
361 | if ((iget_flags & XFS_IGET_CREATE) && | |
362 | xfs_sb_version_hascrc(&mp->m_sb) && | |
363 | !(mp->m_flags & XFS_MOUNT_IKEEP)) { | |
364 | /* initialise the on-disk inode core */ | |
365 | memset(&ip->i_d, 0, sizeof(ip->i_d)); | |
366 | ip->i_d.di_magic = XFS_DINODE_MAGIC; | |
367 | ip->i_d.di_gen = prandom_u32(); | |
368 | if (xfs_sb_version_hascrc(&mp->m_sb)) { | |
369 | ip->i_d.di_version = 3; | |
370 | ip->i_d.di_ino = ip->i_ino; | |
371 | uuid_copy(&ip->i_d.di_uuid, &mp->m_sb.sb_uuid); | |
372 | } else | |
373 | ip->i_d.di_version = 2; | |
374 | return 0; | |
375 | } | |
376 | ||
377 | /* | |
378 | * Get pointers to the on-disk inode and the buffer containing it. | |
379 | */ | |
380 | error = xfs_imap_to_bp(mp, tp, &ip->i_imap, &dip, &bp, 0, iget_flags); | |
381 | if (error) | |
382 | return error; | |
383 | ||
384 | /* even unallocated inodes are verified */ | |
385 | if (!xfs_dinode_verify(mp, ip, dip)) { | |
386 | xfs_alert(mp, "%s: validation failed for inode %lld failed", | |
387 | __func__, ip->i_ino); | |
388 | ||
389 | XFS_CORRUPTION_ERROR(__func__, XFS_ERRLEVEL_LOW, mp, dip); | |
2451337d | 390 | error = -EFSCORRUPTED; |
1fd7115e DC |
391 | goto out_brelse; |
392 | } | |
393 | ||
394 | /* | |
395 | * If the on-disk inode is already linked to a directory | |
396 | * entry, copy all of the inode into the in-core inode. | |
397 | * xfs_iformat_fork() handles copying in the inode format | |
398 | * specific information. | |
399 | * Otherwise, just get the truly permanent information. | |
400 | */ | |
401 | if (dip->di_mode) { | |
402 | xfs_dinode_from_disk(&ip->i_d, dip); | |
403 | error = xfs_iformat_fork(ip, dip); | |
404 | if (error) { | |
405 | #ifdef DEBUG | |
406 | xfs_alert(mp, "%s: xfs_iformat() returned error %d", | |
407 | __func__, error); | |
408 | #endif /* DEBUG */ | |
409 | goto out_brelse; | |
410 | } | |
411 | } else { | |
412 | /* | |
413 | * Partial initialisation of the in-core inode. Just the bits | |
414 | * that xfs_ialloc won't overwrite or relies on being correct. | |
415 | */ | |
416 | ip->i_d.di_magic = be16_to_cpu(dip->di_magic); | |
417 | ip->i_d.di_version = dip->di_version; | |
418 | ip->i_d.di_gen = be32_to_cpu(dip->di_gen); | |
419 | ip->i_d.di_flushiter = be16_to_cpu(dip->di_flushiter); | |
420 | ||
421 | if (dip->di_version == 3) { | |
422 | ip->i_d.di_ino = be64_to_cpu(dip->di_ino); | |
423 | uuid_copy(&ip->i_d.di_uuid, &dip->di_uuid); | |
424 | } | |
425 | ||
426 | /* | |
427 | * Make sure to pull in the mode here as well in | |
428 | * case the inode is released without being used. | |
429 | * This ensures that xfs_inactive() will see that | |
430 | * the inode is already free and not try to mess | |
431 | * with the uninitialized part of it. | |
432 | */ | |
433 | ip->i_d.di_mode = 0; | |
434 | } | |
435 | ||
436 | /* | |
263997a6 DC |
437 | * Automatically convert version 1 inode formats in memory to version 2 |
438 | * inode format. If the inode is modified, it will get logged and | |
439 | * rewritten as a version 2 inode. We can do this because we set the | |
440 | * superblock feature bit for v2 inodes unconditionally during mount | |
441 | * and it means the reast of the code can assume the inode version is 2 | |
442 | * or higher. | |
1fd7115e DC |
443 | */ |
444 | if (ip->i_d.di_version == 1) { | |
263997a6 DC |
445 | ip->i_d.di_version = 2; |
446 | memset(&(ip->i_d.di_pad[0]), 0, sizeof(ip->i_d.di_pad)); | |
1fd7115e DC |
447 | ip->i_d.di_nlink = ip->i_d.di_onlink; |
448 | ip->i_d.di_onlink = 0; | |
449 | xfs_set_projid(ip, 0); | |
450 | } | |
451 | ||
452 | ip->i_delayed_blks = 0; | |
453 | ||
454 | /* | |
455 | * Mark the buffer containing the inode as something to keep | |
456 | * around for a while. This helps to keep recently accessed | |
457 | * meta-data in-core longer. | |
458 | */ | |
459 | xfs_buf_set_ref(bp, XFS_INO_REF); | |
460 | ||
461 | /* | |
462 | * Use xfs_trans_brelse() to release the buffer containing the on-disk | |
463 | * inode, because it was acquired with xfs_trans_read_buf() in | |
464 | * xfs_imap_to_bp() above. If tp is NULL, this is just a normal | |
465 | * brelse(). If we're within a transaction, then xfs_trans_brelse() | |
466 | * will only release the buffer if it is not dirty within the | |
467 | * transaction. It will be OK to release the buffer in this case, | |
468 | * because inodes on disk are never destroyed and we will be locking the | |
469 | * new in-core inode before putting it in the cache where other | |
470 | * processes can find it. Thus we don't have to worry about the inode | |
471 | * being changed just because we released the buffer. | |
472 | */ | |
473 | out_brelse: | |
474 | xfs_trans_brelse(tp, bp); | |
475 | return error; | |
476 | } |