]>
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" |
3ab78df2 | 25 | #include "xfs_defer.h" |
1fd7115e | 26 | #include "xfs_inode.h" |
e9e899a2 | 27 | #include "xfs_errortag.h" |
1fd7115e DC |
28 | #include "xfs_error.h" |
29 | #include "xfs_cksum.h" | |
30 | #include "xfs_icache.h" | |
239880ef | 31 | #include "xfs_trans.h" |
a4fbe6ab | 32 | #include "xfs_ialloc.h" |
a324cbf1 | 33 | #include "xfs_dir2.h" |
1fd7115e | 34 | |
f0e28280 JL |
35 | #include <linux/iversion.h> |
36 | ||
1fd7115e DC |
37 | /* |
38 | * Check that none of the inode's in the buffer have a next | |
39 | * unlinked field of 0. | |
40 | */ | |
41 | #if defined(DEBUG) | |
42 | void | |
43 | xfs_inobp_check( | |
44 | xfs_mount_t *mp, | |
45 | xfs_buf_t *bp) | |
46 | { | |
47 | int i; | |
48 | int j; | |
49 | xfs_dinode_t *dip; | |
50 | ||
51 | j = mp->m_inode_cluster_size >> mp->m_sb.sb_inodelog; | |
52 | ||
53 | for (i = 0; i < j; i++) { | |
88ee2df7 | 54 | dip = xfs_buf_offset(bp, i * mp->m_sb.sb_inodesize); |
1fd7115e DC |
55 | if (!dip->di_next_unlinked) { |
56 | xfs_alert(mp, | |
74ffa796 DC |
57 | "Detected bogus zero next_unlinked field in inode %d buffer 0x%llx.", |
58 | i, (long long)bp->b_bn); | |
1fd7115e DC |
59 | } |
60 | } | |
61 | } | |
62 | #endif | |
63 | ||
8cdcc810 RW |
64 | bool |
65 | xfs_dinode_good_version( | |
66 | struct xfs_mount *mp, | |
67 | __u8 version) | |
68 | { | |
69 | if (xfs_sb_version_hascrc(&mp->m_sb)) | |
70 | return version == 3; | |
71 | ||
72 | return version == 1 || version == 2; | |
73 | } | |
74 | ||
d8914002 DC |
75 | /* |
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. | |
80 | * | |
b79f4a1c DC |
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. | |
7d6a13f0 DC |
87 | * Changes to this readahead error behavour also need to be reflected in |
88 | * xfs_dquot_buf_readahead_verify(). | |
d8914002 | 89 | */ |
1fd7115e DC |
90 | static void |
91 | xfs_inode_buf_verify( | |
d8914002 DC |
92 | struct xfs_buf *bp, |
93 | bool readahead) | |
1fd7115e DC |
94 | { |
95 | struct xfs_mount *mp = bp->b_target->bt_mount; | |
6a96c565 | 96 | xfs_agnumber_t agno; |
1fd7115e DC |
97 | int i; |
98 | int ni; | |
99 | ||
100 | /* | |
101 | * Validate the magic number and version of every inode in the buffer | |
102 | */ | |
6a96c565 | 103 | agno = xfs_daddr_to_agno(mp, XFS_BUF_ADDR(bp)); |
1fd7115e DC |
104 | ni = XFS_BB_TO_FSB(mp, bp->b_length) * mp->m_sb.sb_inopblock; |
105 | for (i = 0; i < ni; i++) { | |
106 | int di_ok; | |
107 | xfs_dinode_t *dip; | |
6a96c565 | 108 | xfs_agino_t unlinked_ino; |
1fd7115e | 109 | |
88ee2df7 | 110 | dip = xfs_buf_offset(bp, (i << mp->m_sb.sb_inodelog)); |
6a96c565 | 111 | unlinked_ino = be32_to_cpu(dip->di_next_unlinked); |
1fd7115e | 112 | di_ok = dip->di_magic == cpu_to_be16(XFS_DINODE_MAGIC) && |
6a96c565 DW |
113 | xfs_dinode_good_version(mp, dip->di_version) && |
114 | (unlinked_ino == NULLAGINO || | |
115 | xfs_verify_agino(mp, agno, unlinked_ino)); | |
1fd7115e | 116 | if (unlikely(XFS_TEST_ERROR(!di_ok, mp, |
9e24cfd0 | 117 | XFS_ERRTAG_ITOBP_INOTOBP))) { |
d8914002 DC |
118 | if (readahead) { |
119 | bp->b_flags &= ~XBF_DONE; | |
b79f4a1c | 120 | xfs_buf_ioerror(bp, -EIO); |
d8914002 DC |
121 | return; |
122 | } | |
123 | ||
1fd7115e | 124 | #ifdef DEBUG |
74ffa796 | 125 | xfs_alert(mp, |
1fd7115e DC |
126 | "bad inode magic/vsn daddr %lld #%d (magic=%x)", |
127 | (unsigned long long)bp->b_bn, i, | |
128 | be16_to_cpu(dip->di_magic)); | |
1fd7115e | 129 | #endif |
6edb1810 DW |
130 | xfs_buf_verifier_error(bp, -EFSCORRUPTED, |
131 | __func__, dip, sizeof(*dip), | |
132 | NULL); | |
6a96c565 | 133 | return; |
1fd7115e DC |
134 | } |
135 | } | |
1fd7115e DC |
136 | } |
137 | ||
138 | ||
139 | static void | |
140 | xfs_inode_buf_read_verify( | |
141 | struct xfs_buf *bp) | |
142 | { | |
d8914002 DC |
143 | xfs_inode_buf_verify(bp, false); |
144 | } | |
145 | ||
146 | static void | |
147 | xfs_inode_buf_readahead_verify( | |
148 | struct xfs_buf *bp) | |
149 | { | |
150 | xfs_inode_buf_verify(bp, true); | |
1fd7115e DC |
151 | } |
152 | ||
153 | static void | |
154 | xfs_inode_buf_write_verify( | |
155 | struct xfs_buf *bp) | |
156 | { | |
d8914002 | 157 | xfs_inode_buf_verify(bp, false); |
1fd7115e DC |
158 | } |
159 | ||
160 | const struct xfs_buf_ops xfs_inode_buf_ops = { | |
233135b7 | 161 | .name = "xfs_inode", |
1fd7115e DC |
162 | .verify_read = xfs_inode_buf_read_verify, |
163 | .verify_write = xfs_inode_buf_write_verify, | |
164 | }; | |
165 | ||
d8914002 | 166 | const struct xfs_buf_ops xfs_inode_buf_ra_ops = { |
233135b7 | 167 | .name = "xxfs_inode_ra", |
d8914002 DC |
168 | .verify_read = xfs_inode_buf_readahead_verify, |
169 | .verify_write = xfs_inode_buf_write_verify, | |
170 | }; | |
171 | ||
1fd7115e DC |
172 | |
173 | /* | |
174 | * This routine is called to map an inode to the buffer containing the on-disk | |
175 | * version of the inode. It returns a pointer to the buffer containing the | |
176 | * on-disk inode in the bpp parameter, and in the dipp parameter it returns a | |
177 | * pointer to the on-disk inode within that buffer. | |
178 | * | |
179 | * If a non-zero error is returned, then the contents of bpp and dipp are | |
180 | * undefined. | |
181 | */ | |
182 | int | |
183 | xfs_imap_to_bp( | |
184 | struct xfs_mount *mp, | |
185 | struct xfs_trans *tp, | |
186 | struct xfs_imap *imap, | |
187 | struct xfs_dinode **dipp, | |
188 | struct xfs_buf **bpp, | |
189 | uint buf_flags, | |
190 | uint iget_flags) | |
191 | { | |
192 | struct xfs_buf *bp; | |
193 | int error; | |
194 | ||
195 | buf_flags |= XBF_UNMAPPED; | |
196 | error = xfs_trans_read_buf(mp, tp, mp->m_ddev_targp, imap->im_blkno, | |
197 | (int)imap->im_len, buf_flags, &bp, | |
198 | &xfs_inode_buf_ops); | |
199 | if (error) { | |
2451337d | 200 | if (error == -EAGAIN) { |
1fd7115e DC |
201 | ASSERT(buf_flags & XBF_TRYLOCK); |
202 | return error; | |
203 | } | |
204 | ||
2451337d | 205 | if (error == -EFSCORRUPTED && |
1fd7115e | 206 | (iget_flags & XFS_IGET_UNTRUSTED)) |
2451337d | 207 | return -EINVAL; |
1fd7115e DC |
208 | |
209 | xfs_warn(mp, "%s: xfs_trans_read_buf() returned error %d.", | |
210 | __func__, error); | |
211 | return error; | |
212 | } | |
213 | ||
214 | *bpp = bp; | |
88ee2df7 | 215 | *dipp = xfs_buf_offset(bp, imap->im_boffset); |
1fd7115e DC |
216 | return 0; |
217 | } | |
218 | ||
638f4416 | 219 | void |
3987848c DC |
220 | xfs_inode_from_disk( |
221 | struct xfs_inode *ip, | |
f8d55aa0 | 222 | struct xfs_dinode *from) |
1fd7115e | 223 | { |
3987848c DC |
224 | struct xfs_icdinode *to = &ip->i_d; |
225 | struct inode *inode = VFS_I(ip); | |
226 | ||
faeb4e47 DC |
227 | |
228 | /* | |
229 | * Convert v1 inodes immediately to v2 inode format as this is the | |
230 | * minimum inode version format we support in the rest of the code. | |
231 | */ | |
c19b3b05 | 232 | to->di_version = from->di_version; |
faeb4e47 | 233 | if (to->di_version == 1) { |
54d7b5c1 | 234 | set_nlink(inode, be16_to_cpu(from->di_onlink)); |
faeb4e47 DC |
235 | to->di_projid_lo = 0; |
236 | to->di_projid_hi = 0; | |
237 | to->di_version = 2; | |
238 | } else { | |
54d7b5c1 | 239 | set_nlink(inode, be32_to_cpu(from->di_nlink)); |
faeb4e47 DC |
240 | to->di_projid_lo = be16_to_cpu(from->di_projid_lo); |
241 | to->di_projid_hi = be16_to_cpu(from->di_projid_hi); | |
242 | } | |
243 | ||
1fd7115e | 244 | to->di_format = from->di_format; |
1fd7115e DC |
245 | to->di_uid = be32_to_cpu(from->di_uid); |
246 | to->di_gid = be32_to_cpu(from->di_gid); | |
1fd7115e | 247 | to->di_flushiter = be16_to_cpu(from->di_flushiter); |
3987848c DC |
248 | |
249 | /* | |
250 | * Time is signed, so need to convert to signed 32 bit before | |
251 | * storing in inode timestamp which may be 64 bit. Otherwise | |
252 | * a time before epoch is converted to a time long after epoch | |
253 | * on 64 bit systems. | |
254 | */ | |
255 | inode->i_atime.tv_sec = (int)be32_to_cpu(from->di_atime.t_sec); | |
256 | inode->i_atime.tv_nsec = (int)be32_to_cpu(from->di_atime.t_nsec); | |
257 | inode->i_mtime.tv_sec = (int)be32_to_cpu(from->di_mtime.t_sec); | |
258 | inode->i_mtime.tv_nsec = (int)be32_to_cpu(from->di_mtime.t_nsec); | |
259 | inode->i_ctime.tv_sec = (int)be32_to_cpu(from->di_ctime.t_sec); | |
260 | inode->i_ctime.tv_nsec = (int)be32_to_cpu(from->di_ctime.t_nsec); | |
9e9a2674 | 261 | inode->i_generation = be32_to_cpu(from->di_gen); |
c19b3b05 | 262 | inode->i_mode = be16_to_cpu(from->di_mode); |
3987848c | 263 | |
1fd7115e DC |
264 | to->di_size = be64_to_cpu(from->di_size); |
265 | to->di_nblocks = be64_to_cpu(from->di_nblocks); | |
266 | to->di_extsize = be32_to_cpu(from->di_extsize); | |
267 | to->di_nextents = be32_to_cpu(from->di_nextents); | |
268 | to->di_anextents = be16_to_cpu(from->di_anextents); | |
269 | to->di_forkoff = from->di_forkoff; | |
270 | to->di_aformat = from->di_aformat; | |
271 | to->di_dmevmask = be32_to_cpu(from->di_dmevmask); | |
272 | to->di_dmstate = be16_to_cpu(from->di_dmstate); | |
273 | to->di_flags = be16_to_cpu(from->di_flags); | |
1fd7115e DC |
274 | |
275 | if (to->di_version == 3) { | |
f0e28280 JL |
276 | inode_set_iversion_queried(inode, |
277 | be64_to_cpu(from->di_changecount)); | |
1fd7115e DC |
278 | to->di_crtime.t_sec = be32_to_cpu(from->di_crtime.t_sec); |
279 | to->di_crtime.t_nsec = be32_to_cpu(from->di_crtime.t_nsec); | |
280 | to->di_flags2 = be64_to_cpu(from->di_flags2); | |
f7ca3522 | 281 | to->di_cowextsize = be32_to_cpu(from->di_cowextsize); |
1fd7115e DC |
282 | } |
283 | } | |
284 | ||
285 | void | |
3987848c DC |
286 | xfs_inode_to_disk( |
287 | struct xfs_inode *ip, | |
93f958f9 DC |
288 | struct xfs_dinode *to, |
289 | xfs_lsn_t lsn) | |
3987848c DC |
290 | { |
291 | struct xfs_icdinode *from = &ip->i_d; | |
292 | struct inode *inode = VFS_I(ip); | |
293 | ||
93f958f9 | 294 | to->di_magic = cpu_to_be16(XFS_DINODE_MAGIC); |
faeb4e47 | 295 | to->di_onlink = 0; |
93f958f9 | 296 | |
93f958f9 | 297 | to->di_version = from->di_version; |
3987848c | 298 | to->di_format = from->di_format; |
3987848c DC |
299 | to->di_uid = cpu_to_be32(from->di_uid); |
300 | to->di_gid = cpu_to_be32(from->di_gid); | |
3987848c DC |
301 | to->di_projid_lo = cpu_to_be16(from->di_projid_lo); |
302 | to->di_projid_hi = cpu_to_be16(from->di_projid_hi); | |
3987848c | 303 | |
93f958f9 | 304 | memset(to->di_pad, 0, sizeof(to->di_pad)); |
3987848c DC |
305 | to->di_atime.t_sec = cpu_to_be32(inode->i_atime.tv_sec); |
306 | to->di_atime.t_nsec = cpu_to_be32(inode->i_atime.tv_nsec); | |
307 | to->di_mtime.t_sec = cpu_to_be32(inode->i_mtime.tv_sec); | |
308 | to->di_mtime.t_nsec = cpu_to_be32(inode->i_mtime.tv_nsec); | |
309 | to->di_ctime.t_sec = cpu_to_be32(inode->i_ctime.tv_sec); | |
310 | to->di_ctime.t_nsec = cpu_to_be32(inode->i_ctime.tv_nsec); | |
54d7b5c1 | 311 | to->di_nlink = cpu_to_be32(inode->i_nlink); |
9e9a2674 | 312 | to->di_gen = cpu_to_be32(inode->i_generation); |
c19b3b05 | 313 | to->di_mode = cpu_to_be16(inode->i_mode); |
3987848c DC |
314 | |
315 | to->di_size = cpu_to_be64(from->di_size); | |
316 | to->di_nblocks = cpu_to_be64(from->di_nblocks); | |
317 | to->di_extsize = cpu_to_be32(from->di_extsize); | |
318 | to->di_nextents = cpu_to_be32(from->di_nextents); | |
319 | to->di_anextents = cpu_to_be16(from->di_anextents); | |
320 | to->di_forkoff = from->di_forkoff; | |
321 | to->di_aformat = from->di_aformat; | |
322 | to->di_dmevmask = cpu_to_be32(from->di_dmevmask); | |
323 | to->di_dmstate = cpu_to_be16(from->di_dmstate); | |
324 | to->di_flags = cpu_to_be16(from->di_flags); | |
3987848c DC |
325 | |
326 | if (from->di_version == 3) { | |
f0e28280 | 327 | to->di_changecount = cpu_to_be64(inode_peek_iversion(inode)); |
3987848c DC |
328 | to->di_crtime.t_sec = cpu_to_be32(from->di_crtime.t_sec); |
329 | to->di_crtime.t_nsec = cpu_to_be32(from->di_crtime.t_nsec); | |
330 | to->di_flags2 = cpu_to_be64(from->di_flags2); | |
f7ca3522 | 331 | to->di_cowextsize = cpu_to_be32(from->di_cowextsize); |
93f958f9 DC |
332 | to->di_ino = cpu_to_be64(ip->i_ino); |
333 | to->di_lsn = cpu_to_be64(lsn); | |
334 | memset(to->di_pad2, 0, sizeof(to->di_pad2)); | |
335 | uuid_copy(&to->di_uuid, &ip->i_mount->m_sb.sb_meta_uuid); | |
3987848c DC |
336 | to->di_flushiter = 0; |
337 | } else { | |
338 | to->di_flushiter = cpu_to_be16(from->di_flushiter); | |
339 | } | |
340 | } | |
341 | ||
342 | void | |
343 | xfs_log_dinode_to_disk( | |
344 | struct xfs_log_dinode *from, | |
345 | struct xfs_dinode *to) | |
1fd7115e DC |
346 | { |
347 | to->di_magic = cpu_to_be16(from->di_magic); | |
348 | to->di_mode = cpu_to_be16(from->di_mode); | |
faeb4e47 | 349 | to->di_version = from->di_version; |
1fd7115e | 350 | to->di_format = from->di_format; |
faeb4e47 | 351 | to->di_onlink = 0; |
1fd7115e DC |
352 | to->di_uid = cpu_to_be32(from->di_uid); |
353 | to->di_gid = cpu_to_be32(from->di_gid); | |
354 | to->di_nlink = cpu_to_be32(from->di_nlink); | |
355 | to->di_projid_lo = cpu_to_be16(from->di_projid_lo); | |
356 | to->di_projid_hi = cpu_to_be16(from->di_projid_hi); | |
357 | memcpy(to->di_pad, from->di_pad, sizeof(to->di_pad)); | |
3987848c | 358 | |
1fd7115e DC |
359 | to->di_atime.t_sec = cpu_to_be32(from->di_atime.t_sec); |
360 | to->di_atime.t_nsec = cpu_to_be32(from->di_atime.t_nsec); | |
361 | to->di_mtime.t_sec = cpu_to_be32(from->di_mtime.t_sec); | |
362 | to->di_mtime.t_nsec = cpu_to_be32(from->di_mtime.t_nsec); | |
363 | to->di_ctime.t_sec = cpu_to_be32(from->di_ctime.t_sec); | |
364 | to->di_ctime.t_nsec = cpu_to_be32(from->di_ctime.t_nsec); | |
3987848c | 365 | |
1fd7115e DC |
366 | to->di_size = cpu_to_be64(from->di_size); |
367 | to->di_nblocks = cpu_to_be64(from->di_nblocks); | |
368 | to->di_extsize = cpu_to_be32(from->di_extsize); | |
369 | to->di_nextents = cpu_to_be32(from->di_nextents); | |
370 | to->di_anextents = cpu_to_be16(from->di_anextents); | |
371 | to->di_forkoff = from->di_forkoff; | |
372 | to->di_aformat = from->di_aformat; | |
373 | to->di_dmevmask = cpu_to_be32(from->di_dmevmask); | |
374 | to->di_dmstate = cpu_to_be16(from->di_dmstate); | |
375 | to->di_flags = cpu_to_be16(from->di_flags); | |
376 | to->di_gen = cpu_to_be32(from->di_gen); | |
377 | ||
378 | if (from->di_version == 3) { | |
379 | to->di_changecount = cpu_to_be64(from->di_changecount); | |
380 | to->di_crtime.t_sec = cpu_to_be32(from->di_crtime.t_sec); | |
381 | to->di_crtime.t_nsec = cpu_to_be32(from->di_crtime.t_nsec); | |
382 | to->di_flags2 = cpu_to_be64(from->di_flags2); | |
f7ca3522 | 383 | to->di_cowextsize = cpu_to_be32(from->di_cowextsize); |
1fd7115e DC |
384 | to->di_ino = cpu_to_be64(from->di_ino); |
385 | to->di_lsn = cpu_to_be64(from->di_lsn); | |
386 | memcpy(to->di_pad2, from->di_pad2, sizeof(to->di_pad2)); | |
387 | uuid_copy(&to->di_uuid, &from->di_uuid); | |
388 | to->di_flushiter = 0; | |
389 | } else { | |
390 | to->di_flushiter = cpu_to_be16(from->di_flushiter); | |
391 | } | |
392 | } | |
393 | ||
a6a781a5 | 394 | xfs_failaddr_t |
1fd7115e DC |
395 | xfs_dinode_verify( |
396 | struct xfs_mount *mp, | |
420fbeb4 | 397 | xfs_ino_t ino, |
1fd7115e DC |
398 | struct xfs_dinode *dip) |
399 | { | |
3c6f46ea | 400 | uint16_t mode; |
c8e156ac DW |
401 | uint16_t flags; |
402 | uint64_t flags2; | |
71493b83 | 403 | uint64_t di_size; |
c8e156ac | 404 | |
1fd7115e | 405 | if (dip->di_magic != cpu_to_be16(XFS_DINODE_MAGIC)) |
a6a781a5 | 406 | return __this_address; |
1fd7115e | 407 | |
50aa90ef DW |
408 | /* Verify v3 integrity information first */ |
409 | if (dip->di_version >= 3) { | |
410 | if (!xfs_sb_version_hascrc(&mp->m_sb)) | |
411 | return __this_address; | |
412 | if (!xfs_verify_cksum((char *)dip, mp->m_sb.sb_inodesize, | |
413 | XFS_DINODE_CRC_OFF)) | |
414 | return __this_address; | |
415 | if (be64_to_cpu(dip->di_ino) != ino) | |
416 | return __this_address; | |
417 | if (!uuid_equal(&dip->di_uuid, &mp->m_sb.sb_meta_uuid)) | |
418 | return __this_address; | |
419 | } | |
1fd7115e | 420 | |
ef388e20 | 421 | /* don't allow invalid i_size */ |
71493b83 DW |
422 | di_size = be64_to_cpu(dip->di_size); |
423 | if (di_size & (1ULL << 63)) | |
a6a781a5 | 424 | return __this_address; |
ef388e20 | 425 | |
3c6f46ea | 426 | mode = be16_to_cpu(dip->di_mode); |
a324cbf1 | 427 | if (mode && xfs_mode_to_ftype(mode) == XFS_DIR3_FT_UNKNOWN) |
a6a781a5 | 428 | return __this_address; |
3c6f46ea AG |
429 | |
430 | /* No zero-length symlinks/dirs. */ | |
71493b83 | 431 | if ((S_ISLNK(mode) || S_ISDIR(mode)) && di_size == 0) |
a6a781a5 | 432 | return __this_address; |
ef388e20 | 433 | |
71493b83 DW |
434 | /* Fork checks carried over from xfs_iformat_fork */ |
435 | if (mode && | |
436 | be32_to_cpu(dip->di_nextents) + be16_to_cpu(dip->di_anextents) > | |
437 | be64_to_cpu(dip->di_nblocks)) | |
438 | return __this_address; | |
439 | ||
440 | if (mode && XFS_DFORK_BOFF(dip) > mp->m_sb.sb_inodesize) | |
441 | return __this_address; | |
442 | ||
443 | flags = be16_to_cpu(dip->di_flags); | |
444 | ||
445 | if (mode && (flags & XFS_DIFLAG_REALTIME) && !mp->m_rtdev_targp) | |
446 | return __this_address; | |
447 | ||
448 | /* Do we have appropriate data fork formats for the mode? */ | |
449 | switch (mode & S_IFMT) { | |
450 | case S_IFIFO: | |
451 | case S_IFCHR: | |
452 | case S_IFBLK: | |
453 | case S_IFSOCK: | |
454 | if (dip->di_format != XFS_DINODE_FMT_DEV) | |
455 | return __this_address; | |
456 | break; | |
457 | case S_IFREG: | |
458 | case S_IFLNK: | |
459 | case S_IFDIR: | |
460 | switch (dip->di_format) { | |
461 | case XFS_DINODE_FMT_LOCAL: | |
462 | /* | |
463 | * no local regular files yet | |
464 | */ | |
465 | if (S_ISREG(mode)) | |
466 | return __this_address; | |
467 | if (di_size > XFS_DFORK_DSIZE(dip, mp)) | |
468 | return __this_address; | |
b42db086 ES |
469 | if (dip->di_nextents) |
470 | return __this_address; | |
71493b83 DW |
471 | /* fall through */ |
472 | case XFS_DINODE_FMT_EXTENTS: | |
473 | case XFS_DINODE_FMT_BTREE: | |
474 | break; | |
475 | default: | |
476 | return __this_address; | |
477 | } | |
478 | break; | |
479 | case 0: | |
480 | /* Uninitialized inode ok. */ | |
481 | break; | |
482 | default: | |
483 | return __this_address; | |
484 | } | |
485 | ||
486 | if (XFS_DFORK_Q(dip)) { | |
487 | switch (dip->di_aformat) { | |
488 | case XFS_DINODE_FMT_LOCAL: | |
b42db086 ES |
489 | if (dip->di_anextents) |
490 | return __this_address; | |
491 | /* fall through */ | |
71493b83 DW |
492 | case XFS_DINODE_FMT_EXTENTS: |
493 | case XFS_DINODE_FMT_BTREE: | |
494 | break; | |
495 | default: | |
496 | return __this_address; | |
497 | } | |
b42db086 ES |
498 | } else { |
499 | /* | |
500 | * If there is no fork offset, this may be a freshly-made inode | |
501 | * in a new disk cluster, in which case di_aformat is zeroed. | |
502 | * Otherwise, such an inode must be in EXTENTS format; this goes | |
503 | * for freed inodes as well. | |
504 | */ | |
505 | switch (dip->di_aformat) { | |
506 | case 0: | |
507 | case XFS_DINODE_FMT_EXTENTS: | |
508 | break; | |
509 | default: | |
510 | return __this_address; | |
511 | } | |
512 | if (dip->di_anextents) | |
513 | return __this_address; | |
71493b83 | 514 | } |
ef388e20 | 515 | |
1fd7115e DC |
516 | /* only version 3 or greater inodes are extensively verified here */ |
517 | if (dip->di_version < 3) | |
a6a781a5 | 518 | return NULL; |
c8e156ac | 519 | |
c8e156ac DW |
520 | flags2 = be64_to_cpu(dip->di_flags2); |
521 | ||
522 | /* don't allow reflink/cowextsize if we don't have reflink */ | |
523 | if ((flags2 & (XFS_DIFLAG2_REFLINK | XFS_DIFLAG2_COWEXTSIZE)) && | |
524 | !xfs_sb_version_hasreflink(&mp->m_sb)) | |
a6a781a5 | 525 | return __this_address; |
c8e156ac | 526 | |
71493b83 DW |
527 | /* only regular files get reflink */ |
528 | if ((flags2 & XFS_DIFLAG2_REFLINK) && (mode & S_IFMT) != S_IFREG) | |
529 | return __this_address; | |
c8e156ac DW |
530 | |
531 | /* don't let reflink and realtime mix */ | |
532 | if ((flags2 & XFS_DIFLAG2_REFLINK) && (flags & XFS_DIFLAG_REALTIME)) | |
a6a781a5 | 533 | return __this_address; |
c8e156ac | 534 | |
4f435ebe DW |
535 | /* don't let reflink and dax mix */ |
536 | if ((flags2 & XFS_DIFLAG2_REFLINK) && (flags2 & XFS_DIFLAG2_DAX)) | |
a6a781a5 | 537 | return __this_address; |
4f435ebe | 538 | |
a6a781a5 | 539 | return NULL; |
1fd7115e DC |
540 | } |
541 | ||
542 | void | |
543 | xfs_dinode_calc_crc( | |
544 | struct xfs_mount *mp, | |
545 | struct xfs_dinode *dip) | |
546 | { | |
c8ce540d | 547 | uint32_t crc; |
1fd7115e DC |
548 | |
549 | if (dip->di_version < 3) | |
550 | return; | |
551 | ||
552 | ASSERT(xfs_sb_version_hascrc(&mp->m_sb)); | |
cae028df | 553 | crc = xfs_start_cksum_update((char *)dip, mp->m_sb.sb_inodesize, |
533b81c8 | 554 | XFS_DINODE_CRC_OFF); |
1fd7115e DC |
555 | dip->di_crc = xfs_end_cksum(crc); |
556 | } | |
557 | ||
558 | /* | |
559 | * Read the disk inode attributes into the in-core inode structure. | |
560 | * | |
561 | * For version 5 superblocks, if we are initialising a new inode and we are not | |
562 | * utilising the XFS_MOUNT_IKEEP inode cluster mode, we can simple build the new | |
563 | * inode core with a random generation number. If we are keeping inodes around, | |
564 | * we need to read the inode cluster to get the existing generation number off | |
565 | * disk. Further, if we are using version 4 superblocks (i.e. v1/v2 inode | |
566 | * format) then log recovery is dependent on the di_flushiter field being | |
567 | * initialised from the current on-disk value and hence we must also read the | |
568 | * inode off disk. | |
569 | */ | |
570 | int | |
571 | xfs_iread( | |
572 | xfs_mount_t *mp, | |
573 | xfs_trans_t *tp, | |
574 | xfs_inode_t *ip, | |
575 | uint iget_flags) | |
576 | { | |
577 | xfs_buf_t *bp; | |
578 | xfs_dinode_t *dip; | |
a6a781a5 | 579 | xfs_failaddr_t fa; |
1fd7115e DC |
580 | int error; |
581 | ||
582 | /* | |
583 | * Fill in the location information in the in-core inode. | |
584 | */ | |
585 | error = xfs_imap(mp, tp, ip->i_ino, &ip->i_imap, iget_flags); | |
586 | if (error) | |
587 | return error; | |
588 | ||
589 | /* shortcut IO on inode allocation if possible */ | |
590 | if ((iget_flags & XFS_IGET_CREATE) && | |
591 | xfs_sb_version_hascrc(&mp->m_sb) && | |
592 | !(mp->m_flags & XFS_MOUNT_IKEEP)) { | |
593 | /* initialise the on-disk inode core */ | |
594 | memset(&ip->i_d, 0, sizeof(ip->i_d)); | |
9e9a2674 | 595 | VFS_I(ip)->i_generation = prandom_u32(); |
fa4493f0 | 596 | ip->i_d.di_version = 3; |
1fd7115e DC |
597 | return 0; |
598 | } | |
599 | ||
600 | /* | |
601 | * Get pointers to the on-disk inode and the buffer containing it. | |
602 | */ | |
603 | error = xfs_imap_to_bp(mp, tp, &ip->i_imap, &dip, &bp, 0, iget_flags); | |
604 | if (error) | |
605 | return error; | |
606 | ||
607 | /* even unallocated inodes are verified */ | |
a6a781a5 DW |
608 | fa = xfs_dinode_verify(mp, ip->i_ino, dip); |
609 | if (fa) { | |
22431bf3 DW |
610 | xfs_inode_verifier_error(ip, -EFSCORRUPTED, "dinode", dip, |
611 | sizeof(*dip), fa); | |
2451337d | 612 | error = -EFSCORRUPTED; |
1fd7115e DC |
613 | goto out_brelse; |
614 | } | |
615 | ||
616 | /* | |
617 | * If the on-disk inode is already linked to a directory | |
618 | * entry, copy all of the inode into the in-core inode. | |
619 | * xfs_iformat_fork() handles copying in the inode format | |
620 | * specific information. | |
621 | * Otherwise, just get the truly permanent information. | |
622 | */ | |
623 | if (dip->di_mode) { | |
3987848c | 624 | xfs_inode_from_disk(ip, dip); |
1fd7115e DC |
625 | error = xfs_iformat_fork(ip, dip); |
626 | if (error) { | |
627 | #ifdef DEBUG | |
628 | xfs_alert(mp, "%s: xfs_iformat() returned error %d", | |
629 | __func__, error); | |
630 | #endif /* DEBUG */ | |
631 | goto out_brelse; | |
632 | } | |
633 | } else { | |
634 | /* | |
635 | * Partial initialisation of the in-core inode. Just the bits | |
636 | * that xfs_ialloc won't overwrite or relies on being correct. | |
637 | */ | |
1fd7115e | 638 | ip->i_d.di_version = dip->di_version; |
9e9a2674 | 639 | VFS_I(ip)->i_generation = be32_to_cpu(dip->di_gen); |
1fd7115e DC |
640 | ip->i_d.di_flushiter = be16_to_cpu(dip->di_flushiter); |
641 | ||
1fd7115e DC |
642 | /* |
643 | * Make sure to pull in the mode here as well in | |
644 | * case the inode is released without being used. | |
645 | * This ensures that xfs_inactive() will see that | |
646 | * the inode is already free and not try to mess | |
647 | * with the uninitialized part of it. | |
648 | */ | |
c19b3b05 | 649 | VFS_I(ip)->i_mode = 0; |
1fd7115e DC |
650 | } |
651 | ||
faeb4e47 | 652 | ASSERT(ip->i_d.di_version >= 2); |
1fd7115e DC |
653 | ip->i_delayed_blks = 0; |
654 | ||
655 | /* | |
656 | * Mark the buffer containing the inode as something to keep | |
657 | * around for a while. This helps to keep recently accessed | |
658 | * meta-data in-core longer. | |
659 | */ | |
660 | xfs_buf_set_ref(bp, XFS_INO_REF); | |
661 | ||
662 | /* | |
663 | * Use xfs_trans_brelse() to release the buffer containing the on-disk | |
664 | * inode, because it was acquired with xfs_trans_read_buf() in | |
665 | * xfs_imap_to_bp() above. If tp is NULL, this is just a normal | |
666 | * brelse(). If we're within a transaction, then xfs_trans_brelse() | |
667 | * will only release the buffer if it is not dirty within the | |
668 | * transaction. It will be OK to release the buffer in this case, | |
669 | * because inodes on disk are never destroyed and we will be locking the | |
670 | * new in-core inode before putting it in the cache where other | |
671 | * processes can find it. Thus we don't have to worry about the inode | |
672 | * being changed just because we released the buffer. | |
673 | */ | |
674 | out_brelse: | |
675 | xfs_trans_brelse(tp, bp); | |
676 | return error; | |
677 | } | |
8bb82bc1 DW |
678 | |
679 | /* | |
680 | * Validate di_extsize hint. | |
681 | * | |
682 | * The rules are documented at xfs_ioctl_setattr_check_extsize(). | |
683 | * These functions must be kept in sync with each other. | |
684 | */ | |
685 | xfs_failaddr_t | |
686 | xfs_inode_validate_extsize( | |
687 | struct xfs_mount *mp, | |
688 | uint32_t extsize, | |
689 | uint16_t mode, | |
690 | uint16_t flags) | |
691 | { | |
692 | bool rt_flag; | |
693 | bool hint_flag; | |
694 | bool inherit_flag; | |
695 | uint32_t extsize_bytes; | |
696 | uint32_t blocksize_bytes; | |
697 | ||
698 | rt_flag = (flags & XFS_DIFLAG_REALTIME); | |
699 | hint_flag = (flags & XFS_DIFLAG_EXTSIZE); | |
700 | inherit_flag = (flags & XFS_DIFLAG_EXTSZINHERIT); | |
701 | extsize_bytes = XFS_FSB_TO_B(mp, extsize); | |
702 | ||
703 | if (rt_flag) | |
704 | blocksize_bytes = mp->m_sb.sb_rextsize << mp->m_sb.sb_blocklog; | |
705 | else | |
706 | blocksize_bytes = mp->m_sb.sb_blocksize; | |
707 | ||
708 | if ((hint_flag || inherit_flag) && !(S_ISDIR(mode) || S_ISREG(mode))) | |
709 | return __this_address; | |
710 | ||
711 | if (hint_flag && !S_ISREG(mode)) | |
712 | return __this_address; | |
713 | ||
714 | if (inherit_flag && !S_ISDIR(mode)) | |
715 | return __this_address; | |
716 | ||
717 | if ((hint_flag || inherit_flag) && extsize == 0) | |
718 | return __this_address; | |
719 | ||
720 | if (!(hint_flag || inherit_flag) && extsize != 0) | |
721 | return __this_address; | |
722 | ||
723 | if (extsize_bytes % blocksize_bytes) | |
724 | return __this_address; | |
725 | ||
726 | if (extsize > MAXEXTLEN) | |
727 | return __this_address; | |
728 | ||
729 | if (!rt_flag && extsize > mp->m_sb.sb_agblocks / 2) | |
730 | return __this_address; | |
731 | ||
732 | return NULL; | |
733 | } | |
734 | ||
735 | /* | |
736 | * Validate di_cowextsize hint. | |
737 | * | |
738 | * The rules are documented at xfs_ioctl_setattr_check_cowextsize(). | |
739 | * These functions must be kept in sync with each other. | |
740 | */ | |
741 | xfs_failaddr_t | |
742 | xfs_inode_validate_cowextsize( | |
743 | struct xfs_mount *mp, | |
744 | uint32_t cowextsize, | |
745 | uint16_t mode, | |
746 | uint16_t flags, | |
747 | uint64_t flags2) | |
748 | { | |
749 | bool rt_flag; | |
750 | bool hint_flag; | |
751 | uint32_t cowextsize_bytes; | |
752 | ||
753 | rt_flag = (flags & XFS_DIFLAG_REALTIME); | |
754 | hint_flag = (flags2 & XFS_DIFLAG2_COWEXTSIZE); | |
755 | cowextsize_bytes = XFS_FSB_TO_B(mp, cowextsize); | |
756 | ||
757 | if (hint_flag && !xfs_sb_version_hasreflink(&mp->m_sb)) | |
758 | return __this_address; | |
759 | ||
760 | if (hint_flag && !(S_ISDIR(mode) || S_ISREG(mode))) | |
761 | return __this_address; | |
762 | ||
763 | if (hint_flag && cowextsize == 0) | |
764 | return __this_address; | |
765 | ||
766 | if (!hint_flag && cowextsize != 0) | |
767 | return __this_address; | |
768 | ||
769 | if (hint_flag && rt_flag) | |
770 | return __this_address; | |
771 | ||
772 | if (cowextsize_bytes % mp->m_sb.sb_blocksize) | |
773 | return __this_address; | |
774 | ||
775 | if (cowextsize > MAXEXTLEN) | |
776 | return __this_address; | |
777 | ||
778 | if (cowextsize > mp->m_sb.sb_agblocks / 2) | |
779 | return __this_address; | |
780 | ||
781 | return NULL; | |
782 | } |