]>
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 | 17 | */ |
9c799827 | 18 | #include "libxfs_priv.h" |
b626fb59 DC |
19 | #include "xfs_fs.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" | |
f944d3d0 | 25 | #include "xfs_defer.h" |
b626fb59 | 26 | #include "xfs_inode.h" |
56d3fc2b | 27 | #include "xfs_errortag.h" |
b626fb59 DC |
28 | #include "xfs_cksum.h" |
29 | #include "xfs_trans.h" | |
30 | #include "xfs_ialloc.h" | |
0962b5b3 | 31 | #include "xfs_dir2.h" |
2bd0ea18 | 32 | |
5e656dbb BN |
33 | /* |
34 | * Check that none of the inode's in the buffer have a next | |
35 | * unlinked field of 0. | |
36 | */ | |
37 | #if defined(DEBUG) | |
2bd0ea18 NS |
38 | void |
39 | xfs_inobp_check( | |
40 | xfs_mount_t *mp, | |
41 | xfs_buf_t *bp) | |
42 | { | |
43 | int i; | |
44 | int j; | |
45 | xfs_dinode_t *dip; | |
46 | ||
47 | j = mp->m_inode_cluster_size >> mp->m_sb.sb_inodelog; | |
48 | ||
49 | for (i = 0; i < j; i++) { | |
92acb899 | 50 | dip = xfs_buf_offset(bp, i * mp->m_sb.sb_inodesize); |
46eca962 | 51 | if (!dip->di_next_unlinked) { |
a2ceac1f | 52 | xfs_alert(mp, |
ea6a00d4 DC |
53 | "Detected bogus zero next_unlinked field in inode %d buffer 0x%llx.", |
54 | i, (long long)bp->b_bn); | |
2bd0ea18 NS |
55 | } |
56 | } | |
57 | } | |
58 | #endif | |
59 | ||
307ae950 RW |
60 | bool |
61 | xfs_dinode_good_version( | |
62 | struct xfs_mount *mp, | |
63 | __u8 version) | |
64 | { | |
65 | if (xfs_sb_version_hascrc(&mp->m_sb)) | |
66 | return version == 3; | |
67 | ||
68 | return version == 1 || version == 2; | |
69 | } | |
70 | ||
f85fc622 DC |
71 | /* |
72 | * If we are doing readahead on an inode buffer, we might be in log recovery | |
73 | * reading an inode allocation buffer that hasn't yet been replayed, and hence | |
74 | * has not had the inode cores stamped into it. Hence for readahead, the buffer | |
75 | * may be potentially invalid. | |
76 | * | |
736e1cf6 DC |
77 | * If the readahead buffer is invalid, we need to mark it with an error and |
78 | * clear the DONE status of the buffer so that a followup read will re-read it | |
79 | * from disk. We don't report the error otherwise to avoid warnings during log | |
80 | * recovery and we don't get unnecssary panics on debug kernels. We use EIO here | |
81 | * because all we want to do is say readahead failed; there is no-one to report | |
82 | * the error to, so this will distinguish it from a non-ra verifier failure. | |
6daba42a DC |
83 | * Changes to this readahead error behavour also need to be reflected in |
84 | * xfs_dquot_buf_readahead_verify(). | |
f85fc622 | 85 | */ |
a2ceac1f DC |
86 | static void |
87 | xfs_inode_buf_verify( | |
f85fc622 DC |
88 | struct xfs_buf *bp, |
89 | bool readahead) | |
5e656dbb | 90 | { |
a2ceac1f | 91 | struct xfs_mount *mp = bp->b_target->bt_mount; |
5e656dbb BN |
92 | int i; |
93 | int ni; | |
5e656dbb BN |
94 | |
95 | /* | |
96 | * Validate the magic number and version of every inode in the buffer | |
5e656dbb | 97 | */ |
a2ceac1f | 98 | ni = XFS_BB_TO_FSB(mp, bp->b_length) * mp->m_sb.sb_inopblock; |
5e656dbb BN |
99 | for (i = 0; i < ni; i++) { |
100 | int di_ok; | |
101 | xfs_dinode_t *dip; | |
102 | ||
92acb899 | 103 | dip = xfs_buf_offset(bp, (i << mp->m_sb.sb_inodelog)); |
a2ceac1f | 104 | di_ok = dip->di_magic == cpu_to_be16(XFS_DINODE_MAGIC) && |
307ae950 | 105 | xfs_dinode_good_version(mp, dip->di_version); |
5e656dbb | 106 | if (unlikely(XFS_TEST_ERROR(!di_ok, mp, |
e2a190dd | 107 | XFS_ERRTAG_ITOBP_INOTOBP))) { |
f85fc622 DC |
108 | if (readahead) { |
109 | bp->b_flags &= ~XBF_DONE; | |
736e1cf6 | 110 | xfs_buf_ioerror(bp, -EIO); |
f85fc622 DC |
111 | return; |
112 | } | |
113 | ||
1e697959 | 114 | xfs_verifier_error(bp, -EFSCORRUPTED, __this_address); |
5e656dbb | 115 | #ifdef DEBUG |
ea6a00d4 | 116 | xfs_alert(mp, |
a2ceac1f DC |
117 | "bad inode magic/vsn daddr %lld #%d (magic=%x)", |
118 | (unsigned long long)bp->b_bn, i, | |
56b2de80 | 119 | be16_to_cpu(dip->di_magic)); |
5e656dbb | 120 | #endif |
5e656dbb BN |
121 | } |
122 | } | |
5e656dbb | 123 | xfs_inobp_check(mp, bp); |
a2ceac1f | 124 | } |
5e656dbb | 125 | |
e6d77a21 | 126 | |
a2ceac1f DC |
127 | static void |
128 | xfs_inode_buf_read_verify( | |
129 | struct xfs_buf *bp) | |
130 | { | |
f85fc622 DC |
131 | xfs_inode_buf_verify(bp, false); |
132 | } | |
133 | ||
134 | static void | |
135 | xfs_inode_buf_readahead_verify( | |
136 | struct xfs_buf *bp) | |
137 | { | |
138 | xfs_inode_buf_verify(bp, true); | |
5e656dbb | 139 | } |
2bd0ea18 | 140 | |
a2ceac1f DC |
141 | static void |
142 | xfs_inode_buf_write_verify( | |
143 | struct xfs_buf *bp) | |
56b2de80 | 144 | { |
f85fc622 | 145 | xfs_inode_buf_verify(bp, false); |
56b2de80 DC |
146 | } |
147 | ||
a2ceac1f | 148 | const struct xfs_buf_ops xfs_inode_buf_ops = { |
a3fac935 | 149 | .name = "xfs_inode", |
a2ceac1f DC |
150 | .verify_read = xfs_inode_buf_read_verify, |
151 | .verify_write = xfs_inode_buf_write_verify, | |
152 | }; | |
153 | ||
f85fc622 | 154 | const struct xfs_buf_ops xfs_inode_buf_ra_ops = { |
a3fac935 | 155 | .name = "xxfs_inode_ra", |
f85fc622 DC |
156 | .verify_read = xfs_inode_buf_readahead_verify, |
157 | .verify_write = xfs_inode_buf_write_verify, | |
158 | }; | |
159 | ||
160 | ||
2bd0ea18 | 161 | /* |
a2ceac1f DC |
162 | * This routine is called to map an inode to the buffer containing the on-disk |
163 | * version of the inode. It returns a pointer to the buffer containing the | |
164 | * on-disk inode in the bpp parameter, and in the dipp parameter it returns a | |
165 | * pointer to the on-disk inode within that buffer. | |
2bd0ea18 | 166 | * |
a2ceac1f DC |
167 | * If a non-zero error is returned, then the contents of bpp and dipp are |
168 | * undefined. | |
2bd0ea18 NS |
169 | */ |
170 | int | |
a2ceac1f DC |
171 | xfs_imap_to_bp( |
172 | struct xfs_mount *mp, | |
173 | struct xfs_trans *tp, | |
174 | struct xfs_imap *imap, | |
175 | struct xfs_dinode **dipp, | |
176 | struct xfs_buf **bpp, | |
177 | uint buf_flags, | |
178 | uint iget_flags) | |
2bd0ea18 | 179 | { |
a2ceac1f DC |
180 | struct xfs_buf *bp; |
181 | int error; | |
2bd0ea18 | 182 | |
a2ceac1f DC |
183 | buf_flags |= XBF_UNMAPPED; |
184 | error = xfs_trans_read_buf(mp, tp, mp->m_ddev_targp, imap->im_blkno, | |
185 | (int)imap->im_len, buf_flags, &bp, | |
186 | &xfs_inode_buf_ops); | |
187 | if (error) { | |
12b53197 | 188 | if (error == -EAGAIN) { |
a2ceac1f DC |
189 | ASSERT(buf_flags & XBF_TRYLOCK); |
190 | return error; | |
191 | } | |
2bd0ea18 | 192 | |
12b53197 | 193 | if (error == -EFSCORRUPTED && |
a2ceac1f | 194 | (iget_flags & XFS_IGET_UNTRUSTED)) |
12b53197 | 195 | return -EINVAL; |
2bd0ea18 | 196 | |
a2ceac1f DC |
197 | xfs_warn(mp, "%s: xfs_trans_read_buf() returned error %d.", |
198 | __func__, error); | |
199 | return error; | |
2bd0ea18 | 200 | } |
2bd0ea18 | 201 | |
2bd0ea18 | 202 | *bpp = bp; |
92acb899 | 203 | *dipp = xfs_buf_offset(bp, imap->im_boffset); |
2bd0ea18 NS |
204 | return 0; |
205 | } | |
206 | ||
5000d01d | 207 | void |
1bc6cbe3 DC |
208 | xfs_inode_from_disk( |
209 | struct xfs_inode *ip, | |
138659f1 | 210 | struct xfs_dinode *from) |
2bd0ea18 | 211 | { |
1bc6cbe3 DC |
212 | struct xfs_icdinode *to = &ip->i_d; |
213 | struct inode *inode = VFS_I(ip); | |
214 | ||
f089fc42 DC |
215 | |
216 | /* | |
217 | * Convert v1 inodes immediately to v2 inode format as this is the | |
218 | * minimum inode version format we support in the rest of the code. | |
219 | */ | |
e37bf53c | 220 | to->di_version = from->di_version; |
f089fc42 | 221 | if (to->di_version == 1) { |
bcbe04c1 | 222 | set_nlink(inode, be16_to_cpu(from->di_onlink)); |
f089fc42 DC |
223 | to->di_projid_lo = 0; |
224 | to->di_projid_hi = 0; | |
225 | to->di_version = 2; | |
226 | } else { | |
bcbe04c1 | 227 | set_nlink(inode, be32_to_cpu(from->di_nlink)); |
f089fc42 DC |
228 | to->di_projid_lo = be16_to_cpu(from->di_projid_lo); |
229 | to->di_projid_hi = be16_to_cpu(from->di_projid_hi); | |
230 | } | |
231 | ||
5e656dbb | 232 | to->di_format = from->di_format; |
5e656dbb BN |
233 | to->di_uid = be32_to_cpu(from->di_uid); |
234 | to->di_gid = be32_to_cpu(from->di_gid); | |
5e656dbb | 235 | to->di_flushiter = be16_to_cpu(from->di_flushiter); |
1bc6cbe3 DC |
236 | |
237 | /* | |
238 | * Time is signed, so need to convert to signed 32 bit before | |
239 | * storing in inode timestamp which may be 64 bit. Otherwise | |
240 | * a time before epoch is converted to a time long after epoch | |
241 | * on 64 bit systems. | |
242 | */ | |
243 | inode->i_atime.tv_sec = (int)be32_to_cpu(from->di_atime.t_sec); | |
244 | inode->i_atime.tv_nsec = (int)be32_to_cpu(from->di_atime.t_nsec); | |
245 | inode->i_mtime.tv_sec = (int)be32_to_cpu(from->di_mtime.t_sec); | |
246 | inode->i_mtime.tv_nsec = (int)be32_to_cpu(from->di_mtime.t_nsec); | |
247 | inode->i_ctime.tv_sec = (int)be32_to_cpu(from->di_ctime.t_sec); | |
248 | inode->i_ctime.tv_nsec = (int)be32_to_cpu(from->di_ctime.t_nsec); | |
6652c253 | 249 | inode->i_generation = be32_to_cpu(from->di_gen); |
e37bf53c | 250 | inode->i_mode = be16_to_cpu(from->di_mode); |
1bc6cbe3 | 251 | |
5e656dbb BN |
252 | to->di_size = be64_to_cpu(from->di_size); |
253 | to->di_nblocks = be64_to_cpu(from->di_nblocks); | |
254 | to->di_extsize = be32_to_cpu(from->di_extsize); | |
255 | to->di_nextents = be32_to_cpu(from->di_nextents); | |
256 | to->di_anextents = be16_to_cpu(from->di_anextents); | |
257 | to->di_forkoff = from->di_forkoff; | |
258 | to->di_aformat = from->di_aformat; | |
259 | to->di_dmevmask = be32_to_cpu(from->di_dmevmask); | |
260 | to->di_dmstate = be16_to_cpu(from->di_dmstate); | |
261 | to->di_flags = be16_to_cpu(from->di_flags); | |
41ce5f36 DC |
262 | |
263 | if (to->di_version == 3) { | |
9abcc5cd | 264 | inode->i_version = be64_to_cpu(from->di_changecount); |
41ce5f36 DC |
265 | to->di_crtime.t_sec = be32_to_cpu(from->di_crtime.t_sec); |
266 | to->di_crtime.t_nsec = be32_to_cpu(from->di_crtime.t_nsec); | |
267 | to->di_flags2 = be64_to_cpu(from->di_flags2); | |
10460994 | 268 | to->di_cowextsize = be32_to_cpu(from->di_cowextsize); |
41ce5f36 | 269 | } |
5e656dbb BN |
270 | } |
271 | ||
272 | void | |
1bc6cbe3 DC |
273 | xfs_inode_to_disk( |
274 | struct xfs_inode *ip, | |
db17aebe DC |
275 | struct xfs_dinode *to, |
276 | xfs_lsn_t lsn) | |
1bc6cbe3 DC |
277 | { |
278 | struct xfs_icdinode *from = &ip->i_d; | |
279 | struct inode *inode = VFS_I(ip); | |
280 | ||
db17aebe | 281 | to->di_magic = cpu_to_be16(XFS_DINODE_MAGIC); |
f089fc42 | 282 | to->di_onlink = 0; |
db17aebe | 283 | |
db17aebe | 284 | to->di_version = from->di_version; |
1bc6cbe3 | 285 | to->di_format = from->di_format; |
1bc6cbe3 DC |
286 | to->di_uid = cpu_to_be32(from->di_uid); |
287 | to->di_gid = cpu_to_be32(from->di_gid); | |
1bc6cbe3 DC |
288 | to->di_projid_lo = cpu_to_be16(from->di_projid_lo); |
289 | to->di_projid_hi = cpu_to_be16(from->di_projid_hi); | |
1bc6cbe3 | 290 | |
db17aebe | 291 | memset(to->di_pad, 0, sizeof(to->di_pad)); |
1bc6cbe3 DC |
292 | to->di_atime.t_sec = cpu_to_be32(inode->i_atime.tv_sec); |
293 | to->di_atime.t_nsec = cpu_to_be32(inode->i_atime.tv_nsec); | |
294 | to->di_mtime.t_sec = cpu_to_be32(inode->i_mtime.tv_sec); | |
295 | to->di_mtime.t_nsec = cpu_to_be32(inode->i_mtime.tv_nsec); | |
296 | to->di_ctime.t_sec = cpu_to_be32(inode->i_ctime.tv_sec); | |
297 | to->di_ctime.t_nsec = cpu_to_be32(inode->i_ctime.tv_nsec); | |
bcbe04c1 | 298 | to->di_nlink = cpu_to_be32(inode->i_nlink); |
6652c253 | 299 | to->di_gen = cpu_to_be32(inode->i_generation); |
e37bf53c | 300 | to->di_mode = cpu_to_be16(inode->i_mode); |
1bc6cbe3 DC |
301 | |
302 | to->di_size = cpu_to_be64(from->di_size); | |
303 | to->di_nblocks = cpu_to_be64(from->di_nblocks); | |
304 | to->di_extsize = cpu_to_be32(from->di_extsize); | |
305 | to->di_nextents = cpu_to_be32(from->di_nextents); | |
306 | to->di_anextents = cpu_to_be16(from->di_anextents); | |
307 | to->di_forkoff = from->di_forkoff; | |
308 | to->di_aformat = from->di_aformat; | |
309 | to->di_dmevmask = cpu_to_be32(from->di_dmevmask); | |
310 | to->di_dmstate = cpu_to_be16(from->di_dmstate); | |
311 | to->di_flags = cpu_to_be16(from->di_flags); | |
1bc6cbe3 DC |
312 | |
313 | if (from->di_version == 3) { | |
9abcc5cd | 314 | to->di_changecount = cpu_to_be64(inode->i_version); |
1bc6cbe3 DC |
315 | to->di_crtime.t_sec = cpu_to_be32(from->di_crtime.t_sec); |
316 | to->di_crtime.t_nsec = cpu_to_be32(from->di_crtime.t_nsec); | |
317 | to->di_flags2 = cpu_to_be64(from->di_flags2); | |
10460994 | 318 | to->di_cowextsize = cpu_to_be32(from->di_cowextsize); |
db17aebe DC |
319 | to->di_ino = cpu_to_be64(ip->i_ino); |
320 | to->di_lsn = cpu_to_be64(lsn); | |
321 | memset(to->di_pad2, 0, sizeof(to->di_pad2)); | |
322 | uuid_copy(&to->di_uuid, &ip->i_mount->m_sb.sb_meta_uuid); | |
1bc6cbe3 DC |
323 | to->di_flushiter = 0; |
324 | } else { | |
325 | to->di_flushiter = cpu_to_be16(from->di_flushiter); | |
326 | } | |
327 | } | |
328 | ||
329 | void | |
330 | xfs_log_dinode_to_disk( | |
331 | struct xfs_log_dinode *from, | |
332 | struct xfs_dinode *to) | |
5e656dbb BN |
333 | { |
334 | to->di_magic = cpu_to_be16(from->di_magic); | |
335 | to->di_mode = cpu_to_be16(from->di_mode); | |
f089fc42 | 336 | to->di_version = from->di_version; |
5e656dbb | 337 | to->di_format = from->di_format; |
f089fc42 | 338 | to->di_onlink = 0; |
5e656dbb BN |
339 | to->di_uid = cpu_to_be32(from->di_uid); |
340 | to->di_gid = cpu_to_be32(from->di_gid); | |
341 | to->di_nlink = cpu_to_be32(from->di_nlink); | |
22bc10ed AM |
342 | to->di_projid_lo = cpu_to_be16(from->di_projid_lo); |
343 | to->di_projid_hi = cpu_to_be16(from->di_projid_hi); | |
5e656dbb | 344 | memcpy(to->di_pad, from->di_pad, sizeof(to->di_pad)); |
1bc6cbe3 | 345 | |
5e656dbb BN |
346 | to->di_atime.t_sec = cpu_to_be32(from->di_atime.t_sec); |
347 | to->di_atime.t_nsec = cpu_to_be32(from->di_atime.t_nsec); | |
348 | to->di_mtime.t_sec = cpu_to_be32(from->di_mtime.t_sec); | |
349 | to->di_mtime.t_nsec = cpu_to_be32(from->di_mtime.t_nsec); | |
350 | to->di_ctime.t_sec = cpu_to_be32(from->di_ctime.t_sec); | |
351 | to->di_ctime.t_nsec = cpu_to_be32(from->di_ctime.t_nsec); | |
1bc6cbe3 | 352 | |
5e656dbb BN |
353 | to->di_size = cpu_to_be64(from->di_size); |
354 | to->di_nblocks = cpu_to_be64(from->di_nblocks); | |
355 | to->di_extsize = cpu_to_be32(from->di_extsize); | |
356 | to->di_nextents = cpu_to_be32(from->di_nextents); | |
357 | to->di_anextents = cpu_to_be16(from->di_anextents); | |
358 | to->di_forkoff = from->di_forkoff; | |
359 | to->di_aformat = from->di_aformat; | |
360 | to->di_dmevmask = cpu_to_be32(from->di_dmevmask); | |
361 | to->di_dmstate = cpu_to_be16(from->di_dmstate); | |
362 | to->di_flags = cpu_to_be16(from->di_flags); | |
363 | to->di_gen = cpu_to_be32(from->di_gen); | |
41ce5f36 DC |
364 | |
365 | if (from->di_version == 3) { | |
366 | to->di_changecount = cpu_to_be64(from->di_changecount); | |
367 | to->di_crtime.t_sec = cpu_to_be32(from->di_crtime.t_sec); | |
368 | to->di_crtime.t_nsec = cpu_to_be32(from->di_crtime.t_nsec); | |
369 | to->di_flags2 = cpu_to_be64(from->di_flags2); | |
10460994 | 370 | to->di_cowextsize = cpu_to_be32(from->di_cowextsize); |
41ce5f36 DC |
371 | to->di_ino = cpu_to_be64(from->di_ino); |
372 | to->di_lsn = cpu_to_be64(from->di_lsn); | |
373 | memcpy(to->di_pad2, from->di_pad2, sizeof(to->di_pad2)); | |
f85fc622 | 374 | uuid_copy(&to->di_uuid, &from->di_uuid); |
fde71651 DC |
375 | to->di_flushiter = 0; |
376 | } else { | |
377 | to->di_flushiter = cpu_to_be16(from->di_flushiter); | |
41ce5f36 DC |
378 | } |
379 | } | |
380 | ||
bc01119d | 381 | xfs_failaddr_t |
41ce5f36 DC |
382 | xfs_dinode_verify( |
383 | struct xfs_mount *mp, | |
e515cca1 | 384 | xfs_ino_t ino, |
41ce5f36 DC |
385 | struct xfs_dinode *dip) |
386 | { | |
3cfabff6 | 387 | uint16_t mode; |
183537ed DW |
388 | uint16_t flags; |
389 | uint64_t flags2; | |
ad42e5a1 | 390 | uint64_t di_size; |
183537ed | 391 | |
41ce5f36 | 392 | if (dip->di_magic != cpu_to_be16(XFS_DINODE_MAGIC)) |
bc01119d | 393 | return __this_address; |
41ce5f36 | 394 | |
98703400 DW |
395 | /* Verify v3 integrity information first */ |
396 | if (dip->di_version >= 3) { | |
397 | if (!xfs_sb_version_hascrc(&mp->m_sb)) | |
398 | return __this_address; | |
399 | if (!xfs_verify_cksum((char *)dip, mp->m_sb.sb_inodesize, | |
400 | XFS_DINODE_CRC_OFF)) | |
401 | return __this_address; | |
402 | if (be64_to_cpu(dip->di_ino) != ino) | |
403 | return __this_address; | |
404 | if (!uuid_equal(&dip->di_uuid, &mp->m_sb.sb_meta_uuid)) | |
405 | return __this_address; | |
406 | } | |
407 | ||
9d85eb44 | 408 | /* don't allow invalid i_size */ |
ad42e5a1 DW |
409 | di_size = be64_to_cpu(dip->di_size); |
410 | if (di_size & (1ULL << 63)) | |
bc01119d | 411 | return __this_address; |
9d85eb44 | 412 | |
3cfabff6 | 413 | mode = be16_to_cpu(dip->di_mode); |
0962b5b3 | 414 | if (mode && xfs_mode_to_ftype(mode) == XFS_DIR3_FT_UNKNOWN) |
bc01119d | 415 | return __this_address; |
3cfabff6 AG |
416 | |
417 | /* No zero-length symlinks/dirs. */ | |
ad42e5a1 | 418 | if ((S_ISLNK(mode) || S_ISDIR(mode)) && di_size == 0) |
bc01119d | 419 | return __this_address; |
9d85eb44 | 420 | |
ad42e5a1 DW |
421 | /* Fork checks carried over from xfs_iformat_fork */ |
422 | if (mode && | |
423 | be32_to_cpu(dip->di_nextents) + be16_to_cpu(dip->di_anextents) > | |
424 | be64_to_cpu(dip->di_nblocks)) | |
425 | return __this_address; | |
426 | ||
427 | if (mode && XFS_DFORK_BOFF(dip) > mp->m_sb.sb_inodesize) | |
428 | return __this_address; | |
429 | ||
430 | flags = be16_to_cpu(dip->di_flags); | |
431 | ||
432 | if (mode && (flags & XFS_DIFLAG_REALTIME) && !mp->m_rtdev_targp) | |
433 | return __this_address; | |
434 | ||
435 | /* Do we have appropriate data fork formats for the mode? */ | |
436 | switch (mode & S_IFMT) { | |
437 | case S_IFIFO: | |
438 | case S_IFCHR: | |
439 | case S_IFBLK: | |
440 | case S_IFSOCK: | |
441 | if (dip->di_format != XFS_DINODE_FMT_DEV) | |
442 | return __this_address; | |
443 | break; | |
444 | case S_IFREG: | |
445 | case S_IFLNK: | |
446 | case S_IFDIR: | |
447 | switch (dip->di_format) { | |
448 | case XFS_DINODE_FMT_LOCAL: | |
449 | /* | |
450 | * no local regular files yet | |
451 | */ | |
452 | if (S_ISREG(mode)) | |
453 | return __this_address; | |
454 | if (di_size > XFS_DFORK_DSIZE(dip, mp)) | |
455 | return __this_address; | |
456 | /* fall through */ | |
457 | case XFS_DINODE_FMT_EXTENTS: | |
458 | case XFS_DINODE_FMT_BTREE: | |
459 | break; | |
460 | default: | |
461 | return __this_address; | |
462 | } | |
463 | break; | |
464 | case 0: | |
465 | /* Uninitialized inode ok. */ | |
466 | break; | |
467 | default: | |
468 | return __this_address; | |
469 | } | |
470 | ||
471 | if (XFS_DFORK_Q(dip)) { | |
472 | switch (dip->di_aformat) { | |
473 | case XFS_DINODE_FMT_LOCAL: | |
474 | case XFS_DINODE_FMT_EXTENTS: | |
475 | case XFS_DINODE_FMT_BTREE: | |
476 | break; | |
477 | default: | |
478 | return __this_address; | |
479 | } | |
480 | } | |
481 | ||
41ce5f36 DC |
482 | /* only version 3 or greater inodes are extensively verified here */ |
483 | if (dip->di_version < 3) | |
bc01119d | 484 | return NULL; |
41ce5f36 | 485 | |
183537ed DW |
486 | flags2 = be64_to_cpu(dip->di_flags2); |
487 | ||
488 | /* don't allow reflink/cowextsize if we don't have reflink */ | |
489 | if ((flags2 & (XFS_DIFLAG2_REFLINK | XFS_DIFLAG2_COWEXTSIZE)) && | |
490 | !xfs_sb_version_hasreflink(&mp->m_sb)) | |
bc01119d | 491 | return __this_address; |
183537ed | 492 | |
ad42e5a1 DW |
493 | /* only regular files get reflink */ |
494 | if ((flags2 & XFS_DIFLAG2_REFLINK) && (mode & S_IFMT) != S_IFREG) | |
495 | return __this_address; | |
496 | ||
183537ed DW |
497 | /* don't let reflink and realtime mix */ |
498 | if ((flags2 & XFS_DIFLAG2_REFLINK) && (flags & XFS_DIFLAG_REALTIME)) | |
bc01119d | 499 | return __this_address; |
183537ed | 500 | |
378ed4e7 DW |
501 | /* don't let reflink and dax mix */ |
502 | if ((flags2 & XFS_DIFLAG2_REFLINK) && (flags2 & XFS_DIFLAG2_DAX)) | |
bc01119d | 503 | return __this_address; |
378ed4e7 | 504 | |
bc01119d | 505 | return NULL; |
41ce5f36 | 506 | } |
e6d77a21 | 507 | |
41ce5f36 DC |
508 | void |
509 | xfs_dinode_calc_crc( | |
510 | struct xfs_mount *mp, | |
511 | struct xfs_dinode *dip) | |
512 | { | |
4a492e72 | 513 | uint32_t crc; |
41ce5f36 DC |
514 | |
515 | if (dip->di_version < 3) | |
516 | return; | |
517 | ||
518 | ASSERT(xfs_sb_version_hascrc(&mp->m_sb)); | |
0bb90214 | 519 | crc = xfs_start_cksum_update((char *)dip, mp->m_sb.sb_inodesize, |
3faddc54 | 520 | XFS_DINODE_CRC_OFF); |
41ce5f36 | 521 | dip->di_crc = xfs_end_cksum(crc); |
2bd0ea18 NS |
522 | } |
523 | ||
a2ceac1f DC |
524 | /* |
525 | * Read the disk inode attributes into the in-core inode structure. | |
ff105f75 DC |
526 | * |
527 | * For version 5 superblocks, if we are initialising a new inode and we are not | |
528 | * utilising the XFS_MOUNT_IKEEP inode cluster mode, we can simple build the new | |
529 | * inode core with a random generation number. If we are keeping inodes around, | |
530 | * we need to read the inode cluster to get the existing generation number off | |
531 | * disk. Further, if we are using version 4 superblocks (i.e. v1/v2 inode | |
532 | * format) then log recovery is dependent on the di_flushiter field being | |
533 | * initialised from the current on-disk value and hence we must also read the | |
534 | * inode off disk. | |
a2ceac1f DC |
535 | */ |
536 | int | |
537 | xfs_iread( | |
538 | xfs_mount_t *mp, | |
539 | xfs_trans_t *tp, | |
540 | xfs_inode_t *ip, | |
541 | uint iget_flags) | |
542 | { | |
543 | xfs_buf_t *bp; | |
544 | xfs_dinode_t *dip; | |
bc01119d | 545 | xfs_failaddr_t fa; |
a2ceac1f DC |
546 | int error; |
547 | ||
548 | /* | |
549 | * Fill in the location information in the in-core inode. | |
550 | */ | |
551 | error = xfs_imap(mp, tp, ip->i_ino, &ip->i_imap, iget_flags); | |
552 | if (error) | |
553 | return error; | |
554 | ||
ff105f75 DC |
555 | /* shortcut IO on inode allocation if possible */ |
556 | if ((iget_flags & XFS_IGET_CREATE) && | |
557 | xfs_sb_version_hascrc(&mp->m_sb) && | |
558 | !(mp->m_flags & XFS_MOUNT_IKEEP)) { | |
559 | /* initialise the on-disk inode core */ | |
560 | memset(&ip->i_d, 0, sizeof(ip->i_d)); | |
6652c253 | 561 | VFS_I(ip)->i_generation = prandom_u32(); |
db17aebe | 562 | if (xfs_sb_version_hascrc(&mp->m_sb)) |
ff105f75 | 563 | ip->i_d.di_version = 3; |
db17aebe | 564 | else |
ff105f75 DC |
565 | ip->i_d.di_version = 2; |
566 | return 0; | |
567 | } | |
568 | ||
a2ceac1f DC |
569 | /* |
570 | * Get pointers to the on-disk inode and the buffer containing it. | |
571 | */ | |
572 | error = xfs_imap_to_bp(mp, tp, &ip->i_imap, &dip, &bp, 0, iget_flags); | |
573 | if (error) | |
574 | return error; | |
575 | ||
41ce5f36 | 576 | /* even unallocated inodes are verified */ |
bc01119d DW |
577 | fa = xfs_dinode_verify(mp, ip->i_ino, dip); |
578 | if (fa) { | |
579 | xfs_alert(mp, "%s: validation failed for inode %lld at %pS", | |
580 | __func__, ip->i_ino, fa); | |
41ce5f36 DC |
581 | |
582 | XFS_CORRUPTION_ERROR(__func__, XFS_ERRLEVEL_LOW, mp, dip); | |
12b53197 | 583 | error = -EFSCORRUPTED; |
a2ceac1f DC |
584 | goto out_brelse; |
585 | } | |
586 | ||
587 | /* | |
588 | * If the on-disk inode is already linked to a directory | |
589 | * entry, copy all of the inode into the in-core inode. | |
5d90ab5a | 590 | * xfs_iformat_fork() handles copying in the inode format |
a2ceac1f DC |
591 | * specific information. |
592 | * Otherwise, just get the truly permanent information. | |
593 | */ | |
594 | if (dip->di_mode) { | |
1bc6cbe3 | 595 | xfs_inode_from_disk(ip, dip); |
5d90ab5a | 596 | error = xfs_iformat_fork(ip, dip); |
a2ceac1f DC |
597 | if (error) { |
598 | #ifdef DEBUG | |
599 | xfs_alert(mp, "%s: xfs_iformat() returned error %d", | |
600 | __func__, error); | |
601 | #endif /* DEBUG */ | |
602 | goto out_brelse; | |
603 | } | |
604 | } else { | |
41ce5f36 DC |
605 | /* |
606 | * Partial initialisation of the in-core inode. Just the bits | |
607 | * that xfs_ialloc won't overwrite or relies on being correct. | |
608 | */ | |
a2ceac1f | 609 | ip->i_d.di_version = dip->di_version; |
6652c253 | 610 | VFS_I(ip)->i_generation = be32_to_cpu(dip->di_gen); |
a2ceac1f | 611 | ip->i_d.di_flushiter = be16_to_cpu(dip->di_flushiter); |
41ce5f36 | 612 | |
a2ceac1f DC |
613 | /* |
614 | * Make sure to pull in the mode here as well in | |
615 | * case the inode is released without being used. | |
616 | * This ensures that xfs_inactive() will see that | |
617 | * the inode is already free and not try to mess | |
618 | * with the uninitialized part of it. | |
619 | */ | |
e37bf53c | 620 | VFS_I(ip)->i_mode = 0; |
a2ceac1f DC |
621 | } |
622 | ||
f089fc42 | 623 | ASSERT(ip->i_d.di_version >= 2); |
a2ceac1f DC |
624 | ip->i_delayed_blks = 0; |
625 | ||
626 | /* | |
627 | * Mark the buffer containing the inode as something to keep | |
628 | * around for a while. This helps to keep recently accessed | |
629 | * meta-data in-core longer. | |
630 | */ | |
631 | xfs_buf_set_ref(bp, XFS_INO_REF); | |
632 | ||
633 | /* | |
1d519883 DC |
634 | * Use xfs_trans_brelse() to release the buffer containing the on-disk |
635 | * inode, because it was acquired with xfs_trans_read_buf() in | |
636 | * xfs_imap_to_bp() above. If tp is NULL, this is just a normal | |
a2ceac1f DC |
637 | * brelse(). If we're within a transaction, then xfs_trans_brelse() |
638 | * will only release the buffer if it is not dirty within the | |
639 | * transaction. It will be OK to release the buffer in this case, | |
1d519883 DC |
640 | * because inodes on disk are never destroyed and we will be locking the |
641 | * new in-core inode before putting it in the cache where other | |
642 | * processes can find it. Thus we don't have to worry about the inode | |
643 | * being changed just because we released the buffer. | |
a2ceac1f DC |
644 | */ |
645 | out_brelse: | |
646 | xfs_trans_brelse(tp, bp); | |
647 | return error; | |
648 | } |