]>
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 DC |
26 | #include "xfs_inode.h" |
27 | #include "xfs_cksum.h" | |
28 | #include "xfs_trans.h" | |
29 | #include "xfs_ialloc.h" | |
2bd0ea18 | 30 | |
5e656dbb BN |
31 | /* |
32 | * Check that none of the inode's in the buffer have a next | |
33 | * unlinked field of 0. | |
34 | */ | |
35 | #if defined(DEBUG) | |
2bd0ea18 NS |
36 | void |
37 | xfs_inobp_check( | |
38 | xfs_mount_t *mp, | |
39 | xfs_buf_t *bp) | |
40 | { | |
41 | int i; | |
42 | int j; | |
43 | xfs_dinode_t *dip; | |
44 | ||
45 | j = mp->m_inode_cluster_size >> mp->m_sb.sb_inodelog; | |
46 | ||
47 | for (i = 0; i < j; i++) { | |
92acb899 | 48 | dip = xfs_buf_offset(bp, i * mp->m_sb.sb_inodesize); |
46eca962 | 49 | if (!dip->di_next_unlinked) { |
a2ceac1f | 50 | xfs_alert(mp, |
ea6a00d4 DC |
51 | "Detected bogus zero next_unlinked field in inode %d buffer 0x%llx.", |
52 | i, (long long)bp->b_bn); | |
2bd0ea18 NS |
53 | } |
54 | } | |
55 | } | |
56 | #endif | |
57 | ||
307ae950 RW |
58 | bool |
59 | xfs_dinode_good_version( | |
60 | struct xfs_mount *mp, | |
61 | __u8 version) | |
62 | { | |
63 | if (xfs_sb_version_hascrc(&mp->m_sb)) | |
64 | return version == 3; | |
65 | ||
66 | return version == 1 || version == 2; | |
67 | } | |
68 | ||
f85fc622 DC |
69 | /* |
70 | * If we are doing readahead on an inode buffer, we might be in log recovery | |
71 | * reading an inode allocation buffer that hasn't yet been replayed, and hence | |
72 | * has not had the inode cores stamped into it. Hence for readahead, the buffer | |
73 | * may be potentially invalid. | |
74 | * | |
736e1cf6 DC |
75 | * If the readahead buffer is invalid, we need to mark it with an error and |
76 | * clear the DONE status of the buffer so that a followup read will re-read it | |
77 | * from disk. We don't report the error otherwise to avoid warnings during log | |
78 | * recovery and we don't get unnecssary panics on debug kernels. We use EIO here | |
79 | * because all we want to do is say readahead failed; there is no-one to report | |
80 | * the error to, so this will distinguish it from a non-ra verifier failure. | |
6daba42a DC |
81 | * Changes to this readahead error behavour also need to be reflected in |
82 | * xfs_dquot_buf_readahead_verify(). | |
f85fc622 | 83 | */ |
a2ceac1f DC |
84 | static void |
85 | xfs_inode_buf_verify( | |
f85fc622 DC |
86 | struct xfs_buf *bp, |
87 | bool readahead) | |
5e656dbb | 88 | { |
a2ceac1f | 89 | struct xfs_mount *mp = bp->b_target->bt_mount; |
5e656dbb BN |
90 | int i; |
91 | int ni; | |
5e656dbb BN |
92 | |
93 | /* | |
94 | * Validate the magic number and version of every inode in the buffer | |
5e656dbb | 95 | */ |
a2ceac1f | 96 | ni = XFS_BB_TO_FSB(mp, bp->b_length) * mp->m_sb.sb_inopblock; |
5e656dbb BN |
97 | for (i = 0; i < ni; i++) { |
98 | int di_ok; | |
99 | xfs_dinode_t *dip; | |
100 | ||
92acb899 | 101 | dip = xfs_buf_offset(bp, (i << mp->m_sb.sb_inodelog)); |
a2ceac1f | 102 | di_ok = dip->di_magic == cpu_to_be16(XFS_DINODE_MAGIC) && |
307ae950 | 103 | xfs_dinode_good_version(mp, dip->di_version); |
5e656dbb BN |
104 | if (unlikely(XFS_TEST_ERROR(!di_ok, mp, |
105 | XFS_ERRTAG_ITOBP_INOTOBP, | |
106 | XFS_RANDOM_ITOBP_INOTOBP))) { | |
f85fc622 DC |
107 | if (readahead) { |
108 | bp->b_flags &= ~XBF_DONE; | |
736e1cf6 | 109 | xfs_buf_ioerror(bp, -EIO); |
f85fc622 DC |
110 | return; |
111 | } | |
112 | ||
12b53197 | 113 | xfs_buf_ioerror(bp, -EFSCORRUPTED); |
45922933 | 114 | xfs_verifier_error(bp); |
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 | ||
a73b88f2 | 381 | bool |
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 | { | |
183537ed DW |
387 | uint16_t flags; |
388 | uint64_t flags2; | |
389 | ||
41ce5f36 DC |
390 | if (dip->di_magic != cpu_to_be16(XFS_DINODE_MAGIC)) |
391 | return false; | |
392 | ||
393 | /* only version 3 or greater inodes are extensively verified here */ | |
394 | if (dip->di_version < 3) | |
395 | return true; | |
396 | ||
397 | if (!xfs_sb_version_hascrc(&mp->m_sb)) | |
398 | return false; | |
399 | if (!xfs_verify_cksum((char *)dip, mp->m_sb.sb_inodesize, | |
3faddc54 | 400 | XFS_DINODE_CRC_OFF)) |
41ce5f36 | 401 | return false; |
e515cca1 | 402 | if (be64_to_cpu(dip->di_ino) != ino) |
41ce5f36 | 403 | return false; |
9c4e12fb | 404 | if (!uuid_equal(&dip->di_uuid, &mp->m_sb.sb_meta_uuid)) |
41ce5f36 | 405 | return false; |
183537ed DW |
406 | |
407 | flags = be16_to_cpu(dip->di_flags); | |
408 | flags2 = be64_to_cpu(dip->di_flags2); | |
409 | ||
410 | /* don't allow reflink/cowextsize if we don't have reflink */ | |
411 | if ((flags2 & (XFS_DIFLAG2_REFLINK | XFS_DIFLAG2_COWEXTSIZE)) && | |
412 | !xfs_sb_version_hasreflink(&mp->m_sb)) | |
413 | return false; | |
414 | ||
415 | /* don't let reflink and realtime mix */ | |
416 | if ((flags2 & XFS_DIFLAG2_REFLINK) && (flags & XFS_DIFLAG_REALTIME)) | |
417 | return false; | |
418 | ||
378ed4e7 DW |
419 | /* don't let reflink and dax mix */ |
420 | if ((flags2 & XFS_DIFLAG2_REFLINK) && (flags2 & XFS_DIFLAG2_DAX)) | |
421 | return false; | |
422 | ||
41ce5f36 DC |
423 | return true; |
424 | } | |
e6d77a21 | 425 | |
41ce5f36 DC |
426 | void |
427 | xfs_dinode_calc_crc( | |
428 | struct xfs_mount *mp, | |
429 | struct xfs_dinode *dip) | |
430 | { | |
431 | __uint32_t crc; | |
432 | ||
433 | if (dip->di_version < 3) | |
434 | return; | |
435 | ||
436 | ASSERT(xfs_sb_version_hascrc(&mp->m_sb)); | |
437 | crc = xfs_start_cksum((char *)dip, mp->m_sb.sb_inodesize, | |
3faddc54 | 438 | XFS_DINODE_CRC_OFF); |
41ce5f36 | 439 | dip->di_crc = xfs_end_cksum(crc); |
2bd0ea18 NS |
440 | } |
441 | ||
a2ceac1f DC |
442 | /* |
443 | * Read the disk inode attributes into the in-core inode structure. | |
ff105f75 DC |
444 | * |
445 | * For version 5 superblocks, if we are initialising a new inode and we are not | |
446 | * utilising the XFS_MOUNT_IKEEP inode cluster mode, we can simple build the new | |
447 | * inode core with a random generation number. If we are keeping inodes around, | |
448 | * we need to read the inode cluster to get the existing generation number off | |
449 | * disk. Further, if we are using version 4 superblocks (i.e. v1/v2 inode | |
450 | * format) then log recovery is dependent on the di_flushiter field being | |
451 | * initialised from the current on-disk value and hence we must also read the | |
452 | * inode off disk. | |
a2ceac1f DC |
453 | */ |
454 | int | |
455 | xfs_iread( | |
456 | xfs_mount_t *mp, | |
457 | xfs_trans_t *tp, | |
458 | xfs_inode_t *ip, | |
459 | uint iget_flags) | |
460 | { | |
461 | xfs_buf_t *bp; | |
462 | xfs_dinode_t *dip; | |
463 | int error; | |
464 | ||
465 | /* | |
466 | * Fill in the location information in the in-core inode. | |
467 | */ | |
468 | error = xfs_imap(mp, tp, ip->i_ino, &ip->i_imap, iget_flags); | |
469 | if (error) | |
470 | return error; | |
471 | ||
ff105f75 DC |
472 | /* shortcut IO on inode allocation if possible */ |
473 | if ((iget_flags & XFS_IGET_CREATE) && | |
474 | xfs_sb_version_hascrc(&mp->m_sb) && | |
475 | !(mp->m_flags & XFS_MOUNT_IKEEP)) { | |
476 | /* initialise the on-disk inode core */ | |
477 | memset(&ip->i_d, 0, sizeof(ip->i_d)); | |
6652c253 | 478 | VFS_I(ip)->i_generation = prandom_u32(); |
db17aebe | 479 | if (xfs_sb_version_hascrc(&mp->m_sb)) |
ff105f75 | 480 | ip->i_d.di_version = 3; |
db17aebe | 481 | else |
ff105f75 DC |
482 | ip->i_d.di_version = 2; |
483 | return 0; | |
484 | } | |
485 | ||
a2ceac1f DC |
486 | /* |
487 | * Get pointers to the on-disk inode and the buffer containing it. | |
488 | */ | |
489 | error = xfs_imap_to_bp(mp, tp, &ip->i_imap, &dip, &bp, 0, iget_flags); | |
490 | if (error) | |
491 | return error; | |
492 | ||
41ce5f36 | 493 | /* even unallocated inodes are verified */ |
e515cca1 | 494 | if (!xfs_dinode_verify(mp, ip->i_ino, dip)) { |
41ce5f36 DC |
495 | xfs_alert(mp, "%s: validation failed for inode %lld failed", |
496 | __func__, ip->i_ino); | |
497 | ||
498 | XFS_CORRUPTION_ERROR(__func__, XFS_ERRLEVEL_LOW, mp, dip); | |
12b53197 | 499 | error = -EFSCORRUPTED; |
a2ceac1f DC |
500 | goto out_brelse; |
501 | } | |
502 | ||
503 | /* | |
504 | * If the on-disk inode is already linked to a directory | |
505 | * entry, copy all of the inode into the in-core inode. | |
5d90ab5a | 506 | * xfs_iformat_fork() handles copying in the inode format |
a2ceac1f DC |
507 | * specific information. |
508 | * Otherwise, just get the truly permanent information. | |
509 | */ | |
510 | if (dip->di_mode) { | |
1bc6cbe3 | 511 | xfs_inode_from_disk(ip, dip); |
5d90ab5a | 512 | error = xfs_iformat_fork(ip, dip); |
a2ceac1f DC |
513 | if (error) { |
514 | #ifdef DEBUG | |
515 | xfs_alert(mp, "%s: xfs_iformat() returned error %d", | |
516 | __func__, error); | |
517 | #endif /* DEBUG */ | |
518 | goto out_brelse; | |
519 | } | |
520 | } else { | |
41ce5f36 DC |
521 | /* |
522 | * Partial initialisation of the in-core inode. Just the bits | |
523 | * that xfs_ialloc won't overwrite or relies on being correct. | |
524 | */ | |
a2ceac1f | 525 | ip->i_d.di_version = dip->di_version; |
6652c253 | 526 | VFS_I(ip)->i_generation = be32_to_cpu(dip->di_gen); |
a2ceac1f | 527 | ip->i_d.di_flushiter = be16_to_cpu(dip->di_flushiter); |
41ce5f36 | 528 | |
a2ceac1f DC |
529 | /* |
530 | * Make sure to pull in the mode here as well in | |
531 | * case the inode is released without being used. | |
532 | * This ensures that xfs_inactive() will see that | |
533 | * the inode is already free and not try to mess | |
534 | * with the uninitialized part of it. | |
535 | */ | |
e37bf53c | 536 | VFS_I(ip)->i_mode = 0; |
a2ceac1f DC |
537 | } |
538 | ||
f089fc42 | 539 | ASSERT(ip->i_d.di_version >= 2); |
a2ceac1f DC |
540 | |
541 | ip->i_delayed_blks = 0; | |
542 | ||
543 | /* | |
544 | * Mark the buffer containing the inode as something to keep | |
545 | * around for a while. This helps to keep recently accessed | |
546 | * meta-data in-core longer. | |
547 | */ | |
548 | xfs_buf_set_ref(bp, XFS_INO_REF); | |
549 | ||
550 | /* | |
1d519883 DC |
551 | * Use xfs_trans_brelse() to release the buffer containing the on-disk |
552 | * inode, because it was acquired with xfs_trans_read_buf() in | |
553 | * xfs_imap_to_bp() above. If tp is NULL, this is just a normal | |
a2ceac1f DC |
554 | * brelse(). If we're within a transaction, then xfs_trans_brelse() |
555 | * will only release the buffer if it is not dirty within the | |
556 | * transaction. It will be OK to release the buffer in this case, | |
1d519883 DC |
557 | * because inodes on disk are never destroyed and we will be locking the |
558 | * new in-core inode before putting it in the cache where other | |
559 | * processes can find it. Thus we don't have to worry about the inode | |
560 | * being changed just because we released the buffer. | |
a2ceac1f DC |
561 | */ |
562 | out_brelse: | |
563 | xfs_trans_brelse(tp, bp); | |
564 | return error; | |
565 | } |