]>
Commit | Line | Data |
---|---|---|
4b1ac34b NB |
1 | /* |
2 | * mdadm - manage Linux "md" devices aka RAID arrays. | |
3 | * | |
4 | * Copyright (C) 2001-2004 Neil Brown <neilb@cse.unsw.edu.au> | |
5 | * | |
6 | * | |
7 | * This program is free software; you can redistribute it and/or modify | |
8 | * it under the terms of the GNU General Public License as published by | |
9 | * the Free Software Foundation; either version 2 of the License, or | |
10 | * (at your option) any later version. | |
11 | * | |
12 | * This program is distributed in the hope that it will be useful, | |
13 | * but WITHOUT ANY WARRANTY; without even the implied warranty of | |
14 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | |
15 | * GNU General Public License for more details. | |
16 | * | |
17 | * You should have received a copy of the GNU General Public License | |
18 | * along with this program; if not, write to the Free Software | |
19 | * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA | |
20 | * | |
21 | * Author: Neil Brown | |
22 | * Email: <neilb@cse.unsw.edu.au> | |
23 | * Paper: Neil Brown | |
24 | * School of Computer Science and Engineering | |
25 | * The University of New South Wales | |
26 | * Sydney, 2052 | |
27 | * Australia | |
28 | */ | |
29 | ||
30 | #include "mdadm.h" | |
31 | ||
32 | /* | |
33 | * All handling for the 0.90.0 version superblock is in | |
34 | * this file. | |
35 | * This includes: | |
36 | * - finding, loading, and writing the superblock. | |
37 | * - initialising a new superblock | |
38 | * - printing the superblock for --examine | |
39 | * - printing part of the superblock for --detail | |
40 | * .. other stuff | |
41 | */ | |
42 | ||
43 | ||
44 | static unsigned long calc_sb0_csum(mdp_super_t *super) | |
45 | { | |
46 | unsigned long csum = super->sb_csum; | |
47 | unsigned long newcsum; | |
48 | super->sb_csum= 0 ; | |
49 | newcsum = calc_csum(super, MD_SB_BYTES); | |
50 | super->sb_csum = csum; | |
51 | return newcsum; | |
52 | } | |
53 | ||
54 | void examine_super0(void *sbv) | |
55 | { | |
56 | mdp_super_t *sb = sbv; | |
57 | time_t atime; | |
58 | int d; | |
59 | char *c; | |
60 | ||
61 | printf(" Magic : %08x\n", sb->md_magic); | |
62 | printf(" Version : %02d.%02d.%02d\n", sb->major_version, sb->minor_version, | |
63 | sb->patch_version); | |
64 | if (sb->minor_version >= 90) | |
65 | printf(" UUID : %08x:%08x:%08x:%08x\n", sb->set_uuid0, sb->set_uuid1, | |
66 | sb->set_uuid2, sb->set_uuid3); | |
67 | else | |
68 | printf(" UUID : %08x\n", sb->set_uuid0); | |
69 | ||
70 | atime = sb->ctime; | |
71 | printf(" Creation Time : %.24s\n", ctime(&atime)); | |
72 | c=map_num(pers, sb->level); | |
73 | printf(" Raid Level : %s\n", c?c:"-unknown-"); | |
74 | if ((int)sb->level >= 0) | |
75 | printf(" Device Size : %d%s\n", sb->size, human_size((long long)sb->size<<10)); | |
76 | printf(" Raid Devices : %d\n", sb->raid_disks); | |
77 | printf(" Total Devices : %d\n", sb->nr_disks); | |
78 | printf("Preferred Minor : %d\n", sb->md_minor); | |
79 | printf("\n"); | |
80 | atime = sb->utime; | |
81 | printf(" Update Time : %.24s\n", ctime(&atime)); | |
82 | printf(" State : %s\n", | |
83 | (sb->state&(1<<MD_SB_CLEAN))?"clean":"active"); | |
84 | printf(" Active Devices : %d\n", sb->active_disks); | |
85 | printf("Working Devices : %d\n", sb->working_disks); | |
86 | printf(" Failed Devices : %d\n", sb->failed_disks); | |
87 | printf(" Spare Devices : %d\n", sb->spare_disks); | |
88 | if (calc_sb0_csum(sb) == sb->sb_csum) | |
89 | printf(" Checksum : %x - correct\n", sb->sb_csum); | |
90 | else | |
91 | printf(" Checksum : %x - expected %lx\n", sb->sb_csum, calc_sb0_csum(sb)); | |
92 | printf(" Events : %d.%d\n", sb->events_hi, sb->events_lo); | |
93 | printf("\n"); | |
94 | if (sb->level == 5) { | |
95 | c = map_num(r5layout, sb->layout); | |
96 | printf(" Layout : %s\n", c?c:"-unknown-"); | |
97 | } | |
98 | switch(sb->level) { | |
99 | case 0: | |
100 | case 4: | |
101 | case 5: | |
102 | printf(" Chunk Size : %dK\n", sb->chunk_size/1024); | |
103 | break; | |
104 | case -1: | |
105 | printf(" Rounding : %dK\n", sb->chunk_size/1024); | |
106 | break; | |
107 | default: break; | |
108 | } | |
109 | printf("\n"); | |
110 | printf(" Number Major Minor RaidDevice State\n"); | |
111 | for (d= -1; d<(signed int)(sb->raid_disks+sb->spare_disks); d++) { | |
112 | mdp_disk_t *dp; | |
113 | char *dv; | |
114 | char nb[5]; | |
115 | if (d>=0) dp = &sb->disks[d]; | |
116 | else dp = &sb->this_disk; | |
117 | sprintf(nb, "%4d", d); | |
118 | printf("%4s %5d %5d %5d %5d ", d < 0 ? "this" : nb, | |
119 | dp->number, dp->major, dp->minor, dp->raid_disk); | |
120 | if (dp->state & (1<<MD_DISK_FAULTY)) printf(" faulty"); | |
121 | if (dp->state & (1<<MD_DISK_ACTIVE)) printf(" active"); | |
122 | if (dp->state & (1<<MD_DISK_SYNC)) printf(" sync"); | |
123 | if (dp->state & (1<<MD_DISK_REMOVED)) printf(" removed"); | |
124 | if (dp->state == 0) printf(" spare"); | |
125 | if ((dv=map_dev(dp->major, dp->minor))) | |
126 | printf(" %s", dv); | |
127 | printf("\n"); | |
128 | if (d == -1) printf("\n"); | |
129 | } | |
130 | } | |
131 | ||
132 | void brief_examine_super0(void *sbv) | |
133 | { | |
134 | mdp_super_t *sb = sbv; | |
135 | char *c=map_num(pers, sb->level); | |
136 | ||
137 | printf("ARRAY %s level=%s num-devices=%d UUID=", | |
138 | get_md_name(sb->md_minor), | |
139 | c?c:"-unknown-", sb->raid_disks); | |
140 | if (sb->minor_version >= 90) | |
141 | printf("%08x:%08x:%08x:%08x", sb->set_uuid0, sb->set_uuid1, | |
142 | sb->set_uuid2, sb->set_uuid3); | |
143 | else | |
144 | printf("%08x", sb->set_uuid0); | |
145 | printf("\n"); | |
146 | } | |
147 | ||
148 | void detail_super0(void *sbv) | |
149 | { | |
150 | mdp_super_t *sb = sbv; | |
151 | printf(" UUID : "); | |
152 | if (sb->minor_version >= 90) | |
153 | printf("%08x:%08x:%08x:%08x", sb->set_uuid0, sb->set_uuid1, | |
154 | sb->set_uuid2, sb->set_uuid3); | |
155 | else | |
156 | printf("%08x", sb->set_uuid0); | |
157 | printf("\n Events : %d.%d\n\n", sb->events_hi, sb->events_lo); | |
158 | } | |
159 | ||
160 | void brief_detail_super0(void *sbv) | |
161 | { | |
162 | mdp_super_t *sb = sbv; | |
163 | printf(" UUID="); | |
164 | if (sb->minor_version >= 90) | |
165 | printf("%08x:%08x:%08x:%08x", sb->set_uuid0, sb->set_uuid1, | |
166 | sb->set_uuid2, sb->set_uuid3); | |
167 | else | |
168 | printf("%08x", sb->set_uuid0); | |
169 | } | |
170 | ||
171 | void uuid_from_super0(int uuid[4], void * sbv) | |
172 | { | |
173 | mdp_super_t *super = sbv; | |
174 | uuid[0] = super->set_uuid0; | |
175 | if (super->minor_version >= 90) { | |
176 | uuid[1] = super->set_uuid1; | |
177 | uuid[2] = super->set_uuid2; | |
178 | uuid[3] = super->set_uuid3; | |
179 | } else { | |
180 | uuid[1] = 0; | |
181 | uuid[2] = 0; | |
182 | uuid[3] = 0; | |
183 | } | |
184 | } | |
185 | ||
186 | void getinfo_super0(struct mdinfo *info, void *sbv) | |
187 | { | |
188 | mdp_super_t *sb = sbv; | |
189 | int working = 0; | |
190 | int i; | |
191 | ||
192 | info->array.major_version = sb->major_version; | |
193 | info->array.minor_version = sb->minor_version; | |
194 | info->array.patch_version = sb->patch_version; | |
195 | info->array.raid_disks = sb->raid_disks; | |
196 | info->array.level = sb->level; | |
197 | info->array.md_minor = sb->md_minor; | |
198 | info->array.ctime = sb->ctime; | |
199 | ||
200 | info->disk.state = sb->this_disk.state; | |
201 | info->disk.major = sb->this_disk.major; | |
202 | info->disk.minor = sb->this_disk.minor; | |
203 | info->disk.raid_disk = sb->this_disk.raid_disk; | |
204 | ||
205 | info->events = md_event(sb); | |
206 | ||
207 | uuid_from_super0(info->uuid, sbv); | |
208 | ||
209 | /* work_disks is calculated rather than read directly */ | |
210 | for (i=0; i < MD_SB_DISKS; i++) | |
211 | if ((sb->disks[i].state & (1<<MD_DISK_SYNC)) && | |
212 | (sb->disks[i].state & (1<<MD_DISK_ACTIVE)) && | |
213 | !(sb->disks[i].state & (1<<MD_DISK_FAULTY))) | |
214 | working ++; | |
215 | info->array.working_disks = working; | |
216 | } | |
217 | ||
218 | ||
219 | int update_super0(struct mdinfo *info, void *sbv, char *update, char *devname, int verbose) | |
220 | { | |
221 | int rv = 0; | |
222 | mdp_super_t *sb = sbv; | |
223 | if (strcmp(update, "sparc2.2")==0 ) { | |
224 | /* 2.2 sparc put the events in the wrong place | |
225 | * So we copy the tail of the superblock | |
226 | * up 4 bytes before continuing | |
227 | */ | |
228 | __u32 *sb32 = (__u32*)sb; | |
229 | memcpy(sb32+MD_SB_GENERIC_CONSTANT_WORDS+7, | |
230 | sb32+MD_SB_GENERIC_CONSTANT_WORDS+7+1, | |
231 | (MD_SB_WORDS - (MD_SB_GENERIC_CONSTANT_WORDS+7+1))*4); | |
232 | fprintf (stderr, Name ": adjusting superblock of %s for 2.2/sparc compatability.\n", | |
233 | devname); | |
234 | } | |
235 | if (strcmp(update, "super-minor") ==0) { | |
236 | sb->md_minor = info->array.md_minor; | |
237 | if (verbose) | |
238 | fprintf(stderr, Name ": updating superblock of %s with minor number %d\n", | |
239 | devname, info->array.md_minor); | |
240 | } | |
241 | if (strcmp(update, "summaries") == 0) { | |
242 | int i; | |
243 | /* set nr_disks, active_disks, working_disks, | |
244 | * failed_disks, spare_disks based on disks[] | |
245 | * array in superblock. | |
246 | * Also make sure extra slots aren't 'failed' | |
247 | */ | |
248 | sb->nr_disks = sb->active_disks = | |
249 | sb->working_disks = sb->failed_disks = | |
250 | sb->spare_disks = 0; | |
251 | for (i=0; i < MD_SB_DISKS ; i++) | |
252 | if (sb->disks[i].major || | |
253 | sb->disks[i].minor) { | |
254 | int state = sb->disks[i].state; | |
255 | if (state & (1<<MD_DISK_REMOVED)) | |
256 | continue; | |
257 | sb->nr_disks++; | |
258 | if (state & (1<<MD_DISK_ACTIVE)) | |
259 | sb->active_disks++; | |
260 | if (state & (1<<MD_DISK_FAULTY)) | |
261 | sb->failed_disks++; | |
262 | else | |
263 | sb->working_disks++; | |
264 | if (state == 0) | |
265 | sb->spare_disks++; | |
266 | } else if (i >= sb->raid_disks && sb->disks[i].number == 0) | |
267 | sb->disks[i].state = 0; | |
268 | } | |
269 | if (strcmp(update, "force")==0) { | |
270 | sb->events_hi = (info->events>>32) & 0xFFFFFFFF; | |
271 | sb->events_lo = (info->events) & 0xFFFFFFFF; | |
272 | if (sb->level == 5 || sb->level == 4 || sb->level == 6) | |
273 | /* need to force clean */ | |
274 | sb->state |= (1 << MD_SB_CLEAN); | |
275 | } | |
276 | if (strcmp(update, "assemble")==0) { | |
277 | int d = info->disk.number; | |
278 | if (sb->disks[d].state != info->disk.state) { | |
279 | sb->disks[d].state = info->disk.state; | |
280 | rv = 1; | |
281 | } | |
282 | } | |
283 | if (strcmp(update, "newdev") == 0) { | |
284 | int d = info->disk.number; | |
285 | memset(&sb->disks[d], 0, sizeof(sb->disks[d])); | |
286 | sb->disks[d].number = d; | |
287 | sb->disks[d].major = info->disk.major; | |
288 | sb->disks[d].minor = info->disk.minor; | |
289 | sb->disks[d].raid_disk = info->disk.raid_disk; | |
290 | sb->disks[d].state = info->disk.state; | |
291 | sb->this_disk = sb->disks[d]; | |
292 | } | |
293 | if (strcmp(update, "grow") == 0) { | |
294 | sb->raid_disks = info->array.raid_disks; | |
295 | sb->nr_disks = info->array.nr_disks; | |
296 | sb->active_disks = info->array.active_disks; | |
297 | sb->working_disks = info->array.working_disks; | |
298 | memset(&sb->disks[info->disk.number], 0, sizeof(sb->disks[0])); | |
299 | sb->disks[info->disk.number].number = info->disk.number; | |
300 | sb->disks[info->disk.number].major = info->disk.major; | |
301 | sb->disks[info->disk.number].minor = info->disk.minor; | |
302 | sb->disks[info->disk.number].raid_disk = info->disk.raid_disk; | |
303 | sb->disks[info->disk.number].state = info->disk.state; | |
304 | if (sb->this_disk.number == info->disk.number) | |
305 | sb->this_disk = sb->disks[info->disk.number]; | |
306 | } | |
307 | if (strcmp(update, "resync") == 0) { | |
308 | /* make sure resync happens */ | |
309 | sb->state &= ~(1<<MD_SB_CLEAN); | |
310 | sb->recovery_cp = 0; | |
311 | } | |
312 | ||
313 | sb->sb_csum = calc_sb0_csum(sb); | |
314 | return rv; | |
315 | } | |
316 | ||
317 | __u64 event_super0(void *sbv) | |
318 | { | |
319 | mdp_super_t *sb = sbv; | |
320 | return md_event(sb); | |
321 | } | |
322 | ||
323 | ||
324 | ||
325 | void init_super0(void **sbp, mdu_array_info_t *info) | |
326 | { | |
327 | mdp_super_t *sb = malloc(MD_SB_BYTES); | |
328 | memset(sb, 0, MD_SB_BYTES); | |
329 | ||
330 | sb->md_magic = MD_SB_MAGIC; | |
331 | sb->major_version = 0; | |
332 | sb->minor_version = 90; | |
333 | sb->patch_version = 0; | |
334 | sb->gvalid_words = 0; /* ignored */ | |
335 | sb->set_uuid0 = random(); | |
336 | sb->ctime = time(0); | |
337 | sb->level = info->level; | |
338 | sb->size = info->size; | |
339 | sb->nr_disks = info->nr_disks; | |
340 | sb->raid_disks = info->raid_disks; | |
341 | sb->md_minor = info->md_minor; | |
342 | sb->not_persistent = 0; | |
343 | sb->set_uuid1 = random(); | |
344 | sb->set_uuid2 = random(); | |
345 | sb->set_uuid3 = random(); | |
346 | ||
347 | sb->utime = sb->ctime; | |
348 | sb->state = info->state; | |
349 | sb->active_disks = info->active_disks; | |
350 | sb->working_disks = info->working_disks; | |
351 | sb->failed_disks = info->failed_disks; | |
352 | sb->events_hi = 0; | |
353 | sb->events_lo = 1; | |
354 | ||
355 | sb->layout = info->layout; | |
356 | sb->chunk_size = info->chunk_size; | |
357 | ||
358 | *sbp = sb; | |
359 | } | |
360 | ||
361 | /* Add a device to the superblock being created */ | |
362 | void add_to_super0(void *sbv, mdu_disk_info_t *dinfo) | |
363 | { | |
364 | mdp_super_t *sb = sbv; | |
365 | mdp_disk_t *dk = &sb->disks[dinfo->number]; | |
366 | ||
367 | dk->number = dinfo->number; | |
368 | dk->major = dinfo->major; | |
369 | dk->minor = dinfo->minor; | |
370 | dk->raid_disk = dinfo->raid_disk; | |
371 | dk->state = dinfo->state; | |
372 | } | |
373 | ||
374 | int store_super0(int fd, mdp_super_t *super) | |
375 | { | |
376 | unsigned long size; | |
377 | unsigned long long dsize; | |
378 | unsigned long long offset; | |
379 | ||
380 | #ifdef BLKGETSIZE64 | |
381 | if (ioctl(fd, BLKGETSIZE64, &dsize) != 0) | |
382 | #endif | |
383 | { | |
384 | if (ioctl(fd, BLKGETSIZE, &size)) | |
385 | return 1; | |
386 | else | |
387 | dsize = ((unsigned long long)size)<<9; | |
388 | } | |
389 | ||
390 | if (dsize < MD_RESERVED_SECTORS*2) | |
391 | return 2; | |
392 | ||
393 | offset = MD_NEW_SIZE_SECTORS(dsize>>9); | |
394 | ||
395 | offset *= 512; | |
396 | ||
397 | if (lseek64(fd, offset, 0)< 0LL) | |
398 | return 3; | |
399 | ||
400 | if (write(fd, super, sizeof(*super)) != sizeof(*super)) | |
401 | return 4; | |
402 | ||
403 | return 0; | |
404 | } | |
405 | ||
406 | int write_init_super0(void *sbv, mdu_disk_info_t *dinfo, char *devname) | |
407 | { | |
408 | mdp_super_t *sb = sbv; | |
409 | int fd = open(devname, O_RDWR, O_EXCL); | |
410 | int rv; | |
411 | ||
412 | if (fd < 0) { | |
413 | fprintf(stderr, Name ": Failed to open %s to write superblock\n", devname); | |
414 | return -1; | |
415 | } | |
416 | ||
417 | sb->this_disk = sb->disks[dinfo->number]; | |
418 | sb->sb_csum = calc_sb0_csum(sb); | |
419 | rv = store_super0(fd, sb); | |
420 | close(fd); | |
421 | if (rv) | |
422 | fprintf(stderr, Name ": failed to write superblock to %s\n", devname); | |
423 | return rv; | |
424 | } | |
425 | ||
426 | int compare_super0(void **firstp, void *secondv) | |
427 | { | |
428 | /* | |
429 | * return: | |
430 | * 0 same, or first was empty, and second was copied | |
431 | * 1 second had wrong number | |
432 | * 2 wrong uuid | |
433 | * 3 wrong other info | |
434 | */ | |
435 | mdp_super_t *first = *firstp; | |
436 | mdp_super_t *second = secondv; | |
437 | ||
438 | int uuid1[4], uuid2[4]; | |
439 | if (second->md_magic != MD_SB_MAGIC) | |
440 | return 1; | |
441 | if (!first) { | |
442 | first = malloc(MD_SB_BYTES); | |
443 | memcpy(first, second, MD_SB_BYTES); | |
444 | *firstp = first; | |
445 | return 0; | |
446 | } | |
447 | ||
448 | uuid_from_super0(uuid1, first); | |
449 | uuid_from_super0(uuid2, second); | |
450 | if (!same_uuid(uuid1, uuid2)) | |
451 | return 2; | |
452 | if (first->major_version != second->major_version || | |
453 | first->minor_version != second->minor_version || | |
454 | first->patch_version != second->patch_version || | |
455 | first->gvalid_words != second->gvalid_words || | |
456 | first->ctime != second->ctime || | |
457 | first->level != second->level || | |
458 | first->size != second->size || | |
459 | first->raid_disks != second->raid_disks ) | |
460 | return 3; | |
461 | ||
462 | return 0; | |
463 | } | |
464 | ||
465 | ||
466 | int load_super0(int fd, void **sbp, char *devname) | |
467 | { | |
468 | /* try to read in the superblock | |
469 | * Return: | |
470 | * 0 on success | |
471 | * 1 on cannot get superblock | |
472 | * 2 on superblock meaningless | |
473 | */ | |
474 | unsigned long size; | |
475 | unsigned long long dsize; | |
476 | unsigned long long offset; | |
477 | mdp_super_t *super; | |
478 | ||
479 | #ifdef BLKGETSIZE64 | |
480 | if (ioctl(fd, BLKGETSIZE64, &dsize) != 0) | |
481 | #endif | |
482 | { | |
483 | if (ioctl(fd, BLKGETSIZE, &size)) { | |
484 | if (devname) | |
485 | fprintf(stderr, Name ": cannot find device size for %s: %s\n", | |
486 | devname, strerror(errno)); | |
487 | return 1; | |
488 | } else | |
489 | dsize = size << 9; | |
490 | } | |
491 | ||
492 | if (dsize < MD_RESERVED_SECTORS*2) { | |
493 | if (devname) | |
494 | fprintf(stderr, Name ": %s is too small for md: size is %ld sectors.\n", | |
495 | devname, size); | |
496 | return 1; | |
497 | } | |
498 | ||
499 | offset = MD_NEW_SIZE_SECTORS(dsize>>9); | |
500 | ||
501 | offset *= 512; | |
502 | ||
503 | ioctl(fd, BLKFLSBUF, 0); /* make sure we read current data */ | |
504 | ||
505 | if (lseek64(fd, offset, 0)< 0LL) { | |
506 | if (devname) | |
507 | fprintf(stderr, Name ": Cannot seek to superblock on %s: %s\n", | |
508 | devname, strerror(errno)); | |
509 | return 1; | |
510 | } | |
511 | ||
512 | super = malloc(MD_SB_BYTES); | |
513 | ||
514 | if (read(fd, super, sizeof(*super)) != MD_SB_BYTES) { | |
515 | if (devname) | |
516 | fprintf(stderr, Name ": Cannot read superblock on %s\n", | |
517 | devname); | |
518 | free(super); | |
519 | return 1; | |
520 | } | |
521 | ||
522 | if (super->md_magic != MD_SB_MAGIC) { | |
523 | if (devname) | |
524 | fprintf(stderr, Name ": No super block found on %s (Expected magic %08x, got %08x)\n", | |
525 | devname, MD_SB_MAGIC, super->md_magic); | |
526 | free(super); | |
527 | return 2; | |
528 | } | |
529 | ||
530 | if (super->major_version != 0) { | |
531 | if (devname) | |
532 | fprintf(stderr, Name ": Cannot interpret superblock on %s - version is %d\n", | |
533 | devname, super->major_version); | |
534 | free(super); | |
535 | return 2; | |
536 | } | |
537 | *sbp = super; | |
538 | return 0; | |
539 | } |