]>
Commit | Line | Data |
---|---|---|
979afcb8 PS |
1 | /* |
2 | * raid6check - extended consistency check for RAID-6 | |
3 | * | |
4 | * Copyright (C) 2011 Piergiorgio Sartor | |
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: Piergiorgio Sartor | |
22 | * Based on "restripe.c" from "mdadm" codebase | |
23 | */ | |
24 | ||
25 | #include "mdadm.h" | |
26 | #include <stdint.h> | |
8d8ab389 PS |
27 | #include <signal.h> |
28 | #include <sys/mman.h> | |
979afcb8 | 29 | |
3b9c9603 PS |
30 | #define CHECK_PAGE_BITS (12) |
31 | #define CHECK_PAGE_SIZE (1 << CHECK_PAGE_BITS) | |
32 | ||
635b5861 BS |
33 | enum repair { |
34 | NO_REPAIR = 0, | |
35 | MANUAL_REPAIR, | |
36 | AUTO_REPAIR | |
37 | }; | |
38 | ||
979afcb8 PS |
39 | int geo_map(int block, unsigned long long stripe, int raid_disks, |
40 | int level, int layout); | |
41 | void qsyndrome(uint8_t *p, uint8_t *q, uint8_t **sources, int disks, int size); | |
42 | void make_tables(void); | |
f2e29ad6 RB |
43 | void ensure_zero_has_size(int chunk_size); |
44 | void raid6_datap_recov(int disks, size_t bytes, int faila, uint8_t **ptrs); | |
45 | void raid6_2data_recov(int disks, size_t bytes, int faila, int failb, | |
46 | uint8_t **ptrs); | |
47 | void xor_blocks(char *target, char **sources, int disks, int size); | |
48 | ||
979afcb8 PS |
49 | /* Collect per stripe consistency information */ |
50 | void raid6_collect(int chunk_size, uint8_t *p, uint8_t *q, | |
51 | char *chunkP, char *chunkQ, int *results) | |
52 | { | |
53 | int i; | |
54 | int data_id; | |
55 | uint8_t Px, Qx; | |
56 | extern uint8_t raid6_gflog[]; | |
57 | ||
58 | for(i = 0; i < chunk_size; i++) { | |
59 | Px = (uint8_t)chunkP[i] ^ (uint8_t)p[i]; | |
60 | Qx = (uint8_t)chunkQ[i] ^ (uint8_t)q[i]; | |
61 | ||
62 | if((Px != 0) && (Qx == 0)) | |
63 | results[i] = -1; | |
64 | ||
65 | if((Px == 0) && (Qx != 0)) | |
66 | results[i] = -2; | |
67 | ||
68 | if((Px != 0) && (Qx != 0)) { | |
69 | data_id = (raid6_gflog[Qx] - raid6_gflog[Px]); | |
70 | if(data_id < 0) data_id += 255; | |
71 | results[i] = data_id; | |
72 | } | |
73 | ||
74 | if((Px == 0) && (Qx == 0)) | |
75 | results[i] = -255; | |
76 | } | |
77 | } | |
78 | ||
3b9c9603 PS |
79 | /* Try to find out if a specific disk has problems in a CHECK_PAGE_SIZE page size */ |
80 | int raid6_stats_blk(int *results, int raid_disks) | |
979afcb8 PS |
81 | { |
82 | int i; | |
83 | int curr_broken_disk = -255; | |
84 | int prev_broken_disk = -255; | |
85 | int broken_status = 0; | |
86 | ||
3b9c9603 | 87 | for(i = 0; i < CHECK_PAGE_SIZE; i++) { |
979afcb8 PS |
88 | |
89 | if(results[i] != -255) | |
90 | curr_broken_disk = results[i]; | |
91 | ||
92 | if(curr_broken_disk >= raid_disks) | |
93 | broken_status = 2; | |
94 | ||
95 | switch(broken_status) { | |
96 | case 0: | |
97 | if(curr_broken_disk != -255) { | |
98 | prev_broken_disk = curr_broken_disk; | |
99 | broken_status = 1; | |
100 | } | |
101 | break; | |
102 | ||
103 | case 1: | |
104 | if(curr_broken_disk != prev_broken_disk) | |
105 | broken_status = 2; | |
106 | break; | |
107 | ||
108 | case 2: | |
109 | default: | |
110 | curr_broken_disk = prev_broken_disk = -65535; | |
111 | break; | |
112 | } | |
113 | } | |
114 | ||
115 | return curr_broken_disk; | |
116 | } | |
117 | ||
3b9c9603 PS |
118 | /* Collect disks status for a strip in CHECK_PAGE_SIZE page size blocks */ |
119 | void raid6_stats(int *disk, int *results, int raid_disks, int chunk_size) | |
120 | { | |
121 | int i, j; | |
122 | ||
123 | for(i = 0, j = 0; i < chunk_size; i += CHECK_PAGE_SIZE, j++) { | |
124 | disk[j] = raid6_stats_blk(&results[i], raid_disks); | |
125 | } | |
126 | } | |
127 | ||
351d7680 RB |
128 | int lock_stripe(struct mdinfo *info, unsigned long long start, |
129 | int chunk_size, int data_disks, sighandler_t *sig) { | |
130 | int rv; | |
131 | if(mlockall(MCL_CURRENT | MCL_FUTURE) != 0) { | |
132 | return 2; | |
133 | } | |
134 | ||
135 | sig[0] = signal(SIGTERM, SIG_IGN); | |
136 | sig[1] = signal(SIGINT, SIG_IGN); | |
137 | sig[2] = signal(SIGQUIT, SIG_IGN); | |
138 | ||
139 | rv = sysfs_set_num(info, NULL, "suspend_lo", start * chunk_size * data_disks); | |
140 | rv |= sysfs_set_num(info, NULL, "suspend_hi", (start + 1) * chunk_size * data_disks); | |
141 | return rv * 256; | |
142 | } | |
143 | ||
144 | int unlock_all_stripes(struct mdinfo *info, sighandler_t *sig) { | |
145 | int rv; | |
146 | rv = sysfs_set_num(info, NULL, "suspend_lo", 0x7FFFFFFFFFFFFFFFULL); | |
147 | rv |= sysfs_set_num(info, NULL, "suspend_hi", 0); | |
148 | rv |= sysfs_set_num(info, NULL, "suspend_lo", 0); | |
149 | ||
150 | signal(SIGQUIT, sig[2]); | |
151 | signal(SIGINT, sig[1]); | |
152 | signal(SIGTERM, sig[0]); | |
153 | ||
154 | if(munlockall() != 0) | |
155 | return 3; | |
156 | return rv * 256; | |
157 | } | |
158 | ||
c22ac3b1 PS |
159 | /* Autorepair */ |
160 | int autorepair(int *disk, int diskP, int diskQ, unsigned long long start, int chunk_size, | |
161 | char *name[], int raid_disks, int data_disks, char **blocks_page, | |
162 | char **blocks, uint8_t *p, char **stripes, int *block_index_for_slot, | |
163 | int *source, unsigned long long *offsets) | |
164 | { | |
165 | int i, j; | |
166 | int pages_to_write_count = 0; | |
167 | int page_to_write[chunk_size >> CHECK_PAGE_BITS]; | |
168 | for(j = 0; j < (chunk_size >> CHECK_PAGE_BITS); j++) { | |
169 | if (disk[j] >= 0) { | |
170 | printf("Auto-repairing slot %d (%s)\n", disk[j], name[disk[j]]); | |
171 | pages_to_write_count++; | |
172 | page_to_write[j] = 1; | |
173 | for(i = 0; i < raid_disks; i++) { | |
174 | blocks_page[i] = blocks[i] + j * CHECK_PAGE_SIZE; | |
175 | } | |
176 | if (disk[j] == diskQ) { | |
177 | qsyndrome(p, (uint8_t*)stripes[diskQ] + j * CHECK_PAGE_SIZE, (uint8_t**)blocks_page, data_disks, CHECK_PAGE_SIZE); | |
178 | } | |
179 | else { | |
180 | char *all_but_failed_blocks[data_disks]; | |
181 | int failed_block_index = block_index_for_slot[disk[j]]; | |
182 | for(i = 0; i < data_disks; i++) { | |
183 | if (failed_block_index == i) { | |
184 | all_but_failed_blocks[i] = stripes[diskP] + j * CHECK_PAGE_SIZE; | |
185 | } | |
186 | else { | |
187 | all_but_failed_blocks[i] = blocks_page[i]; | |
188 | } | |
189 | } | |
190 | xor_blocks(stripes[disk[j]] + j * CHECK_PAGE_SIZE, | |
191 | all_but_failed_blocks, data_disks, CHECK_PAGE_SIZE); | |
192 | } | |
193 | } | |
194 | else { | |
195 | page_to_write[j] = 0; | |
196 | } | |
197 | } | |
198 | ||
199 | if(pages_to_write_count > 0) { | |
200 | int write_res = 0; | |
201 | for(j = 0; j < (chunk_size >> CHECK_PAGE_BITS); j++) { | |
202 | if(page_to_write[j] == 1) { | |
203 | lseek64(source[disk[j]], offsets[disk[j]] + start * chunk_size + j * CHECK_PAGE_SIZE, SEEK_SET); | |
204 | write_res += write(source[disk[j]], stripes[disk[j]] + j * CHECK_PAGE_SIZE, CHECK_PAGE_SIZE); | |
205 | } | |
206 | } | |
207 | ||
208 | if (write_res != (CHECK_PAGE_SIZE * pages_to_write_count)) { | |
209 | fprintf(stderr, "Failed to write a full chunk.\n"); | |
210 | return -1; | |
211 | } | |
212 | } | |
213 | ||
214 | return 0; | |
215 | } | |
216 | ||
497b6d6b PS |
217 | /* Manual repair */ |
218 | int manual_repair(int diskP, int diskQ, int chunk_size, int raid_disks, int data_disks, | |
219 | int failed_disk1, int failed_disk2, unsigned long long start, int *block_index_for_slot, | |
220 | char *name[], char **stripes, char **blocks, uint8_t *p, struct mdinfo *info, sighandler_t *sig, | |
221 | int *source, unsigned long long *offsets) | |
222 | { | |
223 | int err = 0; | |
224 | int i; | |
225 | printf("Repairing stripe %llu\n", start); | |
226 | printf("Assuming slots %d (%s) and %d (%s) are incorrect\n", | |
227 | failed_disk1, name[failed_disk1], | |
228 | failed_disk2, name[failed_disk2]); | |
229 | ||
230 | if (failed_disk1 == diskQ || failed_disk2 == diskQ) { | |
231 | char *all_but_failed_blocks[data_disks]; | |
232 | int failed_data_or_p; | |
233 | int failed_block_index; | |
234 | ||
235 | if (failed_disk1 == diskQ) { | |
236 | failed_data_or_p = failed_disk2; | |
237 | } | |
238 | else { | |
239 | failed_data_or_p = failed_disk1; | |
240 | } | |
241 | printf("Repairing D/P(%d) and Q\n", failed_data_or_p); | |
242 | failed_block_index = block_index_for_slot[failed_data_or_p]; | |
243 | for (i = 0; i < data_disks; i++) { | |
244 | if (failed_block_index == i) { | |
245 | all_but_failed_blocks[i] = stripes[diskP]; | |
246 | } | |
247 | else { | |
248 | all_but_failed_blocks[i] = blocks[i]; | |
249 | } | |
250 | } | |
251 | xor_blocks(stripes[failed_data_or_p], | |
252 | all_but_failed_blocks, data_disks, chunk_size); | |
253 | qsyndrome(p, (uint8_t*)stripes[diskQ], (uint8_t**)blocks, data_disks, chunk_size); | |
254 | } | |
255 | else { | |
256 | ensure_zero_has_size(chunk_size); | |
257 | if (failed_disk1 == diskP || failed_disk2 == diskP) { | |
258 | int failed_data, failed_block_index; | |
259 | if (failed_disk1 == diskP) { | |
260 | failed_data = failed_disk2; | |
261 | } | |
262 | else { | |
263 | failed_data = failed_disk1; | |
264 | } | |
265 | failed_block_index = block_index_for_slot[failed_data]; | |
266 | printf("Repairing D(%d) and P\n", failed_data); | |
267 | raid6_datap_recov(raid_disks, chunk_size, failed_block_index, (uint8_t**)blocks); | |
268 | } | |
269 | else { | |
270 | printf("Repairing D and D\n"); | |
271 | int failed_block_index1 = block_index_for_slot[failed_disk1]; | |
272 | int failed_block_index2 = block_index_for_slot[failed_disk2]; | |
273 | if (failed_block_index1 > failed_block_index2) { | |
274 | int t = failed_block_index1; | |
275 | failed_block_index1 = failed_block_index2; | |
276 | failed_block_index2 = t; | |
277 | } | |
278 | raid6_2data_recov(raid_disks, chunk_size, failed_block_index1, failed_block_index2, (uint8_t**)blocks); | |
279 | } | |
280 | } | |
281 | ||
282 | err = lock_stripe(info, start, chunk_size, data_disks, sig); | |
283 | if(err != 0) { | |
284 | if (err != 2) { | |
285 | return -1; | |
286 | } | |
287 | return -2;; | |
288 | } | |
289 | ||
290 | int write_res1, write_res2; | |
291 | off64_t seek_res; | |
292 | ||
293 | seek_res = lseek64(source[failed_disk1], | |
294 | offsets[failed_disk1] + start * chunk_size, SEEK_SET); | |
295 | if (seek_res < 0) { | |
296 | fprintf(stderr, "lseek failed for failed_disk1\n"); | |
297 | return -1; | |
298 | } | |
299 | write_res1 = write(source[failed_disk1], stripes[failed_disk1], chunk_size); | |
300 | ||
301 | seek_res = lseek64(source[failed_disk2], | |
302 | offsets[failed_disk2] + start * chunk_size, SEEK_SET); | |
303 | if (seek_res < 0) { | |
304 | fprintf(stderr, "lseek failed for failed_disk1\n"); | |
305 | return -1; | |
306 | } | |
307 | write_res2 = write(source[failed_disk2], stripes[failed_disk2], chunk_size); | |
308 | ||
309 | err = unlock_all_stripes(info, sig); | |
310 | if(err != 0) { | |
311 | return -2; | |
312 | } | |
313 | ||
314 | if (write_res1 != chunk_size || write_res2 != chunk_size) { | |
315 | fprintf(stderr, "Failed to write a complete chunk.\n"); | |
316 | return -2; | |
317 | } | |
318 | ||
319 | return 0; | |
320 | } | |
321 | ||
8d8ab389 | 322 | int check_stripes(struct mdinfo *info, int *source, unsigned long long *offsets, |
979afcb8 | 323 | int raid_disks, int chunk_size, int level, int layout, |
f2e29ad6 | 324 | unsigned long long start, unsigned long long length, char *name[], |
635b5861 | 325 | enum repair repair, int failed_disk1, int failed_disk2) |
979afcb8 PS |
326 | { |
327 | /* read the data and p and q blocks, and check we got them right */ | |
503975b9 N |
328 | char *stripe_buf = xmalloc(raid_disks * chunk_size); |
329 | char **stripes = xmalloc(raid_disks * sizeof(char*)); | |
330 | char **blocks = xmalloc(raid_disks * sizeof(char*)); | |
3b9c9603 | 331 | char **blocks_page = xmalloc(raid_disks * sizeof(char*)); |
696e95a1 | 332 | int *block_index_for_slot = xmalloc(raid_disks * sizeof(int)); |
503975b9 N |
333 | uint8_t *p = xmalloc(chunk_size); |
334 | uint8_t *q = xmalloc(chunk_size); | |
335 | int *results = xmalloc(chunk_size * sizeof(int)); | |
351d7680 | 336 | sighandler_t *sig = xmalloc(3 * sizeof(sighandler_t)); |
979afcb8 | 337 | |
3b9c9603 | 338 | int i, j; |
979afcb8 PS |
339 | int diskP, diskQ; |
340 | int data_disks = raid_disks - 2; | |
af3c3750 | 341 | int err = 0; |
979afcb8 PS |
342 | |
343 | extern int tables_ready; | |
344 | ||
345 | if (!tables_ready) | |
346 | make_tables(); | |
347 | ||
348 | for ( i = 0 ; i < raid_disks ; i++) | |
349 | stripes[i] = stripe_buf + i * chunk_size; | |
350 | ||
351 | while (length > 0) { | |
3b9c9603 | 352 | int disk[chunk_size >> CHECK_PAGE_BITS]; |
979afcb8 | 353 | |
351d7680 RB |
354 | err = lock_stripe(info, start, chunk_size, data_disks, sig); |
355 | if(err != 0) { | |
356 | if (err != 2) | |
357 | unlock_all_stripes(info, sig); | |
8d8ab389 PS |
358 | goto exitCheck; |
359 | } | |
979afcb8 | 360 | for (i = 0 ; i < raid_disks ; i++) { |
2161adce BS |
361 | off64_t seek_res = lseek64(source[i], offsets[i] + start * chunk_size, |
362 | SEEK_SET); | |
363 | if (seek_res < 0) { | |
364 | fprintf(stderr, "lseek to source %d failed\n", i); | |
365 | unlock_all_stripes(info, sig); | |
366 | err = -1; | |
367 | goto exitCheck; | |
368 | } | |
2c7b668d BS |
369 | int read_res = read(source[i], stripes[i], chunk_size); |
370 | if (read_res < chunk_size) { | |
371 | fprintf(stderr, "Failed to read complete chunk disk %d, aborting\n", i); | |
372 | unlock_all_stripes(info, sig); | |
2161adce | 373 | err = -1; |
2c7b668d BS |
374 | goto exitCheck; |
375 | } | |
979afcb8 | 376 | } |
8d8ab389 | 377 | |
979afcb8 | 378 | for (i = 0 ; i < data_disks ; i++) { |
af3c3750 | 379 | int disk = geo_map(i, start, raid_disks, level, layout); |
979afcb8 | 380 | blocks[i] = stripes[disk]; |
696e95a1 | 381 | block_index_for_slot[disk] = i; |
979afcb8 PS |
382 | } |
383 | ||
384 | qsyndrome(p, q, (uint8_t**)blocks, data_disks, chunk_size); | |
af3c3750 | 385 | diskP = geo_map(-1, start, raid_disks, level, layout); |
f2e29ad6 RB |
386 | diskQ = geo_map(-2, start, raid_disks, level, layout); |
387 | blocks[data_disks] = stripes[diskP]; | |
696e95a1 | 388 | block_index_for_slot[diskP] = data_disks; |
f2e29ad6 | 389 | blocks[data_disks+1] = stripes[diskQ]; |
696e95a1 | 390 | block_index_for_slot[diskQ] = data_disks+1; |
237e40ce | 391 | |
af3c3750 | 392 | raid6_collect(chunk_size, p, q, stripes[diskP], stripes[diskQ], results); |
3b9c9603 | 393 | raid6_stats(disk, results, raid_disks, chunk_size); |
979afcb8 | 394 | |
3b9c9603 PS |
395 | for(j = 0; j < (chunk_size >> CHECK_PAGE_BITS); j++) { |
396 | if(disk[j] >= -2) { | |
397 | disk[j] = geo_map(disk[j], start, raid_disks, level, layout); | |
398 | } | |
399 | if(disk[j] >= 0) { | |
afc755e9 | 400 | printf("Error detected at stripe %llu, page %d: possible failed disk slot: %d --> %s\n", |
3b9c9603 PS |
401 | start, j, disk[j], name[disk[j]]); |
402 | } | |
403 | if(disk[j] == -65535) { | |
afc755e9 | 404 | printf("Error detected at stripe %llu, page %d: disk slot unknown\n", start, j); |
3b9c9603 | 405 | } |
979afcb8 | 406 | } |
3b9c9603 | 407 | |
3cfd0297 | 408 | if(repair == AUTO_REPAIR) { |
c22ac3b1 PS |
409 | err = autorepair(disk, diskP, diskQ, start, chunk_size, |
410 | name, raid_disks, data_disks, blocks_page, | |
411 | blocks, p, stripes, block_index_for_slot, | |
412 | source, offsets); | |
413 | if(err != 0) { | |
414 | unlock_all_stripes(info, sig); | |
415 | goto exitCheck; | |
3cfd0297 PS |
416 | } |
417 | } | |
418 | ||
419 | err = unlock_all_stripes(info, sig); | |
420 | if(err != 0) { | |
421 | goto exitCheck; | |
422 | } | |
423 | ||
635b5861 | 424 | if(repair == MANUAL_REPAIR) { |
497b6d6b PS |
425 | err = manual_repair(diskP, diskQ, chunk_size, raid_disks, data_disks, |
426 | failed_disk1, failed_disk2, start, block_index_for_slot, | |
427 | name, stripes, blocks, p, info, sig, | |
428 | source, offsets); | |
429 | if(err == -1) { | |
2161adce | 430 | unlock_all_stripes(info, sig); |
2161adce BS |
431 | goto exitCheck; |
432 | } | |
3b9c9603 PS |
433 | } |
434 | ||
af3c3750 PS |
435 | length--; |
436 | start++; | |
979afcb8 PS |
437 | } |
438 | ||
af3c3750 PS |
439 | exitCheck: |
440 | ||
979afcb8 PS |
441 | free(stripe_buf); |
442 | free(stripes); | |
443 | free(blocks); | |
3b9c9603 | 444 | free(blocks_page); |
3a89d754 | 445 | free(block_index_for_slot); |
979afcb8 PS |
446 | free(p); |
447 | free(q); | |
448 | free(results); | |
3a89d754 | 449 | free(sig); |
979afcb8 | 450 | |
af3c3750 | 451 | return err; |
979afcb8 PS |
452 | } |
453 | ||
454 | unsigned long long getnum(char *str, char **err) | |
455 | { | |
456 | char *e; | |
457 | unsigned long long rv = strtoull(str, &e, 10); | |
458 | if (e==str || *e) { | |
459 | *err = str; | |
460 | return 0; | |
461 | } | |
462 | return rv; | |
463 | } | |
464 | ||
465 | int main(int argc, char *argv[]) | |
466 | { | |
a9c2c6c6 | 467 | /* md_device start length */ |
af3c3750 PS |
468 | int *fds = NULL; |
469 | char *buf = NULL; | |
470 | char **disk_name = NULL; | |
471 | unsigned long long *offsets = NULL; | |
472 | int raid_disks = 0; | |
2cf31121 | 473 | int active_disks; |
af3c3750 PS |
474 | int chunk_size = 0; |
475 | int layout = -1; | |
979afcb8 | 476 | int level = 6; |
635b5861 | 477 | enum repair repair = NO_REPAIR; |
2c7b668d BS |
478 | int failed_disk1 = -1; |
479 | int failed_disk2 = -1; | |
979afcb8 PS |
480 | unsigned long long start, length; |
481 | int i; | |
a9c2c6c6 | 482 | int mdfd; |
8d8ab389 | 483 | struct mdinfo *info = NULL, *comp = NULL; |
979afcb8 | 484 | char *err = NULL; |
af3c3750 PS |
485 | int exit_err = 0; |
486 | int close_flag = 0; | |
487 | char *prg = strrchr(argv[0], '/'); | |
488 | ||
489 | if (prg == NULL) | |
490 | prg = argv[0]; | |
491 | else | |
492 | prg++; | |
493 | ||
494 | if (argc < 4) { | |
8a63c731 | 495 | fprintf(stderr, "Usage: %s md_device start_stripe length_stripes [autorepair]\n", prg); |
f2e29ad6 | 496 | fprintf(stderr, " or: %s md_device repair stripe failed_slot_1 failed_slot_2\n", prg); |
af3c3750 PS |
497 | exit_err = 1; |
498 | goto exitHere; | |
979afcb8 PS |
499 | } |
500 | ||
a9c2c6c6 PS |
501 | mdfd = open(argv[1], O_RDONLY); |
502 | if(mdfd < 0) { | |
503 | perror(argv[1]); | |
e7b84f9d | 504 | fprintf(stderr, "%s: cannot open %s\n", prg, argv[1]); |
af3c3750 PS |
505 | exit_err = 2; |
506 | goto exitHere; | |
a9c2c6c6 PS |
507 | } |
508 | ||
f8fcf7a1 | 509 | info = sysfs_read(mdfd, NULL, |
a9c2c6c6 PS |
510 | GET_LEVEL| |
511 | GET_LAYOUT| | |
512 | GET_DISKS| | |
2cf31121 | 513 | GET_DEGRADED | |
a9c2c6c6 PS |
514 | GET_COMPONENT| |
515 | GET_CHUNK| | |
516 | GET_DEVS| | |
517 | GET_OFFSET| | |
518 | GET_SIZE); | |
519 | ||
8d8ab389 PS |
520 | if(info == NULL) { |
521 | fprintf(stderr, "%s: Error reading sysfs information of %s\n", prg, argv[1]); | |
522 | exit_err = 9; | |
523 | goto exitHere; | |
524 | } | |
525 | ||
a9c2c6c6 PS |
526 | if(info->array.level != level) { |
527 | fprintf(stderr, "%s: %s not a RAID-6\n", prg, argv[1]); | |
af3c3750 PS |
528 | exit_err = 3; |
529 | goto exitHere; | |
a9c2c6c6 PS |
530 | } |
531 | ||
2cf31121 PS |
532 | if(info->array.failed_disks > 0) { |
533 | fprintf(stderr, "%s: %s degraded array\n", prg, argv[1]); | |
534 | exit_err = 8; | |
535 | goto exitHere; | |
536 | } | |
537 | ||
a9c2c6c6 PS |
538 | printf("layout: %d\n", info->array.layout); |
539 | printf("disks: %d\n", info->array.raid_disks); | |
af3c3750 PS |
540 | printf("component size: %llu\n", info->component_size * 512); |
541 | printf("total stripes: %llu\n", (info->component_size * 512) / info->array.chunk_size); | |
a9c2c6c6 PS |
542 | printf("chunk size: %d\n", info->array.chunk_size); |
543 | printf("\n"); | |
544 | ||
545 | comp = info->devs; | |
2cf31121 | 546 | for(i = 0, active_disks = 0; active_disks < info->array.raid_disks; i++) { |
a9c2c6c6 | 547 | printf("disk: %d - offset: %llu - size: %llu - name: %s - slot: %d\n", |
af3c3750 | 548 | i, comp->data_offset * 512, comp->component_size * 512, |
a9c2c6c6 PS |
549 | map_dev(comp->disk.major, comp->disk.minor, 0), |
550 | comp->disk.raid_disk); | |
2cf31121 PS |
551 | if(comp->disk.raid_disk >= 0) |
552 | active_disks++; | |
a9c2c6c6 PS |
553 | comp = comp->next; |
554 | } | |
555 | printf("\n"); | |
556 | ||
557 | close(mdfd); | |
558 | ||
559 | raid_disks = info->array.raid_disks; | |
560 | chunk_size = info->array.chunk_size; | |
561 | layout = info->array.layout; | |
f2e29ad6 RB |
562 | if (strcmp(argv[2], "repair")==0) { |
563 | if (argc < 6) { | |
564 | fprintf(stderr, "For repair mode, call %s md_device repair stripe failed_slot_1 failed_slot_2\n", prg); | |
565 | exit_err = 1; | |
566 | goto exitHere; | |
567 | } | |
635b5861 | 568 | repair = MANUAL_REPAIR; |
f2e29ad6 RB |
569 | start = getnum(argv[3], &err); |
570 | length = 1; | |
571 | failed_disk1 = getnum(argv[4], &err); | |
572 | failed_disk2 = getnum(argv[5], &err); | |
573 | ||
b67e45b8 | 574 | if(failed_disk1 >= info->array.raid_disks) { |
f2e29ad6 RB |
575 | fprintf(stderr, "%s: failed_slot_1 index is higher than number of devices in raid\n", prg); |
576 | exit_err = 4; | |
577 | goto exitHere; | |
578 | } | |
b67e45b8 | 579 | if(failed_disk2 >= info->array.raid_disks) { |
f2e29ad6 RB |
580 | fprintf(stderr, "%s: failed_slot_2 index is higher than number of devices in raid\n", prg); |
581 | exit_err = 4; | |
582 | goto exitHere; | |
583 | } | |
584 | if(failed_disk1 == failed_disk2) { | |
585 | fprintf(stderr, "%s: failed_slot_1 and failed_slot_2 are the same\n", prg); | |
586 | exit_err = 4; | |
587 | goto exitHere; | |
588 | } | |
589 | } | |
590 | else { | |
591 | start = getnum(argv[2], &err); | |
592 | length = getnum(argv[3], &err); | |
8a63c731 | 593 | if (argc >= 5 && strcmp(argv[4], "autorepair")==0) |
635b5861 | 594 | repair = AUTO_REPAIR; |
f2e29ad6 | 595 | } |
a9c2c6c6 | 596 | |
979afcb8 | 597 | if (err) { |
a9c2c6c6 | 598 | fprintf(stderr, "%s: Bad number: %s\n", prg, err); |
af3c3750 PS |
599 | exit_err = 4; |
600 | goto exitHere; | |
979afcb8 | 601 | } |
a9c2c6c6 | 602 | |
af3c3750 PS |
603 | if(start > ((info->component_size * 512) / chunk_size)) { |
604 | start = (info->component_size * 512) / chunk_size; | |
605 | fprintf(stderr, "%s: start beyond disks size\n", prg); | |
606 | } | |
a9c2c6c6 | 607 | |
af3c3750 PS |
608 | if((length == 0) || |
609 | ((length + start) > ((info->component_size * 512) / chunk_size))) { | |
610 | length = (info->component_size * 512) / chunk_size - start; | |
979afcb8 | 611 | } |
a9c2c6c6 | 612 | |
503975b9 N |
613 | disk_name = xmalloc(raid_disks * sizeof(*disk_name)); |
614 | fds = xmalloc(raid_disks * sizeof(*fds)); | |
615 | offsets = xcalloc(raid_disks, sizeof(*offsets)); | |
616 | buf = xmalloc(raid_disks * chunk_size); | |
af3c3750 | 617 | |
af3c3750 PS |
618 | for(i=0; i<raid_disks; i++) { |
619 | fds[i] = -1; | |
620 | } | |
621 | close_flag = 1; | |
979afcb8 | 622 | |
a9c2c6c6 | 623 | comp = info->devs; |
2cf31121 | 624 | for (i=0, active_disks=0; active_disks<raid_disks; i++) { |
a9c2c6c6 | 625 | int disk_slot = comp->disk.raid_disk; |
2cf31121 PS |
626 | if(disk_slot >= 0) { |
627 | disk_name[disk_slot] = map_dev(comp->disk.major, comp->disk.minor, 0); | |
628 | offsets[disk_slot] = comp->data_offset * 512; | |
e645b341 | 629 | fds[disk_slot] = open(disk_name[disk_slot], O_RDWR | O_SYNC); |
2cf31121 PS |
630 | if (fds[disk_slot] < 0) { |
631 | perror(disk_name[disk_slot]); | |
632 | fprintf(stderr,"%s: cannot open %s\n", prg, disk_name[disk_slot]); | |
633 | exit_err = 6; | |
634 | goto exitHere; | |
635 | } | |
636 | active_disks++; | |
979afcb8 | 637 | } |
a9c2c6c6 | 638 | comp = comp->next; |
979afcb8 PS |
639 | } |
640 | ||
8d8ab389 | 641 | int rv = check_stripes(info, fds, offsets, |
979afcb8 | 642 | raid_disks, chunk_size, level, layout, |
f2e29ad6 | 643 | start, length, disk_name, repair, failed_disk1, failed_disk2); |
979afcb8 PS |
644 | if (rv != 0) { |
645 | fprintf(stderr, | |
a9c2c6c6 | 646 | "%s: check_stripes returned %d\n", prg, rv); |
af3c3750 PS |
647 | exit_err = 7; |
648 | goto exitHere; | |
979afcb8 PS |
649 | } |
650 | ||
af3c3750 PS |
651 | exitHere: |
652 | ||
653 | if (close_flag) | |
654 | for(i = 0; i < raid_disks; i++) | |
655 | close(fds[i]); | |
656 | ||
a9c2c6c6 | 657 | free(disk_name); |
979afcb8 PS |
658 | free(fds); |
659 | free(offsets); | |
660 | free(buf); | |
661 | ||
af3c3750 | 662 | exit(exit_err); |
979afcb8 | 663 | } |