b.liu | e958203 | 2025-04-17 19:18:16 +0800 | [diff] [blame] | 1 | // SPDX-License-Identifier: GPL-2.0+ |
| 2 | /* |
| 3 | * linux/fs/jbd2/recovery.c |
| 4 | * |
| 5 | * Written by Stephen C. Tweedie <sct@redhat.com>, 1999 |
| 6 | * |
| 7 | * Copyright 1999-2000 Red Hat Software --- All Rights Reserved |
| 8 | * |
| 9 | * Journal recovery routines for the generic filesystem journaling code; |
| 10 | * part of the ext2fs journaling system. |
| 11 | */ |
| 12 | |
| 13 | #ifndef __KERNEL__ |
| 14 | #include "jfs_user.h" |
| 15 | #else |
| 16 | #include <linux/time.h> |
| 17 | #include <linux/fs.h> |
| 18 | #include <linux/jbd2.h> |
| 19 | #include <linux/errno.h> |
| 20 | #include <linux/crc32.h> |
| 21 | #include <linux/blkdev.h> |
| 22 | #endif |
| 23 | |
| 24 | /* |
| 25 | * Maintain information about the progress of the recovery job, so that |
| 26 | * the different passes can carry information between them. |
| 27 | */ |
| 28 | struct recovery_info |
| 29 | { |
| 30 | tid_t start_transaction; |
| 31 | tid_t end_transaction; |
| 32 | |
| 33 | int nr_replays; |
| 34 | int nr_revokes; |
| 35 | int nr_revoke_hits; |
| 36 | }; |
| 37 | |
| 38 | enum passtype {PASS_SCAN, PASS_REVOKE, PASS_REPLAY}; |
| 39 | static int do_one_pass(journal_t *journal, |
| 40 | struct recovery_info *info, enum passtype pass); |
| 41 | static int scan_revoke_records(journal_t *, struct buffer_head *, |
| 42 | tid_t, struct recovery_info *); |
| 43 | |
| 44 | #ifdef __KERNEL__ |
| 45 | |
| 46 | /* Release readahead buffers after use */ |
| 47 | static void journal_brelse_array(struct buffer_head *b[], int n) |
| 48 | { |
| 49 | while (--n >= 0) |
| 50 | brelse (b[n]); |
| 51 | } |
| 52 | |
| 53 | |
| 54 | /* |
| 55 | * When reading from the journal, we are going through the block device |
| 56 | * layer directly and so there is no readahead being done for us. We |
| 57 | * need to implement any readahead ourselves if we want it to happen at |
| 58 | * all. Recovery is basically one long sequential read, so make sure we |
| 59 | * do the IO in reasonably large chunks. |
| 60 | * |
| 61 | * This is not so critical that we need to be enormously clever about |
| 62 | * the readahead size, though. 128K is a purely arbitrary, good-enough |
| 63 | * fixed value. |
| 64 | */ |
| 65 | |
| 66 | #define MAXBUF 8 |
| 67 | static int do_readahead(journal_t *journal, unsigned int start) |
| 68 | { |
| 69 | int err; |
| 70 | unsigned int max, nbufs, next; |
| 71 | unsigned long long blocknr; |
| 72 | struct buffer_head *bh; |
| 73 | |
| 74 | struct buffer_head * bufs[MAXBUF]; |
| 75 | |
| 76 | /* Do up to 128K of readahead */ |
| 77 | max = start + (128 * 1024 / journal->j_blocksize); |
| 78 | if (max > journal->j_maxlen) |
| 79 | max = journal->j_maxlen; |
| 80 | |
| 81 | /* Do the readahead itself. We'll submit MAXBUF buffer_heads at |
| 82 | * a time to the block device IO layer. */ |
| 83 | |
| 84 | nbufs = 0; |
| 85 | |
| 86 | for (next = start; next < max; next++) { |
| 87 | err = jbd2_journal_bmap(journal, next, &blocknr); |
| 88 | |
| 89 | if (err) { |
| 90 | printk(KERN_ERR "JBD2: bad block at offset %u\n", |
| 91 | next); |
| 92 | goto failed; |
| 93 | } |
| 94 | |
| 95 | bh = __getblk(journal->j_dev, blocknr, journal->j_blocksize); |
| 96 | if (!bh) { |
| 97 | err = -ENOMEM; |
| 98 | goto failed; |
| 99 | } |
| 100 | |
| 101 | if (!buffer_uptodate(bh) && !buffer_locked(bh)) { |
| 102 | bufs[nbufs++] = bh; |
| 103 | if (nbufs == MAXBUF) { |
| 104 | ll_rw_block(REQ_OP_READ, 0, nbufs, bufs); |
| 105 | journal_brelse_array(bufs, nbufs); |
| 106 | nbufs = 0; |
| 107 | } |
| 108 | } else |
| 109 | brelse(bh); |
| 110 | } |
| 111 | |
| 112 | if (nbufs) |
| 113 | ll_rw_block(REQ_OP_READ, 0, nbufs, bufs); |
| 114 | err = 0; |
| 115 | |
| 116 | failed: |
| 117 | if (nbufs) |
| 118 | journal_brelse_array(bufs, nbufs); |
| 119 | return err; |
| 120 | } |
| 121 | |
| 122 | #endif /* __KERNEL__ */ |
| 123 | |
| 124 | |
| 125 | /* |
| 126 | * Read a block from the journal |
| 127 | */ |
| 128 | |
| 129 | static int jread(struct buffer_head **bhp, journal_t *journal, |
| 130 | unsigned int offset) |
| 131 | { |
| 132 | int err; |
| 133 | unsigned long long blocknr; |
| 134 | struct buffer_head *bh; |
| 135 | |
| 136 | *bhp = NULL; |
| 137 | |
| 138 | if (offset >= journal->j_maxlen) { |
| 139 | printk(KERN_ERR "JBD2: corrupted journal superblock\n"); |
| 140 | return -EFSCORRUPTED; |
| 141 | } |
| 142 | |
| 143 | err = jbd2_journal_bmap(journal, offset, &blocknr); |
| 144 | |
| 145 | if (err) { |
| 146 | printk(KERN_ERR "JBD2: bad block at offset %u\n", |
| 147 | offset); |
| 148 | return err; |
| 149 | } |
| 150 | |
| 151 | bh = __getblk(journal->j_dev, blocknr, journal->j_blocksize); |
| 152 | if (!bh) |
| 153 | return -ENOMEM; |
| 154 | |
| 155 | if (!buffer_uptodate(bh)) { |
| 156 | /* If this is a brand new buffer, start readahead. |
| 157 | Otherwise, we assume we are already reading it. */ |
| 158 | if (!buffer_req(bh)) |
| 159 | do_readahead(journal, offset); |
| 160 | wait_on_buffer(bh); |
| 161 | } |
| 162 | |
| 163 | if (!buffer_uptodate(bh)) { |
| 164 | printk(KERN_ERR "JBD2: Failed to read block at offset %u\n", |
| 165 | offset); |
| 166 | brelse(bh); |
| 167 | return -EIO; |
| 168 | } |
| 169 | |
| 170 | *bhp = bh; |
| 171 | return 0; |
| 172 | } |
| 173 | |
| 174 | static int jbd2_descriptor_block_csum_verify(journal_t *j, void *buf) |
| 175 | { |
| 176 | struct jbd2_journal_block_tail *tail; |
| 177 | __be32 provided; |
| 178 | __u32 calculated; |
| 179 | |
| 180 | if (!jbd2_journal_has_csum_v2or3(j)) |
| 181 | return 1; |
| 182 | |
| 183 | tail = (struct jbd2_journal_block_tail *)(buf + j->j_blocksize - |
| 184 | sizeof(struct jbd2_journal_block_tail)); |
| 185 | provided = tail->t_checksum; |
| 186 | tail->t_checksum = 0; |
| 187 | calculated = jbd2_chksum(j, j->j_csum_seed, buf, j->j_blocksize); |
| 188 | tail->t_checksum = provided; |
| 189 | |
| 190 | return provided == cpu_to_be32(calculated); |
| 191 | } |
| 192 | |
| 193 | /* |
| 194 | * Count the number of in-use tags in a journal descriptor block. |
| 195 | */ |
| 196 | |
| 197 | static int count_tags(journal_t *journal, struct buffer_head *bh) |
| 198 | { |
| 199 | char * tagp; |
| 200 | journal_block_tag_t * tag; |
| 201 | int nr = 0, size = journal->j_blocksize; |
| 202 | int tag_bytes = journal_tag_bytes(journal); |
| 203 | |
| 204 | if (jbd2_journal_has_csum_v2or3(journal)) |
| 205 | size -= sizeof(struct jbd2_journal_block_tail); |
| 206 | |
| 207 | tagp = &bh->b_data[sizeof(journal_header_t)]; |
| 208 | |
| 209 | while ((tagp - bh->b_data + tag_bytes) <= size) { |
| 210 | tag = (journal_block_tag_t *) tagp; |
| 211 | |
| 212 | nr++; |
| 213 | tagp += tag_bytes; |
| 214 | if (!(tag->t_flags & cpu_to_be16(JBD2_FLAG_SAME_UUID))) |
| 215 | tagp += 16; |
| 216 | |
| 217 | if (tag->t_flags & cpu_to_be16(JBD2_FLAG_LAST_TAG)) |
| 218 | break; |
| 219 | } |
| 220 | |
| 221 | return nr; |
| 222 | } |
| 223 | |
| 224 | |
| 225 | /* Make sure we wrap around the log correctly! */ |
| 226 | #define wrap(journal, var) \ |
| 227 | do { \ |
| 228 | if (var >= (journal)->j_last) \ |
| 229 | var -= ((journal)->j_last - (journal)->j_first); \ |
| 230 | } while (0) |
| 231 | |
| 232 | /** |
| 233 | * jbd2_journal_recover - recovers a on-disk journal |
| 234 | * @journal: the journal to recover |
| 235 | * |
| 236 | * The primary function for recovering the log contents when mounting a |
| 237 | * journaled device. |
| 238 | * |
| 239 | * Recovery is done in three passes. In the first pass, we look for the |
| 240 | * end of the log. In the second, we assemble the list of revoke |
| 241 | * blocks. In the third and final pass, we replay any un-revoked blocks |
| 242 | * in the log. |
| 243 | */ |
| 244 | int jbd2_journal_recover(journal_t *journal) |
| 245 | { |
| 246 | int err, err2; |
| 247 | journal_superblock_t * sb; |
| 248 | |
| 249 | struct recovery_info info; |
| 250 | errseq_t wb_err; |
| 251 | struct address_space *mapping; |
| 252 | |
| 253 | memset(&info, 0, sizeof(info)); |
| 254 | sb = journal->j_superblock; |
| 255 | |
| 256 | /* |
| 257 | * The journal superblock's s_start field (the current log head) |
| 258 | * is always zero if, and only if, the journal was cleanly |
| 259 | * unmounted. |
| 260 | */ |
| 261 | |
| 262 | if (!sb->s_start) { |
| 263 | jbd_debug(1, "No recovery required, last transaction %d\n", |
| 264 | be32_to_cpu(sb->s_sequence)); |
| 265 | journal->j_transaction_sequence = be32_to_cpu(sb->s_sequence) + 1; |
| 266 | return 0; |
| 267 | } |
| 268 | |
| 269 | wb_err = 0; |
| 270 | mapping = journal->j_fs_dev->bd_inode->i_mapping; |
| 271 | errseq_check_and_advance(&mapping->wb_err, &wb_err); |
| 272 | err = do_one_pass(journal, &info, PASS_SCAN); |
| 273 | if (!err) |
| 274 | err = do_one_pass(journal, &info, PASS_REVOKE); |
| 275 | if (!err) |
| 276 | err = do_one_pass(journal, &info, PASS_REPLAY); |
| 277 | |
| 278 | jbd_debug(1, "JBD2: recovery, exit status %d, " |
| 279 | "recovered transactions %u to %u\n", |
| 280 | err, info.start_transaction, info.end_transaction); |
| 281 | jbd_debug(1, "JBD2: Replayed %d and revoked %d/%d blocks\n", |
| 282 | info.nr_replays, info.nr_revoke_hits, info.nr_revokes); |
| 283 | |
| 284 | /* Restart the log at the next transaction ID, thus invalidating |
| 285 | * any existing commit records in the log. */ |
| 286 | journal->j_transaction_sequence = ++info.end_transaction; |
| 287 | |
| 288 | jbd2_journal_clear_revoke(journal); |
| 289 | err2 = sync_blockdev(journal->j_fs_dev); |
| 290 | if (!err) |
| 291 | err = err2; |
| 292 | err2 = errseq_check_and_advance(&mapping->wb_err, &wb_err); |
| 293 | if (!err) |
| 294 | err = err2; |
| 295 | /* Make sure all replayed data is on permanent storage */ |
| 296 | if (journal->j_flags & JBD2_BARRIER) { |
| 297 | err2 = blkdev_issue_flush(journal->j_fs_dev, GFP_KERNEL, NULL); |
| 298 | if (!err) |
| 299 | err = err2; |
| 300 | } |
| 301 | return err; |
| 302 | } |
| 303 | |
| 304 | /** |
| 305 | * jbd2_journal_skip_recovery - Start journal and wipe exiting records |
| 306 | * @journal: journal to startup |
| 307 | * |
| 308 | * Locate any valid recovery information from the journal and set up the |
| 309 | * journal structures in memory to ignore it (presumably because the |
| 310 | * caller has evidence that it is out of date). |
| 311 | * This function doesn't appear to be exported.. |
| 312 | * |
| 313 | * We perform one pass over the journal to allow us to tell the user how |
| 314 | * much recovery information is being erased, and to let us initialise |
| 315 | * the journal transaction sequence numbers to the next unused ID. |
| 316 | */ |
| 317 | int jbd2_journal_skip_recovery(journal_t *journal) |
| 318 | { |
| 319 | int err; |
| 320 | |
| 321 | struct recovery_info info; |
| 322 | |
| 323 | memset (&info, 0, sizeof(info)); |
| 324 | |
| 325 | err = do_one_pass(journal, &info, PASS_SCAN); |
| 326 | |
| 327 | if (err) { |
| 328 | printk(KERN_ERR "JBD2: error %d scanning journal\n", err); |
| 329 | ++journal->j_transaction_sequence; |
| 330 | } else { |
| 331 | #ifdef CONFIG_JBD2_DEBUG |
| 332 | int dropped = info.end_transaction - |
| 333 | be32_to_cpu(journal->j_superblock->s_sequence); |
| 334 | jbd_debug(1, |
| 335 | "JBD2: ignoring %d transaction%s from the journal.\n", |
| 336 | dropped, (dropped == 1) ? "" : "s"); |
| 337 | #endif |
| 338 | journal->j_transaction_sequence = ++info.end_transaction; |
| 339 | } |
| 340 | |
| 341 | journal->j_tail = 0; |
| 342 | return err; |
| 343 | } |
| 344 | |
| 345 | static inline unsigned long long read_tag_block(journal_t *journal, |
| 346 | journal_block_tag_t *tag) |
| 347 | { |
| 348 | unsigned long long block = be32_to_cpu(tag->t_blocknr); |
| 349 | if (jbd2_has_feature_64bit(journal)) |
| 350 | block |= (u64)be32_to_cpu(tag->t_blocknr_high) << 32; |
| 351 | return block; |
| 352 | } |
| 353 | |
| 354 | /* |
| 355 | * calc_chksums calculates the checksums for the blocks described in the |
| 356 | * descriptor block. |
| 357 | */ |
| 358 | static int calc_chksums(journal_t *journal, struct buffer_head *bh, |
| 359 | unsigned long *next_log_block, __u32 *crc32_sum) |
| 360 | { |
| 361 | int i, num_blks, err; |
| 362 | unsigned long io_block; |
| 363 | struct buffer_head *obh; |
| 364 | |
| 365 | num_blks = count_tags(journal, bh); |
| 366 | /* Calculate checksum of the descriptor block. */ |
| 367 | *crc32_sum = crc32_be(*crc32_sum, (void *)bh->b_data, bh->b_size); |
| 368 | |
| 369 | for (i = 0; i < num_blks; i++) { |
| 370 | io_block = (*next_log_block)++; |
| 371 | wrap(journal, *next_log_block); |
| 372 | err = jread(&obh, journal, io_block); |
| 373 | if (err) { |
| 374 | printk(KERN_ERR "JBD2: IO error %d recovering block " |
| 375 | "%lu in log\n", err, io_block); |
| 376 | return 1; |
| 377 | } else { |
| 378 | *crc32_sum = crc32_be(*crc32_sum, (void *)obh->b_data, |
| 379 | obh->b_size); |
| 380 | } |
| 381 | put_bh(obh); |
| 382 | } |
| 383 | return 0; |
| 384 | } |
| 385 | |
| 386 | static int jbd2_commit_block_csum_verify(journal_t *j, void *buf) |
| 387 | { |
| 388 | struct commit_header *h; |
| 389 | __be32 provided; |
| 390 | __u32 calculated; |
| 391 | |
| 392 | if (!jbd2_journal_has_csum_v2or3(j)) |
| 393 | return 1; |
| 394 | |
| 395 | h = buf; |
| 396 | provided = h->h_chksum[0]; |
| 397 | h->h_chksum[0] = 0; |
| 398 | calculated = jbd2_chksum(j, j->j_csum_seed, buf, j->j_blocksize); |
| 399 | h->h_chksum[0] = provided; |
| 400 | |
| 401 | return provided == cpu_to_be32(calculated); |
| 402 | } |
| 403 | |
| 404 | static int jbd2_block_tag_csum_verify(journal_t *j, journal_block_tag_t *tag, |
| 405 | void *buf, __u32 sequence) |
| 406 | { |
| 407 | journal_block_tag3_t *tag3 = (journal_block_tag3_t *)tag; |
| 408 | __u32 csum32; |
| 409 | __be32 seq; |
| 410 | |
| 411 | if (!jbd2_journal_has_csum_v2or3(j)) |
| 412 | return 1; |
| 413 | |
| 414 | seq = cpu_to_be32(sequence); |
| 415 | csum32 = jbd2_chksum(j, j->j_csum_seed, (__u8 *)&seq, sizeof(seq)); |
| 416 | csum32 = jbd2_chksum(j, csum32, buf, j->j_blocksize); |
| 417 | |
| 418 | if (jbd2_has_feature_csum3(j)) |
| 419 | return tag3->t_checksum == cpu_to_be32(csum32); |
| 420 | else |
| 421 | return tag->t_checksum == cpu_to_be16(csum32); |
| 422 | } |
| 423 | |
| 424 | static int do_one_pass(journal_t *journal, |
| 425 | struct recovery_info *info, enum passtype pass) |
| 426 | { |
| 427 | unsigned int first_commit_ID, next_commit_ID; |
| 428 | unsigned long next_log_block; |
| 429 | int err, success = 0; |
| 430 | journal_superblock_t * sb; |
| 431 | journal_header_t * tmp; |
| 432 | struct buffer_head * bh; |
| 433 | unsigned int sequence; |
| 434 | int blocktype; |
| 435 | int tag_bytes = journal_tag_bytes(journal); |
| 436 | __u32 crc32_sum = ~0; /* Transactional Checksums */ |
| 437 | int descr_csum_size = 0; |
| 438 | int block_error = 0; |
| 439 | |
| 440 | /* |
| 441 | * First thing is to establish what we expect to find in the log |
| 442 | * (in terms of transaction IDs), and where (in terms of log |
| 443 | * block offsets): query the superblock. |
| 444 | */ |
| 445 | |
| 446 | sb = journal->j_superblock; |
| 447 | next_commit_ID = be32_to_cpu(sb->s_sequence); |
| 448 | next_log_block = be32_to_cpu(sb->s_start); |
| 449 | |
| 450 | first_commit_ID = next_commit_ID; |
| 451 | if (pass == PASS_SCAN) |
| 452 | info->start_transaction = first_commit_ID; |
| 453 | |
| 454 | jbd_debug(1, "Starting recovery pass %d\n", pass); |
| 455 | |
| 456 | /* |
| 457 | * Now we walk through the log, transaction by transaction, |
| 458 | * making sure that each transaction has a commit block in the |
| 459 | * expected place. Each complete transaction gets replayed back |
| 460 | * into the main filesystem. |
| 461 | */ |
| 462 | |
| 463 | while (1) { |
| 464 | int flags; |
| 465 | char * tagp; |
| 466 | journal_block_tag_t * tag; |
| 467 | struct buffer_head * obh; |
| 468 | struct buffer_head * nbh; |
| 469 | |
| 470 | cond_resched(); |
| 471 | |
| 472 | /* If we already know where to stop the log traversal, |
| 473 | * check right now that we haven't gone past the end of |
| 474 | * the log. */ |
| 475 | |
| 476 | if (pass != PASS_SCAN) |
| 477 | if (tid_geq(next_commit_ID, info->end_transaction)) |
| 478 | break; |
| 479 | |
| 480 | jbd_debug(2, "Scanning for sequence ID %u at %lu/%lu\n", |
| 481 | next_commit_ID, next_log_block, journal->j_last); |
| 482 | |
| 483 | /* Skip over each chunk of the transaction looking |
| 484 | * either the next descriptor block or the final commit |
| 485 | * record. */ |
| 486 | |
| 487 | jbd_debug(3, "JBD2: checking block %ld\n", next_log_block); |
| 488 | err = jread(&bh, journal, next_log_block); |
| 489 | if (err) |
| 490 | goto failed; |
| 491 | |
| 492 | next_log_block++; |
| 493 | wrap(journal, next_log_block); |
| 494 | |
| 495 | /* What kind of buffer is it? |
| 496 | * |
| 497 | * If it is a descriptor block, check that it has the |
| 498 | * expected sequence number. Otherwise, we're all done |
| 499 | * here. */ |
| 500 | |
| 501 | tmp = (journal_header_t *)bh->b_data; |
| 502 | |
| 503 | if (tmp->h_magic != cpu_to_be32(JBD2_MAGIC_NUMBER)) { |
| 504 | brelse(bh); |
| 505 | break; |
| 506 | } |
| 507 | |
| 508 | blocktype = be32_to_cpu(tmp->h_blocktype); |
| 509 | sequence = be32_to_cpu(tmp->h_sequence); |
| 510 | jbd_debug(3, "Found magic %d, sequence %d\n", |
| 511 | blocktype, sequence); |
| 512 | |
| 513 | if (sequence != next_commit_ID) { |
| 514 | brelse(bh); |
| 515 | break; |
| 516 | } |
| 517 | |
| 518 | /* OK, we have a valid descriptor block which matches |
| 519 | * all of the sequence number checks. What are we going |
| 520 | * to do with it? That depends on the pass... */ |
| 521 | |
| 522 | switch(blocktype) { |
| 523 | case JBD2_DESCRIPTOR_BLOCK: |
| 524 | /* Verify checksum first */ |
| 525 | if (jbd2_journal_has_csum_v2or3(journal)) |
| 526 | descr_csum_size = |
| 527 | sizeof(struct jbd2_journal_block_tail); |
| 528 | if (descr_csum_size > 0 && |
| 529 | !jbd2_descriptor_block_csum_verify(journal, |
| 530 | bh->b_data)) { |
| 531 | printk(KERN_ERR "JBD2: Invalid checksum " |
| 532 | "recovering block %lu in log\n", |
| 533 | next_log_block); |
| 534 | err = -EFSBADCRC; |
| 535 | brelse(bh); |
| 536 | goto failed; |
| 537 | } |
| 538 | |
| 539 | /* If it is a valid descriptor block, replay it |
| 540 | * in pass REPLAY; if journal_checksums enabled, then |
| 541 | * calculate checksums in PASS_SCAN, otherwise, |
| 542 | * just skip over the blocks it describes. */ |
| 543 | if (pass != PASS_REPLAY) { |
| 544 | if (pass == PASS_SCAN && |
| 545 | jbd2_has_feature_checksum(journal) && |
| 546 | !info->end_transaction) { |
| 547 | if (calc_chksums(journal, bh, |
| 548 | &next_log_block, |
| 549 | &crc32_sum)) { |
| 550 | put_bh(bh); |
| 551 | break; |
| 552 | } |
| 553 | put_bh(bh); |
| 554 | continue; |
| 555 | } |
| 556 | next_log_block += count_tags(journal, bh); |
| 557 | wrap(journal, next_log_block); |
| 558 | put_bh(bh); |
| 559 | continue; |
| 560 | } |
| 561 | |
| 562 | /* A descriptor block: we can now write all of |
| 563 | * the data blocks. Yay, useful work is finally |
| 564 | * getting done here! */ |
| 565 | |
| 566 | tagp = &bh->b_data[sizeof(journal_header_t)]; |
| 567 | while ((tagp - bh->b_data + tag_bytes) |
| 568 | <= journal->j_blocksize - descr_csum_size) { |
| 569 | unsigned long io_block; |
| 570 | |
| 571 | tag = (journal_block_tag_t *) tagp; |
| 572 | flags = be16_to_cpu(tag->t_flags); |
| 573 | |
| 574 | io_block = next_log_block++; |
| 575 | wrap(journal, next_log_block); |
| 576 | err = jread(&obh, journal, io_block); |
| 577 | if (err) { |
| 578 | /* Recover what we can, but |
| 579 | * report failure at the end. */ |
| 580 | success = err; |
| 581 | printk(KERN_ERR |
| 582 | "JBD2: IO error %d recovering " |
| 583 | "block %ld in log\n", |
| 584 | err, io_block); |
| 585 | } else { |
| 586 | unsigned long long blocknr; |
| 587 | |
| 588 | J_ASSERT(obh != NULL); |
| 589 | blocknr = read_tag_block(journal, |
| 590 | tag); |
| 591 | |
| 592 | /* If the block has been |
| 593 | * revoked, then we're all done |
| 594 | * here. */ |
| 595 | if (jbd2_journal_test_revoke |
| 596 | (journal, blocknr, |
| 597 | next_commit_ID)) { |
| 598 | brelse(obh); |
| 599 | ++info->nr_revoke_hits; |
| 600 | goto skip_write; |
| 601 | } |
| 602 | |
| 603 | /* Look for block corruption */ |
| 604 | if (!jbd2_block_tag_csum_verify( |
| 605 | journal, tag, obh->b_data, |
| 606 | be32_to_cpu(tmp->h_sequence))) { |
| 607 | brelse(obh); |
| 608 | success = -EFSBADCRC; |
| 609 | printk(KERN_ERR "JBD2: Invalid " |
| 610 | "checksum recovering " |
| 611 | "data block %llu in " |
| 612 | "log\n", blocknr); |
| 613 | block_error = 1; |
| 614 | goto skip_write; |
| 615 | } |
| 616 | |
| 617 | /* Find a buffer for the new |
| 618 | * data being restored */ |
| 619 | nbh = __getblk(journal->j_fs_dev, |
| 620 | blocknr, |
| 621 | journal->j_blocksize); |
| 622 | if (nbh == NULL) { |
| 623 | printk(KERN_ERR |
| 624 | "JBD2: Out of memory " |
| 625 | "during recovery.\n"); |
| 626 | err = -ENOMEM; |
| 627 | brelse(bh); |
| 628 | brelse(obh); |
| 629 | goto failed; |
| 630 | } |
| 631 | |
| 632 | lock_buffer(nbh); |
| 633 | memcpy(nbh->b_data, obh->b_data, |
| 634 | journal->j_blocksize); |
| 635 | if (flags & JBD2_FLAG_ESCAPE) { |
| 636 | *((__be32 *)nbh->b_data) = |
| 637 | cpu_to_be32(JBD2_MAGIC_NUMBER); |
| 638 | } |
| 639 | |
| 640 | BUFFER_TRACE(nbh, "marking dirty"); |
| 641 | set_buffer_uptodate(nbh); |
| 642 | mark_buffer_dirty(nbh); |
| 643 | BUFFER_TRACE(nbh, "marking uptodate"); |
| 644 | ++info->nr_replays; |
| 645 | /* ll_rw_block(WRITE, 1, &nbh); */ |
| 646 | unlock_buffer(nbh); |
| 647 | brelse(obh); |
| 648 | brelse(nbh); |
| 649 | } |
| 650 | |
| 651 | skip_write: |
| 652 | tagp += tag_bytes; |
| 653 | if (!(flags & JBD2_FLAG_SAME_UUID)) |
| 654 | tagp += 16; |
| 655 | |
| 656 | if (flags & JBD2_FLAG_LAST_TAG) |
| 657 | break; |
| 658 | } |
| 659 | |
| 660 | brelse(bh); |
| 661 | continue; |
| 662 | |
| 663 | case JBD2_COMMIT_BLOCK: |
| 664 | /* How to differentiate between interrupted commit |
| 665 | * and journal corruption ? |
| 666 | * |
| 667 | * {nth transaction} |
| 668 | * Checksum Verification Failed |
| 669 | * | |
| 670 | * ____________________ |
| 671 | * | | |
| 672 | * async_commit sync_commit |
| 673 | * | | |
| 674 | * | GO TO NEXT "Journal Corruption" |
| 675 | * | TRANSACTION |
| 676 | * | |
| 677 | * {(n+1)th transanction} |
| 678 | * | |
| 679 | * _______|______________ |
| 680 | * | | |
| 681 | * Commit block found Commit block not found |
| 682 | * | | |
| 683 | * "Journal Corruption" | |
| 684 | * _____________|_________ |
| 685 | * | | |
| 686 | * nth trans corrupt OR nth trans |
| 687 | * and (n+1)th interrupted interrupted |
| 688 | * before commit block |
| 689 | * could reach the disk. |
| 690 | * (Cannot find the difference in above |
| 691 | * mentioned conditions. Hence assume |
| 692 | * "Interrupted Commit".) |
| 693 | */ |
| 694 | |
| 695 | /* Found an expected commit block: if checksums |
| 696 | * are present verify them in PASS_SCAN; else not |
| 697 | * much to do other than move on to the next sequence |
| 698 | * number. */ |
| 699 | if (pass == PASS_SCAN && |
| 700 | jbd2_has_feature_checksum(journal)) { |
| 701 | int chksum_err, chksum_seen; |
| 702 | struct commit_header *cbh = |
| 703 | (struct commit_header *)bh->b_data; |
| 704 | unsigned found_chksum = |
| 705 | be32_to_cpu(cbh->h_chksum[0]); |
| 706 | |
| 707 | chksum_err = chksum_seen = 0; |
| 708 | |
| 709 | if (info->end_transaction) { |
| 710 | journal->j_failed_commit = |
| 711 | info->end_transaction; |
| 712 | brelse(bh); |
| 713 | break; |
| 714 | } |
| 715 | |
| 716 | if (crc32_sum == found_chksum && |
| 717 | cbh->h_chksum_type == JBD2_CRC32_CHKSUM && |
| 718 | cbh->h_chksum_size == |
| 719 | JBD2_CRC32_CHKSUM_SIZE) |
| 720 | chksum_seen = 1; |
| 721 | else if (!(cbh->h_chksum_type == 0 && |
| 722 | cbh->h_chksum_size == 0 && |
| 723 | found_chksum == 0 && |
| 724 | !chksum_seen)) |
| 725 | /* |
| 726 | * If fs is mounted using an old kernel and then |
| 727 | * kernel with journal_chksum is used then we |
| 728 | * get a situation where the journal flag has |
| 729 | * checksum flag set but checksums are not |
| 730 | * present i.e chksum = 0, in the individual |
| 731 | * commit blocks. |
| 732 | * Hence to avoid checksum failures, in this |
| 733 | * situation, this extra check is added. |
| 734 | */ |
| 735 | chksum_err = 1; |
| 736 | |
| 737 | if (chksum_err) { |
| 738 | info->end_transaction = next_commit_ID; |
| 739 | |
| 740 | if (!jbd2_has_feature_async_commit(journal)) { |
| 741 | journal->j_failed_commit = |
| 742 | next_commit_ID; |
| 743 | brelse(bh); |
| 744 | break; |
| 745 | } |
| 746 | } |
| 747 | crc32_sum = ~0; |
| 748 | } |
| 749 | if (pass == PASS_SCAN && |
| 750 | !jbd2_commit_block_csum_verify(journal, |
| 751 | bh->b_data)) { |
| 752 | info->end_transaction = next_commit_ID; |
| 753 | |
| 754 | if (!jbd2_has_feature_async_commit(journal)) { |
| 755 | journal->j_failed_commit = |
| 756 | next_commit_ID; |
| 757 | brelse(bh); |
| 758 | break; |
| 759 | } |
| 760 | } |
| 761 | brelse(bh); |
| 762 | next_commit_ID++; |
| 763 | continue; |
| 764 | |
| 765 | case JBD2_REVOKE_BLOCK: |
| 766 | /* If we aren't in the REVOKE pass, then we can |
| 767 | * just skip over this block. */ |
| 768 | if (pass != PASS_REVOKE) { |
| 769 | brelse(bh); |
| 770 | continue; |
| 771 | } |
| 772 | |
| 773 | err = scan_revoke_records(journal, bh, |
| 774 | next_commit_ID, info); |
| 775 | brelse(bh); |
| 776 | if (err) |
| 777 | goto failed; |
| 778 | continue; |
| 779 | |
| 780 | default: |
| 781 | jbd_debug(3, "Unrecognised magic %d, end of scan.\n", |
| 782 | blocktype); |
| 783 | brelse(bh); |
| 784 | goto done; |
| 785 | } |
| 786 | } |
| 787 | |
| 788 | done: |
| 789 | /* |
| 790 | * We broke out of the log scan loop: either we came to the |
| 791 | * known end of the log or we found an unexpected block in the |
| 792 | * log. If the latter happened, then we know that the "current" |
| 793 | * transaction marks the end of the valid log. |
| 794 | */ |
| 795 | |
| 796 | if (pass == PASS_SCAN) { |
| 797 | if (!info->end_transaction) |
| 798 | info->end_transaction = next_commit_ID; |
| 799 | } else { |
| 800 | /* It's really bad news if different passes end up at |
| 801 | * different places (but possible due to IO errors). */ |
| 802 | if (info->end_transaction != next_commit_ID) { |
| 803 | printk(KERN_ERR "JBD2: recovery pass %d ended at " |
| 804 | "transaction %u, expected %u\n", |
| 805 | pass, next_commit_ID, info->end_transaction); |
| 806 | if (!success) |
| 807 | success = -EIO; |
| 808 | } |
| 809 | } |
| 810 | if (block_error && success == 0) |
| 811 | success = -EIO; |
| 812 | return success; |
| 813 | |
| 814 | failed: |
| 815 | return err; |
| 816 | } |
| 817 | |
| 818 | /* Scan a revoke record, marking all blocks mentioned as revoked. */ |
| 819 | |
| 820 | static int scan_revoke_records(journal_t *journal, struct buffer_head *bh, |
| 821 | tid_t sequence, struct recovery_info *info) |
| 822 | { |
| 823 | jbd2_journal_revoke_header_t *header; |
| 824 | int offset, max; |
| 825 | int csum_size = 0; |
| 826 | __u32 rcount; |
| 827 | int record_len = 4; |
| 828 | |
| 829 | header = (jbd2_journal_revoke_header_t *) bh->b_data; |
| 830 | offset = sizeof(jbd2_journal_revoke_header_t); |
| 831 | rcount = be32_to_cpu(header->r_count); |
| 832 | |
| 833 | if (!jbd2_descriptor_block_csum_verify(journal, header)) |
| 834 | return -EFSBADCRC; |
| 835 | |
| 836 | if (jbd2_journal_has_csum_v2or3(journal)) |
| 837 | csum_size = sizeof(struct jbd2_journal_block_tail); |
| 838 | if (rcount > journal->j_blocksize - csum_size) |
| 839 | return -EINVAL; |
| 840 | max = rcount; |
| 841 | |
| 842 | if (jbd2_has_feature_64bit(journal)) |
| 843 | record_len = 8; |
| 844 | |
| 845 | while (offset + record_len <= max) { |
| 846 | unsigned long long blocknr; |
| 847 | int err; |
| 848 | |
| 849 | if (record_len == 4) |
| 850 | blocknr = be32_to_cpu(* ((__be32 *) (bh->b_data+offset))); |
| 851 | else |
| 852 | blocknr = be64_to_cpu(* ((__be64 *) (bh->b_data+offset))); |
| 853 | offset += record_len; |
| 854 | err = jbd2_journal_set_revoke(journal, blocknr, sequence); |
| 855 | if (err) |
| 856 | return err; |
| 857 | ++info->nr_revokes; |
| 858 | } |
| 859 | return 0; |
| 860 | } |