Blame view
fs/jbd/revoke.c
21.3 KB
1da177e4c Linux-2.6.12-rc2 |
1 |
/* |
588626996 fix file specific... |
2 |
* linux/fs/jbd/revoke.c |
ae6ddcc5f [PATCH] ext3 and ... |
3 |
* |
1da177e4c Linux-2.6.12-rc2 |
4 5 6 7 8 9 10 11 12 13 14 15 16 17 |
* Written by Stephen C. Tweedie <sct@redhat.com>, 2000 * * Copyright 2000 Red Hat corp --- All Rights Reserved * * This file is part of the Linux kernel and is made available under * the terms of the GNU General Public License, version 2, or at your * option, any later version, incorporated herein by reference. * * Journal revoke routines for the generic filesystem journaling code; * part of the ext2fs journaling system. * * Revoke is the mechanism used to prevent old log records for deleted * metadata from being replayed on top of newer data using the same * blocks. The revoke mechanism is used in two separate places: |
ae6ddcc5f [PATCH] ext3 and ... |
18 |
* |
1da177e4c Linux-2.6.12-rc2 |
19 20 |
* + Commit: during commit we write the entire list of the current * transaction's revoked blocks to the journal |
ae6ddcc5f [PATCH] ext3 and ... |
21 |
* |
1da177e4c Linux-2.6.12-rc2 |
22 23 24 25 26 27 28 29 30 31 |
* + Recovery: during recovery we record the transaction ID of all * revoked blocks. If there are multiple revoke records in the log * for a single block, only the last one counts, and if there is a log * entry for a block beyond the last revoke, then that log entry still * gets replayed. * * We can get interactions between revokes and new log data within a * single transaction: * * Block is revoked and then journaled: |
ae6ddcc5f [PATCH] ext3 and ... |
32 |
* The desired end result is the journaling of the new block, so we |
1da177e4c Linux-2.6.12-rc2 |
33 34 35 36 37 38 39 40 41 42 43 |
* cancel the revoke before the transaction commits. * * Block is journaled and then revoked: * The revoke must take precedence over the write of the block, so we * need either to cancel the journal entry or to write the revoke * later in the log than the log block. In this case, we choose the * latter: journaling a block cancels any revoke record for that block * in the current transaction, so any revoke for that block in the * transaction must have happened after the block was journaled and so * the revoke must take precedence. * |
ae6ddcc5f [PATCH] ext3 and ... |
44 |
* Block is revoked and then written as data: |
1da177e4c Linux-2.6.12-rc2 |
45 46 47 48 49 |
* The data write is allowed to succeed, but the revoke is _not_ * cancelled. We still need to prevent old log records from * overwriting the new data. We don't even need to clear the revoke * bit here. * |
8c111b3f5 jbd: clear revoke... |
50 51 52 53 |
* We cache revoke status of a buffer in the current transaction in b_states * bits. As the name says, revokevalid flag indicates that the cached revoke * status of a buffer is valid and we can rely on the cached status. * |
1da177e4c Linux-2.6.12-rc2 |
54 55 56 57 58 59 60 |
* Revoke information on buffers is a tri-state value: * * RevokeValid clear: no cached revoke status, need to look it up * RevokeValid set, Revoked clear: * buffer has not been revoked, and cancel_revoke * need do nothing. * RevokeValid set, Revoked set: |
ae6ddcc5f [PATCH] ext3 and ... |
61 |
* buffer has been revoked. |
324338794 jbd: update locki... |
62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 |
* * Locking rules: * We keep two hash tables of revoke records. One hashtable belongs to the * running transaction (is pointed to by journal->j_revoke), the other one * belongs to the committing transaction. Accesses to the second hash table * happen only from the kjournald and no other thread touches this table. Also * journal_switch_revoke_table() which switches which hashtable belongs to the * running and which to the committing transaction is called only from * kjournald. Therefore we need no locks when accessing the hashtable belonging * to the committing transaction. * * All users operating on the hash table belonging to the running transaction * have a handle to the transaction. Therefore they are safe from kjournald * switching hash tables under them. For operations on the lists of entries in * the hash table j_revoke_lock is used. * |
25985edce Fix common misspe... |
78 |
* Finally, also replay code uses the hash tables but at this moment no one else |
324338794 jbd: update locki... |
79 80 |
* can touch them (filesystem isn't mounted yet) and hence no locking is * needed. |
1da177e4c Linux-2.6.12-rc2 |
81 82 83 84 85 86 87 88 89 90 91 |
*/ #ifndef __KERNEL__ #include "jfs_user.h" #else #include <linux/time.h> #include <linux/fs.h> #include <linux/jbd.h> #include <linux/errno.h> #include <linux/slab.h> #include <linux/list.h> |
1da177e4c Linux-2.6.12-rc2 |
92 |
#include <linux/init.h> |
38d726d15 jbd: use SWRITE_S... |
93 |
#include <linux/bio.h> |
1da177e4c Linux-2.6.12-rc2 |
94 |
#endif |
f482394cc is_power_of_2(): jbd |
95 |
#include <linux/log2.h> |
1da177e4c Linux-2.6.12-rc2 |
96 |
|
e18b890bb [PATCH] slab: rem... |
97 98 |
static struct kmem_cache *revoke_record_cache; static struct kmem_cache *revoke_table_cache; |
1da177e4c Linux-2.6.12-rc2 |
99 100 101 102 |
/* Each revoke record represents one single revoked block. During journal replay, this involves recording the transaction ID of the last transaction to revoke this block. */ |
ae6ddcc5f [PATCH] ext3 and ... |
103 |
struct jbd_revoke_record_s |
1da177e4c Linux-2.6.12-rc2 |
104 105 106 |
{ struct list_head hash; tid_t sequence; /* Used for recovery only */ |
9c28cbcce jbd: Journal bloc... |
107 |
unsigned int blocknr; |
1da177e4c Linux-2.6.12-rc2 |
108 109 110 111 112 113 114 115 |
}; /* The revoke table is just a simple hash table of revoke records. */ struct jbd_revoke_table_s { /* It is conceivable that we might want a larger hash table * for recovery. Must be a power of two. */ |
ae6ddcc5f [PATCH] ext3 and ... |
116 117 |
int hash_size; int hash_shift; |
1da177e4c Linux-2.6.12-rc2 |
118 119 120 121 122 123 124 |
struct list_head *hash_table; }; #ifdef __KERNEL__ static void write_one_revoke_record(journal_t *, transaction_t *, struct journal_head **, int *, |
38d726d15 jbd: use SWRITE_S... |
125 126 |
struct jbd_revoke_record_s *, int); static void flush_descriptor(journal_t *, struct journal_head *, int, int); |
1da177e4c Linux-2.6.12-rc2 |
127 128 129 130 131 |
#endif /* Utility functions to maintain the revoke table */ /* Borrowed from buffer.c: this is a tried and tested block hash function */ |
9c28cbcce jbd: Journal bloc... |
132 |
static inline int hash(journal_t *journal, unsigned int block) |
1da177e4c Linux-2.6.12-rc2 |
133 134 135 136 137 138 139 140 |
{ struct jbd_revoke_table_s *table = journal->j_revoke; int hash_shift = table->hash_shift; return ((block << (hash_shift - 6)) ^ (block >> 13) ^ (block << (hash_shift - 12))) & (table->hash_size - 1); } |
9c28cbcce jbd: Journal bloc... |
141 |
static int insert_revoke_hash(journal_t *journal, unsigned int blocknr, |
022a4a7bb [PATCH] fs/jbd/: ... |
142 |
tid_t seq) |
1da177e4c Linux-2.6.12-rc2 |
143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 |
{ struct list_head *hash_list; struct jbd_revoke_record_s *record; repeat: record = kmem_cache_alloc(revoke_record_cache, GFP_NOFS); if (!record) goto oom; record->sequence = seq; record->blocknr = blocknr; hash_list = &journal->j_revoke->hash_table[hash(journal, blocknr)]; spin_lock(&journal->j_revoke_lock); list_add(&record->hash, hash_list); spin_unlock(&journal->j_revoke_lock); return 0; oom: if (!journal_oom_retry) return -ENOMEM; |
08fc99bfc jbd: replace rema... |
163 164 |
jbd_debug(1, "ENOMEM in %s, retrying ", __func__); |
1da177e4c Linux-2.6.12-rc2 |
165 166 167 168 169 170 171 |
yield(); goto repeat; } /* Find a revoke record in the journal's hash table. */ static struct jbd_revoke_record_s *find_revoke_record(journal_t *journal, |
9c28cbcce jbd: Journal bloc... |
172 |
unsigned int blocknr) |
1da177e4c Linux-2.6.12-rc2 |
173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 |
{ struct list_head *hash_list; struct jbd_revoke_record_s *record; hash_list = &journal->j_revoke->hash_table[hash(journal, blocknr)]; spin_lock(&journal->j_revoke_lock); record = (struct jbd_revoke_record_s *) hash_list->next; while (&(record->hash) != hash_list) { if (record->blocknr == blocknr) { spin_unlock(&journal->j_revoke_lock); return record; } record = (struct jbd_revoke_record_s *) record->hash.next; } spin_unlock(&journal->j_revoke_lock); return NULL; } |
1984bb763 jbd: tidy up revo... |
191 192 193 194 195 196 197 198 199 200 201 |
void journal_destroy_revoke_caches(void) { if (revoke_record_cache) { kmem_cache_destroy(revoke_record_cache); revoke_record_cache = NULL; } if (revoke_table_cache) { kmem_cache_destroy(revoke_table_cache); revoke_table_cache = NULL; } } |
1da177e4c Linux-2.6.12-rc2 |
202 203 |
int __init journal_init_revoke_caches(void) { |
1984bb763 jbd: tidy up revo... |
204 205 |
J_ASSERT(!revoke_record_cache); J_ASSERT(!revoke_table_cache); |
1da177e4c Linux-2.6.12-rc2 |
206 207 |
revoke_record_cache = kmem_cache_create("revoke_record", sizeof(struct jbd_revoke_record_s), |
e12ba74d8 Group short-lived... |
208 209 210 |
0, SLAB_HWCACHE_ALIGN|SLAB_TEMPORARY, NULL); |
1076d17ac jbd/jbd2 NULL noise |
211 |
if (!revoke_record_cache) |
1984bb763 jbd: tidy up revo... |
212 |
goto record_cache_failure; |
1da177e4c Linux-2.6.12-rc2 |
213 214 215 |
revoke_table_cache = kmem_cache_create("revoke_table", sizeof(struct jbd_revoke_table_s), |
e12ba74d8 Group short-lived... |
216 |
0, SLAB_TEMPORARY, NULL); |
1984bb763 jbd: tidy up revo... |
217 218 |
if (!revoke_table_cache) goto table_cache_failure; |
1da177e4c Linux-2.6.12-rc2 |
219 |
return 0; |
1da177e4c Linux-2.6.12-rc2 |
220 |
|
1984bb763 jbd: tidy up revo... |
221 222 223 224 |
table_cache_failure: journal_destroy_revoke_caches(); record_cache_failure: return -ENOMEM; |
1da177e4c Linux-2.6.12-rc2 |
225 |
} |
f4d79ca2f jbd: eliminate du... |
226 |
static struct jbd_revoke_table_s *journal_init_revoke_table(int hash_size) |
1da177e4c Linux-2.6.12-rc2 |
227 |
{ |
f4d79ca2f jbd: eliminate du... |
228 229 230 |
int shift = 0; int tmp = hash_size; struct jbd_revoke_table_s *table; |
1da177e4c Linux-2.6.12-rc2 |
231 |
|
f4d79ca2f jbd: eliminate du... |
232 233 234 |
table = kmem_cache_alloc(revoke_table_cache, GFP_KERNEL); if (!table) goto out; |
1da177e4c Linux-2.6.12-rc2 |
235 |
|
1da177e4c Linux-2.6.12-rc2 |
236 237 |
while((tmp >>= 1UL) != 0UL) shift++; |
f4d79ca2f jbd: eliminate du... |
238 239 240 |
table->hash_size = hash_size; table->hash_shift = shift; table->hash_table = |
1da177e4c Linux-2.6.12-rc2 |
241 |
kmalloc(hash_size * sizeof(struct list_head), GFP_KERNEL); |
f4d79ca2f jbd: eliminate du... |
242 243 244 245 |
if (!table->hash_table) { kmem_cache_free(revoke_table_cache, table); table = NULL; goto out; |
1da177e4c Linux-2.6.12-rc2 |
246 247 248 |
} for (tmp = 0; tmp < hash_size; tmp++) |
f4d79ca2f jbd: eliminate du... |
249 |
INIT_LIST_HEAD(&table->hash_table[tmp]); |
1da177e4c Linux-2.6.12-rc2 |
250 |
|
f4d79ca2f jbd: eliminate du... |
251 252 253 254 255 256 257 258 259 260 261 262 |
out: return table; } static void journal_destroy_revoke_table(struct jbd_revoke_table_s *table) { int i; struct list_head *hash_list; for (i = 0; i < table->hash_size; i++) { hash_list = &table->hash_table[i]; J_ASSERT(list_empty(hash_list)); |
1da177e4c Linux-2.6.12-rc2 |
263 |
} |
f4d79ca2f jbd: eliminate du... |
264 265 266 |
kfree(table->hash_table); kmem_cache_free(revoke_table_cache, table); } |
1da177e4c Linux-2.6.12-rc2 |
267 |
|
f4d79ca2f jbd: eliminate du... |
268 269 270 271 |
/* Initialise the revoke table for a given journal to a given size. */ int journal_init_revoke(journal_t *journal, int hash_size) { J_ASSERT(journal->j_revoke_table[0] == NULL); |
f482394cc is_power_of_2(): jbd |
272 |
J_ASSERT(is_power_of_2(hash_size)); |
1da177e4c Linux-2.6.12-rc2 |
273 |
|
f4d79ca2f jbd: eliminate du... |
274 275 276 |
journal->j_revoke_table[0] = journal_init_revoke_table(hash_size); if (!journal->j_revoke_table[0]) goto fail0; |
1da177e4c Linux-2.6.12-rc2 |
277 |
|
f4d79ca2f jbd: eliminate du... |
278 279 280 |
journal->j_revoke_table[1] = journal_init_revoke_table(hash_size); if (!journal->j_revoke_table[1]) goto fail1; |
1da177e4c Linux-2.6.12-rc2 |
281 |
|
f4d79ca2f jbd: eliminate du... |
282 |
journal->j_revoke = journal->j_revoke_table[1]; |
1da177e4c Linux-2.6.12-rc2 |
283 284 285 286 |
spin_lock_init(&journal->j_revoke_lock); return 0; |
1da177e4c Linux-2.6.12-rc2 |
287 |
|
f4d79ca2f jbd: eliminate du... |
288 289 290 291 292 |
fail1: journal_destroy_revoke_table(journal->j_revoke_table[0]); fail0: return -ENOMEM; } |
1da177e4c Linux-2.6.12-rc2 |
293 |
|
f4d79ca2f jbd: eliminate du... |
294 |
/* Destroy a journal's revoke table. The table must already be empty! */ |
1da177e4c Linux-2.6.12-rc2 |
295 296 |
void journal_destroy_revoke(journal_t *journal) { |
1da177e4c Linux-2.6.12-rc2 |
297 |
journal->j_revoke = NULL; |
f4d79ca2f jbd: eliminate du... |
298 299 300 301 |
if (journal->j_revoke_table[0]) journal_destroy_revoke_table(journal->j_revoke_table[0]); if (journal->j_revoke_table[1]) journal_destroy_revoke_table(journal->j_revoke_table[1]); |
1da177e4c Linux-2.6.12-rc2 |
302 303 304 305 |
} #ifdef __KERNEL__ |
ae6ddcc5f [PATCH] ext3 and ... |
306 |
/* |
1da177e4c Linux-2.6.12-rc2 |
307 308 309 310 |
* journal_revoke: revoke a given buffer_head from the journal. This * prevents the block from being replayed during recovery if we take a * crash after this current transaction commits. Any subsequent * metadata writes of the buffer in this transaction cancel the |
ae6ddcc5f [PATCH] ext3 and ... |
311 |
* revoke. |
1da177e4c Linux-2.6.12-rc2 |
312 313 314 315 316 |
* * Note that this call may block --- it is up to the caller to make * sure that there are no further calls to journal_write_metadata * before the revoke is complete. In ext3, this implies calling the * revoke before clearing the block bitmap when we are deleting |
ae6ddcc5f [PATCH] ext3 and ... |
317 |
* metadata. |
1da177e4c Linux-2.6.12-rc2 |
318 319 320 |
* * Revoke performs a journal_forget on any buffer_head passed in as a * parameter, but does _not_ forget the buffer_head if the bh was only |
ae6ddcc5f [PATCH] ext3 and ... |
321 |
* found implicitly. |
1da177e4c Linux-2.6.12-rc2 |
322 323 324 325 326 327 328 |
* * bh_in may not be a journalled buffer - it may have come off * the hash tables without an attached journal_head. * * If bh_in is non-zero, journal_revoke() will decrement its b_count * by one. */ |
9c28cbcce jbd: Journal bloc... |
329 |
int journal_revoke(handle_t *handle, unsigned int blocknr, |
1da177e4c Linux-2.6.12-rc2 |
330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 |
struct buffer_head *bh_in) { struct buffer_head *bh = NULL; journal_t *journal; struct block_device *bdev; int err; might_sleep(); if (bh_in) BUFFER_TRACE(bh_in, "enter"); journal = handle->h_transaction->t_journal; if (!journal_set_features(journal, 0, 0, JFS_FEATURE_INCOMPAT_REVOKE)){ J_ASSERT (!"Cannot set revoke feature!"); return -EINVAL; } bdev = journal->j_fs_dev; bh = bh_in; if (!bh) { bh = __find_get_block(bdev, blocknr, journal->j_blocksize); if (bh) BUFFER_TRACE(bh, "found on hash"); } #ifdef JBD_EXPENSIVE_CHECKING else { struct buffer_head *bh2; /* If there is a different buffer_head lying around in * memory anywhere... */ bh2 = __find_get_block(bdev, blocknr, journal->j_blocksize); if (bh2) { /* ... and it has RevokeValid status... */ if (bh2 != bh && buffer_revokevalid(bh2)) /* ...then it better be revoked too, * since it's illegal to create a revoke * record against a buffer_head which is * not marked revoked --- that would * risk missing a subsequent revoke * cancel. */ J_ASSERT_BH(bh2, buffer_revoked(bh2)); put_bh(bh2); } } #endif /* We really ought not ever to revoke twice in a row without first having the revoke cancelled: it's illegal to free a block twice without allocating it in between! */ if (bh) { if (!J_EXPECT_BH(bh, !buffer_revoked(bh), "inconsistent data on disk")) { if (!bh_in) brelse(bh); return -EIO; } set_buffer_revoked(bh); set_buffer_revokevalid(bh); if (bh_in) { BUFFER_TRACE(bh_in, "call journal_forget"); journal_forget(handle, bh_in); } else { BUFFER_TRACE(bh, "call brelse"); __brelse(bh); } } |
9c28cbcce jbd: Journal bloc... |
397 398 |
jbd_debug(2, "insert revoke for block %u, bh_in=%p ", blocknr, bh_in); |
1da177e4c Linux-2.6.12-rc2 |
399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 |
err = insert_revoke_hash(journal, blocknr, handle->h_transaction->t_tid); BUFFER_TRACE(bh_in, "exit"); return err; } /* * Cancel an outstanding revoke. For use only internally by the * journaling code (called from journal_get_write_access). * * We trust buffer_revoked() on the buffer if the buffer is already * being journaled: if there is no revoke pending on the buffer, then we * don't do anything here. * * This would break if it were possible for a buffer to be revoked and * discarded, and then reallocated within the same transaction. In such * a case we would have lost the revoked bit, but when we arrived here * the second time we would still have a pending revoke to cancel. So, * do not trust the Revoked bit on buffers unless RevokeValid is also * set. |
1da177e4c Linux-2.6.12-rc2 |
419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 |
*/ int journal_cancel_revoke(handle_t *handle, struct journal_head *jh) { struct jbd_revoke_record_s *record; journal_t *journal = handle->h_transaction->t_journal; int need_cancel; int did_revoke = 0; /* akpm: debug */ struct buffer_head *bh = jh2bh(jh); jbd_debug(4, "journal_head %p, cancelling revoke ", jh); /* Is the existing Revoke bit valid? If so, we trust it, and * only perform the full cancel if the revoke bit is set. If * not, we can't trust the revoke bit, and we need to do the * full search for a revoke record. */ if (test_set_buffer_revokevalid(bh)) { need_cancel = test_clear_buffer_revoked(bh); } else { need_cancel = 1; clear_buffer_revoked(bh); } if (need_cancel) { record = find_revoke_record(journal, bh->b_blocknr); if (record) { jbd_debug(4, "cancelled existing revoke on " "blocknr %llu ", (unsigned long long)bh->b_blocknr); spin_lock(&journal->j_revoke_lock); list_del(&record->hash); spin_unlock(&journal->j_revoke_lock); kmem_cache_free(revoke_record_cache, record); did_revoke = 1; } } #ifdef JBD_EXPENSIVE_CHECKING /* There better not be one left behind by now! */ record = find_revoke_record(journal, bh->b_blocknr); J_ASSERT_JH(jh, record == NULL); #endif /* Finally, have we just cleared revoke on an unhashed * buffer_head? If so, we'd better make sure we clear the * revoked status on any hashed alias too, otherwise the revoke * state machine will get very upset later on. */ if (need_cancel) { struct buffer_head *bh2; bh2 = __find_get_block(bh->b_bdev, bh->b_blocknr, bh->b_size); if (bh2) { if (bh2 != bh) clear_buffer_revoked(bh2); __brelse(bh2); } } return did_revoke; } |
8c111b3f5 jbd: clear revoke... |
477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 |
/* * journal_clear_revoked_flags clears revoked flag of buffers in * revoke table to reflect there is no revoked buffer in the next * transaction which is going to be started. */ void journal_clear_buffer_revoked_flags(journal_t *journal) { struct jbd_revoke_table_s *revoke = journal->j_revoke; int i = 0; for (i = 0; i < revoke->hash_size; i++) { struct list_head *hash_list; struct list_head *list_entry; hash_list = &revoke->hash_table[i]; list_for_each(list_entry, hash_list) { struct jbd_revoke_record_s *record; struct buffer_head *bh; record = (struct jbd_revoke_record_s *)list_entry; bh = __find_get_block(journal->j_fs_dev, record->blocknr, journal->j_blocksize); if (bh) { clear_buffer_revoked(bh); __brelse(bh); } } } } |
1da177e4c Linux-2.6.12-rc2 |
506 507 508 509 510 511 512 513 514 515 516 517 |
/* journal_switch_revoke table select j_revoke for next transaction * we do not want to suspend any processing until all revokes are * written -bzzz */ void journal_switch_revoke_table(journal_t *journal) { int i; if (journal->j_revoke == journal->j_revoke_table[0]) journal->j_revoke = journal->j_revoke_table[1]; else journal->j_revoke = journal->j_revoke_table[0]; |
ae6ddcc5f [PATCH] ext3 and ... |
518 |
for (i = 0; i < journal->j_revoke->hash_size; i++) |
1da177e4c Linux-2.6.12-rc2 |
519 520 521 522 523 524 |
INIT_LIST_HEAD(&journal->j_revoke->hash_table[i]); } /* * Write revoke records to the journal for all entries in the current * revoke hash, deleting the entries as we go. |
1da177e4c Linux-2.6.12-rc2 |
525 |
*/ |
ae6ddcc5f [PATCH] ext3 and ... |
526 |
void journal_write_revoke_records(journal_t *journal, |
38d726d15 jbd: use SWRITE_S... |
527 |
transaction_t *transaction, int write_op) |
1da177e4c Linux-2.6.12-rc2 |
528 529 530 531 532 533 |
{ struct journal_head *descriptor; struct jbd_revoke_record_s *record; struct jbd_revoke_table_s *revoke; struct list_head *hash_list; int i, offset, count; |
ae6ddcc5f [PATCH] ext3 and ... |
534 |
descriptor = NULL; |
1da177e4c Linux-2.6.12-rc2 |
535 536 537 538 539 540 541 542 543 544 545 |
offset = 0; count = 0; /* select revoke table for committing transaction */ revoke = journal->j_revoke == journal->j_revoke_table[0] ? journal->j_revoke_table[1] : journal->j_revoke_table[0]; for (i = 0; i < revoke->hash_size; i++) { hash_list = &revoke->hash_table[i]; while (!list_empty(hash_list)) { |
ae6ddcc5f [PATCH] ext3 and ... |
546 |
record = (struct jbd_revoke_record_s *) |
1da177e4c Linux-2.6.12-rc2 |
547 548 |
hash_list->next; write_one_revoke_record(journal, transaction, |
ae6ddcc5f [PATCH] ext3 and ... |
549 |
&descriptor, &offset, |
38d726d15 jbd: use SWRITE_S... |
550 |
record, write_op); |
1da177e4c Linux-2.6.12-rc2 |
551 552 553 554 555 556 |
count++; list_del(&record->hash); kmem_cache_free(revoke_record_cache, record); } } if (descriptor) |
38d726d15 jbd: use SWRITE_S... |
557 |
flush_descriptor(journal, descriptor, offset, write_op); |
1da177e4c Linux-2.6.12-rc2 |
558 559 560 |
jbd_debug(1, "Wrote %d revoke records ", count); } |
ae6ddcc5f [PATCH] ext3 and ... |
561 |
/* |
1da177e4c Linux-2.6.12-rc2 |
562 |
* Write out one revoke record. We need to create a new descriptor |
ae6ddcc5f [PATCH] ext3 and ... |
563 |
* block if the old one is full or if we have not already created one. |
1da177e4c Linux-2.6.12-rc2 |
564 |
*/ |
ae6ddcc5f [PATCH] ext3 and ... |
565 |
static void write_one_revoke_record(journal_t *journal, |
1da177e4c Linux-2.6.12-rc2 |
566 |
transaction_t *transaction, |
ae6ddcc5f [PATCH] ext3 and ... |
567 |
struct journal_head **descriptorp, |
1da177e4c Linux-2.6.12-rc2 |
568 |
int *offsetp, |
38d726d15 jbd: use SWRITE_S... |
569 570 |
struct jbd_revoke_record_s *record, int write_op) |
1da177e4c Linux-2.6.12-rc2 |
571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 |
{ struct journal_head *descriptor; int offset; journal_header_t *header; /* If we are already aborting, this all becomes a noop. We still need to go round the loop in journal_write_revoke_records in order to free all of the revoke records: only the IO to the journal is omitted. */ if (is_journal_aborted(journal)) return; descriptor = *descriptorp; offset = *offsetp; /* Make sure we have a descriptor with space left for the record */ if (descriptor) { if (offset == journal->j_blocksize) { |
38d726d15 jbd: use SWRITE_S... |
589 |
flush_descriptor(journal, descriptor, offset, write_op); |
1da177e4c Linux-2.6.12-rc2 |
590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 |
descriptor = NULL; } } if (!descriptor) { descriptor = journal_get_descriptor_buffer(journal); if (!descriptor) return; header = (journal_header_t *) &jh2bh(descriptor)->b_data[0]; header->h_magic = cpu_to_be32(JFS_MAGIC_NUMBER); header->h_blocktype = cpu_to_be32(JFS_REVOKE_BLOCK); header->h_sequence = cpu_to_be32(transaction->t_tid); /* Record it so that we can wait for IO completion later */ JBUFFER_TRACE(descriptor, "file as BJ_LogCtl"); journal_file_buffer(descriptor, transaction, BJ_LogCtl); offset = sizeof(journal_revoke_header_t); *descriptorp = descriptor; } |
ae6ddcc5f [PATCH] ext3 and ... |
610 |
* ((__be32 *)(&jh2bh(descriptor)->b_data[offset])) = |
1da177e4c Linux-2.6.12-rc2 |
611 612 613 614 |
cpu_to_be32(record->blocknr); offset += 4; *offsetp = offset; } |
ae6ddcc5f [PATCH] ext3 and ... |
615 |
/* |
1da177e4c Linux-2.6.12-rc2 |
616 617 618 619 620 |
* Flush a revoke descriptor out to the journal. If we are aborting, * this is a noop; otherwise we are generating a buffer which needs to * be waited for during commit, so it has to go onto the appropriate * journal buffer list. */ |
ae6ddcc5f [PATCH] ext3 and ... |
621 622 |
static void flush_descriptor(journal_t *journal, struct journal_head *descriptor, |
38d726d15 jbd: use SWRITE_S... |
623 |
int offset, int write_op) |
1da177e4c Linux-2.6.12-rc2 |
624 625 626 627 628 629 630 631 632 633 634 635 636 637 |
{ journal_revoke_header_t *header; struct buffer_head *bh = jh2bh(descriptor); if (is_journal_aborted(journal)) { put_bh(bh); return; } header = (journal_revoke_header_t *) jh2bh(descriptor)->b_data; header->r_count = cpu_to_be32(offset); set_buffer_jwrite(bh); BUFFER_TRACE(bh, "write"); set_buffer_dirty(bh); |
9cb569d60 remove SWRITE* I/... |
638 |
write_dirty_buffer(bh, write_op); |
1da177e4c Linux-2.6.12-rc2 |
639 640 |
} #endif |
ae6ddcc5f [PATCH] ext3 and ... |
641 |
/* |
1da177e4c Linux-2.6.12-rc2 |
642 643 644 645 646 647 648 649 650 651 |
* Revoke support for recovery. * * Recovery needs to be able to: * * record all revoke records, including the tid of the latest instance * of each revoke in the journal * * check whether a given block in a given transaction should be replayed * (ie. has not been revoked by a revoke record in that or a subsequent * transaction) |
ae6ddcc5f [PATCH] ext3 and ... |
652 |
* |
1da177e4c Linux-2.6.12-rc2 |
653 654 655 656 657 658 659 |
* empty the revoke table after recovery. */ /* * First, setting revoke records. We create a new revoke record for * every block ever revoked in the log as we scan it for recovery, and * we update the existing records if we find multiple revokes for a |
ae6ddcc5f [PATCH] ext3 and ... |
660 |
* single block. |
1da177e4c Linux-2.6.12-rc2 |
661 |
*/ |
ae6ddcc5f [PATCH] ext3 and ... |
662 |
int journal_set_revoke(journal_t *journal, |
9c28cbcce jbd: Journal bloc... |
663 |
unsigned int blocknr, |
1da177e4c Linux-2.6.12-rc2 |
664 665 666 667 668 669 670 671 672 673 674 |
tid_t sequence) { struct jbd_revoke_record_s *record; record = find_revoke_record(journal, blocknr); if (record) { /* If we have multiple occurrences, only record the * latest sequence number in the hashed record */ if (tid_gt(sequence, record->sequence)) record->sequence = sequence; return 0; |
ae6ddcc5f [PATCH] ext3 and ... |
675 |
} |
1da177e4c Linux-2.6.12-rc2 |
676 677 |
return insert_revoke_hash(journal, blocknr, sequence); } |
ae6ddcc5f [PATCH] ext3 and ... |
678 |
/* |
1da177e4c Linux-2.6.12-rc2 |
679 680 681 682 683 |
* Test revoke records. For a given block referenced in the log, has * that block been revoked? A revoke record with a given transaction * sequence number revokes all blocks in that transaction and earlier * ones, but later transactions still need replayed. */ |
ae6ddcc5f [PATCH] ext3 and ... |
684 |
int journal_test_revoke(journal_t *journal, |
9c28cbcce jbd: Journal bloc... |
685 |
unsigned int blocknr, |
1da177e4c Linux-2.6.12-rc2 |
686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 |
tid_t sequence) { struct jbd_revoke_record_s *record; record = find_revoke_record(journal, blocknr); if (!record) return 0; if (tid_gt(sequence, record->sequence)) return 0; return 1; } /* * Finally, once recovery is over, we need to clear the revoke table so * that it can be reused by the running filesystem. */ void journal_clear_revoke(journal_t *journal) { int i; struct list_head *hash_list; struct jbd_revoke_record_s *record; struct jbd_revoke_table_s *revoke; revoke = journal->j_revoke; for (i = 0; i < revoke->hash_size; i++) { hash_list = &revoke->hash_table[i]; while (!list_empty(hash_list)) { record = (struct jbd_revoke_record_s*) hash_list->next; list_del(&record->hash); kmem_cache_free(revoke_record_cache, record); } } } |