1 /* -*- mode: c; c-basic-offset: 8; -*- 2 * vim: noexpandtab sw=8 ts=8 sts=0: 3 * 4 * journal.c 5 * 6 * Defines functions of journalling api 7 * 8 * Copyright (C) 2003, 2004 Oracle. All rights reserved. 9 * 10 * This program is free software; you can redistribute it and/or 11 * modify it under the terms of the GNU General Public 12 * License as published by the Free Software Foundation; either 13 * version 2 of the License, or (at your option) any later version. 14 * 15 * This program is distributed in the hope that it will be useful, 16 * but WITHOUT ANY WARRANTY; without even the implied warranty of 17 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU 18 * General Public License for more details. 19 * 20 * You should have received a copy of the GNU General Public 21 * License along with this program; if not, write to the 22 * Free Software Foundation, Inc., 59 Temple Place - Suite 330, 23 * Boston, MA 021110-1307, USA. 24 */ 25 26 #include <linux/fs.h> 27 #include <linux/types.h> 28 #include <linux/slab.h> 29 #include <linux/highmem.h> 30 #include <linux/kthread.h> 31 #include <linux/time.h> 32 #include <linux/random.h> 33 34 #include <cluster/masklog.h> 35 36 #include "ocfs2.h" 37 38 #include "alloc.h" 39 #include "blockcheck.h" 40 #include "dir.h" 41 #include "dlmglue.h" 42 #include "extent_map.h" 43 #include "heartbeat.h" 44 #include "inode.h" 45 #include "journal.h" 46 #include "localalloc.h" 47 #include "slot_map.h" 48 #include "super.h" 49 #include "sysfile.h" 50 #include "uptodate.h" 51 #include "quota.h" 52 53 #include "buffer_head_io.h" 54 #include "ocfs2_trace.h" 55 56 DEFINE_SPINLOCK(trans_inc_lock); 57 58 #define ORPHAN_SCAN_SCHEDULE_TIMEOUT 300000 59 60 static int ocfs2_force_read_journal(struct inode *inode); 61 static int ocfs2_recover_node(struct ocfs2_super *osb, 62 int node_num, int slot_num); 63 static int __ocfs2_recovery_thread(void *arg); 64 static int ocfs2_commit_cache(struct ocfs2_super *osb); 65 static int __ocfs2_wait_on_mount(struct ocfs2_super *osb, int quota); 66 static int ocfs2_journal_toggle_dirty(struct ocfs2_super *osb, 67 int dirty, int replayed); 68 static int ocfs2_trylock_journal(struct ocfs2_super *osb, 69 int slot_num); 70 static int ocfs2_recover_orphans(struct ocfs2_super *osb, 71 int slot); 72 static int ocfs2_commit_thread(void *arg); 73 static void ocfs2_queue_recovery_completion(struct ocfs2_journal *journal, 74 int slot_num, 75 struct ocfs2_dinode *la_dinode, 76 struct ocfs2_dinode *tl_dinode, 77 struct ocfs2_quota_recovery *qrec); 78 79 static inline int ocfs2_wait_on_mount(struct ocfs2_super *osb) 80 { 81 return __ocfs2_wait_on_mount(osb, 0); 82 } 83 84 static inline int ocfs2_wait_on_quotas(struct ocfs2_super *osb) 85 { 86 return __ocfs2_wait_on_mount(osb, 1); 87 } 88 89 /* 90 * This replay_map is to track online/offline slots, so we could recover 91 * offline slots during recovery and mount 92 */ 93 94 enum ocfs2_replay_state { 95 REPLAY_UNNEEDED = 0, /* Replay is not needed, so ignore this map */ 96 REPLAY_NEEDED, /* Replay slots marked in rm_replay_slots */ 97 REPLAY_DONE /* Replay was already queued */ 98 }; 99 100 struct ocfs2_replay_map { 101 unsigned int rm_slots; 102 enum ocfs2_replay_state rm_state; 103 unsigned char rm_replay_slots[0]; 104 }; 105 106 void ocfs2_replay_map_set_state(struct ocfs2_super *osb, int state) 107 { 108 if (!osb->replay_map) 109 return; 110 111 /* If we've already queued the replay, we don't have any more to do */ 112 if (osb->replay_map->rm_state == REPLAY_DONE) 113 return; 114 115 osb->replay_map->rm_state = state; 116 } 117 118 int ocfs2_compute_replay_slots(struct ocfs2_super *osb) 119 { 120 struct ocfs2_replay_map *replay_map; 121 int i, node_num; 122 123 /* If replay map is already set, we don't do it again */ 124 if (osb->replay_map) 125 return 0; 126 127 replay_map = kzalloc(sizeof(struct ocfs2_replay_map) + 128 (osb->max_slots * sizeof(char)), GFP_KERNEL); 129 130 if (!replay_map) { 131 mlog_errno(-ENOMEM); 132 return -ENOMEM; 133 } 134 135 spin_lock(&osb->osb_lock); 136 137 replay_map->rm_slots = osb->max_slots; 138 replay_map->rm_state = REPLAY_UNNEEDED; 139 140 /* set rm_replay_slots for offline slot(s) */ 141 for (i = 0; i < replay_map->rm_slots; i++) { 142 if (ocfs2_slot_to_node_num_locked(osb, i, &node_num) == -ENOENT) 143 replay_map->rm_replay_slots[i] = 1; 144 } 145 146 osb->replay_map = replay_map; 147 spin_unlock(&osb->osb_lock); 148 return 0; 149 } 150 151 void ocfs2_queue_replay_slots(struct ocfs2_super *osb) 152 { 153 struct ocfs2_replay_map *replay_map = osb->replay_map; 154 int i; 155 156 if (!replay_map) 157 return; 158 159 if (replay_map->rm_state != REPLAY_NEEDED) 160 return; 161 162 for (i = 0; i < replay_map->rm_slots; i++) 163 if (replay_map->rm_replay_slots[i]) 164 ocfs2_queue_recovery_completion(osb->journal, i, NULL, 165 NULL, NULL); 166 replay_map->rm_state = REPLAY_DONE; 167 } 168 169 void ocfs2_free_replay_slots(struct ocfs2_super *osb) 170 { 171 struct ocfs2_replay_map *replay_map = osb->replay_map; 172 173 if (!osb->replay_map) 174 return; 175 176 kfree(replay_map); 177 osb->replay_map = NULL; 178 } 179 180 int ocfs2_recovery_init(struct ocfs2_super *osb) 181 { 182 struct ocfs2_recovery_map *rm; 183 184 mutex_init(&osb->recovery_lock); 185 osb->disable_recovery = 0; 186 osb->recovery_thread_task = NULL; 187 init_waitqueue_head(&osb->recovery_event); 188 189 rm = kzalloc(sizeof(struct ocfs2_recovery_map) + 190 osb->max_slots * sizeof(unsigned int), 191 GFP_KERNEL); 192 if (!rm) { 193 mlog_errno(-ENOMEM); 194 return -ENOMEM; 195 } 196 197 rm->rm_entries = (unsigned int *)((char *)rm + 198 sizeof(struct ocfs2_recovery_map)); 199 osb->recovery_map = rm; 200 201 return 0; 202 } 203 204 /* we can't grab the goofy sem lock from inside wait_event, so we use 205 * memory barriers to make sure that we'll see the null task before 206 * being woken up */ 207 static int ocfs2_recovery_thread_running(struct ocfs2_super *osb) 208 { 209 mb(); 210 return osb->recovery_thread_task != NULL; 211 } 212 213 void ocfs2_recovery_exit(struct ocfs2_super *osb) 214 { 215 struct ocfs2_recovery_map *rm; 216 217 /* disable any new recovery threads and wait for any currently 218 * running ones to exit. Do this before setting the vol_state. */ 219 mutex_lock(&osb->recovery_lock); 220 osb->disable_recovery = 1; 221 mutex_unlock(&osb->recovery_lock); 222 wait_event(osb->recovery_event, !ocfs2_recovery_thread_running(osb)); 223 224 /* At this point, we know that no more recovery threads can be 225 * launched, so wait for any recovery completion work to 226 * complete. */ 227 flush_workqueue(ocfs2_wq); 228 229 /* 230 * Now that recovery is shut down, and the osb is about to be 231 * freed, the osb_lock is not taken here. 232 */ 233 rm = osb->recovery_map; 234 /* XXX: Should we bug if there are dirty entries? */ 235 236 kfree(rm); 237 } 238 239 static int __ocfs2_recovery_map_test(struct ocfs2_super *osb, 240 unsigned int node_num) 241 { 242 int i; 243 struct ocfs2_recovery_map *rm = osb->recovery_map; 244 245 assert_spin_locked(&osb->osb_lock); 246 247 for (i = 0; i < rm->rm_used; i++) { 248 if (rm->rm_entries[i] == node_num) 249 return 1; 250 } 251 252 return 0; 253 } 254 255 /* Behaves like test-and-set. Returns the previous value */ 256 static int ocfs2_recovery_map_set(struct ocfs2_super *osb, 257 unsigned int node_num) 258 { 259 struct ocfs2_recovery_map *rm = osb->recovery_map; 260 261 spin_lock(&osb->osb_lock); 262 if (__ocfs2_recovery_map_test(osb, node_num)) { 263 spin_unlock(&osb->osb_lock); 264 return 1; 265 } 266 267 /* XXX: Can this be exploited? Not from o2dlm... */ 268 BUG_ON(rm->rm_used >= osb->max_slots); 269 270 rm->rm_entries[rm->rm_used] = node_num; 271 rm->rm_used++; 272 spin_unlock(&osb->osb_lock); 273 274 return 0; 275 } 276 277 static void ocfs2_recovery_map_clear(struct ocfs2_super *osb, 278 unsigned int node_num) 279 { 280 int i; 281 struct ocfs2_recovery_map *rm = osb->recovery_map; 282 283 spin_lock(&osb->osb_lock); 284 285 for (i = 0; i < rm->rm_used; i++) { 286 if (rm->rm_entries[i] == node_num) 287 break; 288 } 289 290 if (i < rm->rm_used) { 291 /* XXX: be careful with the pointer math */ 292 memmove(&(rm->rm_entries[i]), &(rm->rm_entries[i + 1]), 293 (rm->rm_used - i - 1) * sizeof(unsigned int)); 294 rm->rm_used--; 295 } 296 297 spin_unlock(&osb->osb_lock); 298 } 299 300 static int ocfs2_commit_cache(struct ocfs2_super *osb) 301 { 302 int status = 0; 303 unsigned int flushed; 304 struct ocfs2_journal *journal = NULL; 305 306 journal = osb->journal; 307 308 /* Flush all pending commits and checkpoint the journal. */ 309 down_write(&journal->j_trans_barrier); 310 311 flushed = atomic_read(&journal->j_num_trans); 312 trace_ocfs2_commit_cache_begin(flushed); 313 if (flushed == 0) { 314 up_write(&journal->j_trans_barrier); 315 goto finally; 316 } 317 318 jbd2_journal_lock_updates(journal->j_journal); 319 status = jbd2_journal_flush(journal->j_journal); 320 jbd2_journal_unlock_updates(journal->j_journal); 321 if (status < 0) { 322 up_write(&journal->j_trans_barrier); 323 mlog_errno(status); 324 goto finally; 325 } 326 327 ocfs2_inc_trans_id(journal); 328 329 flushed = atomic_read(&journal->j_num_trans); 330 atomic_set(&journal->j_num_trans, 0); 331 up_write(&journal->j_trans_barrier); 332 333 trace_ocfs2_commit_cache_end(journal->j_trans_id, flushed); 334 335 ocfs2_wake_downconvert_thread(osb); 336 wake_up(&journal->j_checkpointed); 337 finally: 338 return status; 339 } 340 341 handle_t *ocfs2_start_trans(struct ocfs2_super *osb, int max_buffs) 342 { 343 journal_t *journal = osb->journal->j_journal; 344 handle_t *handle; 345 346 BUG_ON(!osb || !osb->journal->j_journal); 347 348 if (ocfs2_is_hard_readonly(osb)) 349 return ERR_PTR(-EROFS); 350 351 BUG_ON(osb->journal->j_state == OCFS2_JOURNAL_FREE); 352 BUG_ON(max_buffs <= 0); 353 354 /* Nested transaction? Just return the handle... */ 355 if (journal_current_handle()) 356 return jbd2_journal_start(journal, max_buffs); 357 358 sb_start_intwrite(osb->sb); 359 360 down_read(&osb->journal->j_trans_barrier); 361 362 handle = jbd2_journal_start(journal, max_buffs); 363 if (IS_ERR(handle)) { 364 up_read(&osb->journal->j_trans_barrier); 365 sb_end_intwrite(osb->sb); 366 367 mlog_errno(PTR_ERR(handle)); 368 369 if (is_journal_aborted(journal)) { 370 ocfs2_abort(osb->sb, "Detected aborted journal"); 371 handle = ERR_PTR(-EROFS); 372 } 373 } else { 374 if (!ocfs2_mount_local(osb)) 375 atomic_inc(&(osb->journal->j_num_trans)); 376 } 377 378 return handle; 379 } 380 381 int ocfs2_commit_trans(struct ocfs2_super *osb, 382 handle_t *handle) 383 { 384 int ret, nested; 385 struct ocfs2_journal *journal = osb->journal; 386 387 BUG_ON(!handle); 388 389 nested = handle->h_ref > 1; 390 ret = jbd2_journal_stop(handle); 391 if (ret < 0) 392 mlog_errno(ret); 393 394 if (!nested) { 395 up_read(&journal->j_trans_barrier); 396 sb_end_intwrite(osb->sb); 397 } 398 399 return ret; 400 } 401 402 /* 403 * 'nblocks' is what you want to add to the current transaction. 404 * 405 * This might call jbd2_journal_restart() which will commit dirty buffers 406 * and then restart the transaction. Before calling 407 * ocfs2_extend_trans(), any changed blocks should have been 408 * dirtied. After calling it, all blocks which need to be changed must 409 * go through another set of journal_access/journal_dirty calls. 410 * 411 * WARNING: This will not release any semaphores or disk locks taken 412 * during the transaction, so make sure they were taken *before* 413 * start_trans or we'll have ordering deadlocks. 414 * 415 * WARNING2: Note that we do *not* drop j_trans_barrier here. This is 416 * good because transaction ids haven't yet been recorded on the 417 * cluster locks associated with this handle. 418 */ 419 int ocfs2_extend_trans(handle_t *handle, int nblocks) 420 { 421 int status, old_nblocks; 422 423 BUG_ON(!handle); 424 BUG_ON(nblocks < 0); 425 426 if (!nblocks) 427 return 0; 428 429 old_nblocks = handle->h_buffer_credits; 430 431 trace_ocfs2_extend_trans(old_nblocks, nblocks); 432 433 #ifdef CONFIG_OCFS2_DEBUG_FS 434 status = 1; 435 #else 436 status = jbd2_journal_extend(handle, nblocks); 437 if (status < 0) { 438 mlog_errno(status); 439 goto bail; 440 } 441 #endif 442 443 if (status > 0) { 444 trace_ocfs2_extend_trans_restart(old_nblocks + nblocks); 445 status = jbd2_journal_restart(handle, 446 old_nblocks + nblocks); 447 if (status < 0) { 448 mlog_errno(status); 449 goto bail; 450 } 451 } 452 453 status = 0; 454 bail: 455 return status; 456 } 457 458 struct ocfs2_triggers { 459 struct jbd2_buffer_trigger_type ot_triggers; 460 int ot_offset; 461 }; 462 463 static inline struct ocfs2_triggers *to_ocfs2_trigger(struct jbd2_buffer_trigger_type *triggers) 464 { 465 return container_of(triggers, struct ocfs2_triggers, ot_triggers); 466 } 467 468 static void ocfs2_frozen_trigger(struct jbd2_buffer_trigger_type *triggers, 469 struct buffer_head *bh, 470 void *data, size_t size) 471 { 472 struct ocfs2_triggers *ot = to_ocfs2_trigger(triggers); 473 474 /* 475 * We aren't guaranteed to have the superblock here, so we 476 * must unconditionally compute the ecc data. 477 * __ocfs2_journal_access() will only set the triggers if 478 * metaecc is enabled. 479 */ 480 ocfs2_block_check_compute(data, size, data + ot->ot_offset); 481 } 482 483 /* 484 * Quota blocks have their own trigger because the struct ocfs2_block_check 485 * offset depends on the blocksize. 486 */ 487 static void ocfs2_dq_frozen_trigger(struct jbd2_buffer_trigger_type *triggers, 488 struct buffer_head *bh, 489 void *data, size_t size) 490 { 491 struct ocfs2_disk_dqtrailer *dqt = 492 ocfs2_block_dqtrailer(size, data); 493 494 /* 495 * We aren't guaranteed to have the superblock here, so we 496 * must unconditionally compute the ecc data. 497 * __ocfs2_journal_access() will only set the triggers if 498 * metaecc is enabled. 499 */ 500 ocfs2_block_check_compute(data, size, &dqt->dq_check); 501 } 502 503 /* 504 * Directory blocks also have their own trigger because the 505 * struct ocfs2_block_check offset depends on the blocksize. 506 */ 507 static void ocfs2_db_frozen_trigger(struct jbd2_buffer_trigger_type *triggers, 508 struct buffer_head *bh, 509 void *data, size_t size) 510 { 511 struct ocfs2_dir_block_trailer *trailer = 512 ocfs2_dir_trailer_from_size(size, data); 513 514 /* 515 * We aren't guaranteed to have the superblock here, so we 516 * must unconditionally compute the ecc data. 517 * __ocfs2_journal_access() will only set the triggers if 518 * metaecc is enabled. 519 */ 520 ocfs2_block_check_compute(data, size, &trailer->db_check); 521 } 522 523 static void ocfs2_abort_trigger(struct jbd2_buffer_trigger_type *triggers, 524 struct buffer_head *bh) 525 { 526 mlog(ML_ERROR, 527 "ocfs2_abort_trigger called by JBD2. bh = 0x%lx, " 528 "bh->b_blocknr = %llu\n", 529 (unsigned long)bh, 530 (unsigned long long)bh->b_blocknr); 531 532 /* We aren't guaranteed to have the superblock here - but if we 533 * don't, it'll just crash. */ 534 ocfs2_error(bh->b_assoc_map->host->i_sb, 535 "JBD2 has aborted our journal, ocfs2 cannot continue\n"); 536 } 537 538 static struct ocfs2_triggers di_triggers = { 539 .ot_triggers = { 540 .t_frozen = ocfs2_frozen_trigger, 541 .t_abort = ocfs2_abort_trigger, 542 }, 543 .ot_offset = offsetof(struct ocfs2_dinode, i_check), 544 }; 545 546 static struct ocfs2_triggers eb_triggers = { 547 .ot_triggers = { 548 .t_frozen = ocfs2_frozen_trigger, 549 .t_abort = ocfs2_abort_trigger, 550 }, 551 .ot_offset = offsetof(struct ocfs2_extent_block, h_check), 552 }; 553 554 static struct ocfs2_triggers rb_triggers = { 555 .ot_triggers = { 556 .t_frozen = ocfs2_frozen_trigger, 557 .t_abort = ocfs2_abort_trigger, 558 }, 559 .ot_offset = offsetof(struct ocfs2_refcount_block, rf_check), 560 }; 561 562 static struct ocfs2_triggers gd_triggers = { 563 .ot_triggers = { 564 .t_frozen = ocfs2_frozen_trigger, 565 .t_abort = ocfs2_abort_trigger, 566 }, 567 .ot_offset = offsetof(struct ocfs2_group_desc, bg_check), 568 }; 569 570 static struct ocfs2_triggers db_triggers = { 571 .ot_triggers = { 572 .t_frozen = ocfs2_db_frozen_trigger, 573 .t_abort = ocfs2_abort_trigger, 574 }, 575 }; 576 577 static struct ocfs2_triggers xb_triggers = { 578 .ot_triggers = { 579 .t_frozen = ocfs2_frozen_trigger, 580 .t_abort = ocfs2_abort_trigger, 581 }, 582 .ot_offset = offsetof(struct ocfs2_xattr_block, xb_check), 583 }; 584 585 static struct ocfs2_triggers dq_triggers = { 586 .ot_triggers = { 587 .t_frozen = ocfs2_dq_frozen_trigger, 588 .t_abort = ocfs2_abort_trigger, 589 }, 590 }; 591 592 static struct ocfs2_triggers dr_triggers = { 593 .ot_triggers = { 594 .t_frozen = ocfs2_frozen_trigger, 595 .t_abort = ocfs2_abort_trigger, 596 }, 597 .ot_offset = offsetof(struct ocfs2_dx_root_block, dr_check), 598 }; 599 600 static struct ocfs2_triggers dl_triggers = { 601 .ot_triggers = { 602 .t_frozen = ocfs2_frozen_trigger, 603 .t_abort = ocfs2_abort_trigger, 604 }, 605 .ot_offset = offsetof(struct ocfs2_dx_leaf, dl_check), 606 }; 607 608 static int __ocfs2_journal_access(handle_t *handle, 609 struct ocfs2_caching_info *ci, 610 struct buffer_head *bh, 611 struct ocfs2_triggers *triggers, 612 int type) 613 { 614 int status; 615 struct ocfs2_super *osb = 616 OCFS2_SB(ocfs2_metadata_cache_get_super(ci)); 617 618 BUG_ON(!ci || !ci->ci_ops); 619 BUG_ON(!handle); 620 BUG_ON(!bh); 621 622 trace_ocfs2_journal_access( 623 (unsigned long long)ocfs2_metadata_cache_owner(ci), 624 (unsigned long long)bh->b_blocknr, type, bh->b_size); 625 626 /* we can safely remove this assertion after testing. */ 627 if (!buffer_uptodate(bh)) { 628 mlog(ML_ERROR, "giving me a buffer that's not uptodate!\n"); 629 mlog(ML_ERROR, "b_blocknr=%llu\n", 630 (unsigned long long)bh->b_blocknr); 631 BUG(); 632 } 633 634 /* Set the current transaction information on the ci so 635 * that the locking code knows whether it can drop it's locks 636 * on this ci or not. We're protected from the commit 637 * thread updating the current transaction id until 638 * ocfs2_commit_trans() because ocfs2_start_trans() took 639 * j_trans_barrier for us. */ 640 ocfs2_set_ci_lock_trans(osb->journal, ci); 641 642 ocfs2_metadata_cache_io_lock(ci); 643 switch (type) { 644 case OCFS2_JOURNAL_ACCESS_CREATE: 645 case OCFS2_JOURNAL_ACCESS_WRITE: 646 status = jbd2_journal_get_write_access(handle, bh); 647 break; 648 649 case OCFS2_JOURNAL_ACCESS_UNDO: 650 status = jbd2_journal_get_undo_access(handle, bh); 651 break; 652 653 default: 654 status = -EINVAL; 655 mlog(ML_ERROR, "Unknown access type!\n"); 656 } 657 if (!status && ocfs2_meta_ecc(osb) && triggers) 658 jbd2_journal_set_triggers(bh, &triggers->ot_triggers); 659 ocfs2_metadata_cache_io_unlock(ci); 660 661 if (status < 0) 662 mlog(ML_ERROR, "Error %d getting %d access to buffer!\n", 663 status, type); 664 665 return status; 666 } 667 668 int ocfs2_journal_access_di(handle_t *handle, struct ocfs2_caching_info *ci, 669 struct buffer_head *bh, int type) 670 { 671 return __ocfs2_journal_access(handle, ci, bh, &di_triggers, type); 672 } 673 674 int ocfs2_journal_access_eb(handle_t *handle, struct ocfs2_caching_info *ci, 675 struct buffer_head *bh, int type) 676 { 677 return __ocfs2_journal_access(handle, ci, bh, &eb_triggers, type); 678 } 679 680 int ocfs2_journal_access_rb(handle_t *handle, struct ocfs2_caching_info *ci, 681 struct buffer_head *bh, int type) 682 { 683 return __ocfs2_journal_access(handle, ci, bh, &rb_triggers, 684 type); 685 } 686 687 int ocfs2_journal_access_gd(handle_t *handle, struct ocfs2_caching_info *ci, 688 struct buffer_head *bh, int type) 689 { 690 return __ocfs2_journal_access(handle, ci, bh, &gd_triggers, type); 691 } 692 693 int ocfs2_journal_access_db(handle_t *handle, struct ocfs2_caching_info *ci, 694 struct buffer_head *bh, int type) 695 { 696 return __ocfs2_journal_access(handle, ci, bh, &db_triggers, type); 697 } 698 699 int ocfs2_journal_access_xb(handle_t *handle, struct ocfs2_caching_info *ci, 700 struct buffer_head *bh, int type) 701 { 702 return __ocfs2_journal_access(handle, ci, bh, &xb_triggers, type); 703 } 704 705 int ocfs2_journal_access_dq(handle_t *handle, struct ocfs2_caching_info *ci, 706 struct buffer_head *bh, int type) 707 { 708 return __ocfs2_journal_access(handle, ci, bh, &dq_triggers, type); 709 } 710 711 int ocfs2_journal_access_dr(handle_t *handle, struct ocfs2_caching_info *ci, 712 struct buffer_head *bh, int type) 713 { 714 return __ocfs2_journal_access(handle, ci, bh, &dr_triggers, type); 715 } 716 717 int ocfs2_journal_access_dl(handle_t *handle, struct ocfs2_caching_info *ci, 718 struct buffer_head *bh, int type) 719 { 720 return __ocfs2_journal_access(handle, ci, bh, &dl_triggers, type); 721 } 722 723 int ocfs2_journal_access(handle_t *handle, struct ocfs2_caching_info *ci, 724 struct buffer_head *bh, int type) 725 { 726 return __ocfs2_journal_access(handle, ci, bh, NULL, type); 727 } 728 729 void ocfs2_journal_dirty(handle_t *handle, struct buffer_head *bh) 730 { 731 int status; 732 733 trace_ocfs2_journal_dirty((unsigned long long)bh->b_blocknr); 734 735 status = jbd2_journal_dirty_metadata(handle, bh); 736 BUG_ON(status); 737 } 738 739 #define OCFS2_DEFAULT_COMMIT_INTERVAL (HZ * JBD2_DEFAULT_MAX_COMMIT_AGE) 740 741 void ocfs2_set_journal_params(struct ocfs2_super *osb) 742 { 743 journal_t *journal = osb->journal->j_journal; 744 unsigned long commit_interval = OCFS2_DEFAULT_COMMIT_INTERVAL; 745 746 if (osb->osb_commit_interval) 747 commit_interval = osb->osb_commit_interval; 748 749 write_lock(&journal->j_state_lock); 750 journal->j_commit_interval = commit_interval; 751 if (osb->s_mount_opt & OCFS2_MOUNT_BARRIER) 752 journal->j_flags |= JBD2_BARRIER; 753 else 754 journal->j_flags &= ~JBD2_BARRIER; 755 write_unlock(&journal->j_state_lock); 756 } 757 758 int ocfs2_journal_init(struct ocfs2_journal *journal, int *dirty) 759 { 760 int status = -1; 761 struct inode *inode = NULL; /* the journal inode */ 762 journal_t *j_journal = NULL; 763 struct ocfs2_dinode *di = NULL; 764 struct buffer_head *bh = NULL; 765 struct ocfs2_super *osb; 766 int inode_lock = 0; 767 768 BUG_ON(!journal); 769 770 osb = journal->j_osb; 771 772 /* already have the inode for our journal */ 773 inode = ocfs2_get_system_file_inode(osb, JOURNAL_SYSTEM_INODE, 774 osb->slot_num); 775 if (inode == NULL) { 776 status = -EACCES; 777 mlog_errno(status); 778 goto done; 779 } 780 if (is_bad_inode(inode)) { 781 mlog(ML_ERROR, "access error (bad inode)\n"); 782 iput(inode); 783 inode = NULL; 784 status = -EACCES; 785 goto done; 786 } 787 788 SET_INODE_JOURNAL(inode); 789 OCFS2_I(inode)->ip_open_count++; 790 791 /* Skip recovery waits here - journal inode metadata never 792 * changes in a live cluster so it can be considered an 793 * exception to the rule. */ 794 status = ocfs2_inode_lock_full(inode, &bh, 1, OCFS2_META_LOCK_RECOVERY); 795 if (status < 0) { 796 if (status != -ERESTARTSYS) 797 mlog(ML_ERROR, "Could not get lock on journal!\n"); 798 goto done; 799 } 800 801 inode_lock = 1; 802 di = (struct ocfs2_dinode *)bh->b_data; 803 804 if (inode->i_size < OCFS2_MIN_JOURNAL_SIZE) { 805 mlog(ML_ERROR, "Journal file size (%lld) is too small!\n", 806 inode->i_size); 807 status = -EINVAL; 808 goto done; 809 } 810 811 trace_ocfs2_journal_init(inode->i_size, 812 (unsigned long long)inode->i_blocks, 813 OCFS2_I(inode)->ip_clusters); 814 815 /* call the kernels journal init function now */ 816 j_journal = jbd2_journal_init_inode(inode); 817 if (j_journal == NULL) { 818 mlog(ML_ERROR, "Linux journal layer error\n"); 819 status = -EINVAL; 820 goto done; 821 } 822 823 trace_ocfs2_journal_init_maxlen(j_journal->j_maxlen); 824 825 *dirty = (le32_to_cpu(di->id1.journal1.ij_flags) & 826 OCFS2_JOURNAL_DIRTY_FL); 827 828 journal->j_journal = j_journal; 829 journal->j_inode = inode; 830 journal->j_bh = bh; 831 832 ocfs2_set_journal_params(osb); 833 834 journal->j_state = OCFS2_JOURNAL_LOADED; 835 836 status = 0; 837 done: 838 if (status < 0) { 839 if (inode_lock) 840 ocfs2_inode_unlock(inode, 1); 841 brelse(bh); 842 if (inode) { 843 OCFS2_I(inode)->ip_open_count--; 844 iput(inode); 845 } 846 } 847 848 return status; 849 } 850 851 static void ocfs2_bump_recovery_generation(struct ocfs2_dinode *di) 852 { 853 le32_add_cpu(&(di->id1.journal1.ij_recovery_generation), 1); 854 } 855 856 static u32 ocfs2_get_recovery_generation(struct ocfs2_dinode *di) 857 { 858 return le32_to_cpu(di->id1.journal1.ij_recovery_generation); 859 } 860 861 static int ocfs2_journal_toggle_dirty(struct ocfs2_super *osb, 862 int dirty, int replayed) 863 { 864 int status; 865 unsigned int flags; 866 struct ocfs2_journal *journal = osb->journal; 867 struct buffer_head *bh = journal->j_bh; 868 struct ocfs2_dinode *fe; 869 870 fe = (struct ocfs2_dinode *)bh->b_data; 871 872 /* The journal bh on the osb always comes from ocfs2_journal_init() 873 * and was validated there inside ocfs2_inode_lock_full(). It's a 874 * code bug if we mess it up. */ 875 BUG_ON(!OCFS2_IS_VALID_DINODE(fe)); 876 877 flags = le32_to_cpu(fe->id1.journal1.ij_flags); 878 if (dirty) 879 flags |= OCFS2_JOURNAL_DIRTY_FL; 880 else 881 flags &= ~OCFS2_JOURNAL_DIRTY_FL; 882 fe->id1.journal1.ij_flags = cpu_to_le32(flags); 883 884 if (replayed) 885 ocfs2_bump_recovery_generation(fe); 886 887 ocfs2_compute_meta_ecc(osb->sb, bh->b_data, &fe->i_check); 888 status = ocfs2_write_block(osb, bh, INODE_CACHE(journal->j_inode)); 889 if (status < 0) 890 mlog_errno(status); 891 892 return status; 893 } 894 895 /* 896 * If the journal has been kmalloc'd it needs to be freed after this 897 * call. 898 */ 899 void ocfs2_journal_shutdown(struct ocfs2_super *osb) 900 { 901 struct ocfs2_journal *journal = NULL; 902 int status = 0; 903 struct inode *inode = NULL; 904 int num_running_trans = 0; 905 906 BUG_ON(!osb); 907 908 journal = osb->journal; 909 if (!journal) 910 goto done; 911 912 inode = journal->j_inode; 913 914 if (journal->j_state != OCFS2_JOURNAL_LOADED) 915 goto done; 916 917 /* need to inc inode use count - jbd2_journal_destroy will iput. */ 918 if (!igrab(inode)) 919 BUG(); 920 921 num_running_trans = atomic_read(&(osb->journal->j_num_trans)); 922 trace_ocfs2_journal_shutdown(num_running_trans); 923 924 /* Do a commit_cache here. It will flush our journal, *and* 925 * release any locks that are still held. 926 * set the SHUTDOWN flag and release the trans lock. 927 * the commit thread will take the trans lock for us below. */ 928 journal->j_state = OCFS2_JOURNAL_IN_SHUTDOWN; 929 930 /* The OCFS2_JOURNAL_IN_SHUTDOWN will signal to commit_cache to not 931 * drop the trans_lock (which we want to hold until we 932 * completely destroy the journal. */ 933 if (osb->commit_task) { 934 /* Wait for the commit thread */ 935 trace_ocfs2_journal_shutdown_wait(osb->commit_task); 936 kthread_stop(osb->commit_task); 937 osb->commit_task = NULL; 938 } 939 940 BUG_ON(atomic_read(&(osb->journal->j_num_trans)) != 0); 941 942 if (ocfs2_mount_local(osb)) { 943 jbd2_journal_lock_updates(journal->j_journal); 944 status = jbd2_journal_flush(journal->j_journal); 945 jbd2_journal_unlock_updates(journal->j_journal); 946 if (status < 0) 947 mlog_errno(status); 948 } 949 950 if (status == 0) { 951 /* 952 * Do not toggle if flush was unsuccessful otherwise 953 * will leave dirty metadata in a "clean" journal 954 */ 955 status = ocfs2_journal_toggle_dirty(osb, 0, 0); 956 if (status < 0) 957 mlog_errno(status); 958 } 959 960 /* Shutdown the kernel journal system */ 961 jbd2_journal_destroy(journal->j_journal); 962 journal->j_journal = NULL; 963 964 OCFS2_I(inode)->ip_open_count--; 965 966 /* unlock our journal */ 967 ocfs2_inode_unlock(inode, 1); 968 969 brelse(journal->j_bh); 970 journal->j_bh = NULL; 971 972 journal->j_state = OCFS2_JOURNAL_FREE; 973 974 // up_write(&journal->j_trans_barrier); 975 done: 976 if (inode) 977 iput(inode); 978 } 979 980 static void ocfs2_clear_journal_error(struct super_block *sb, 981 journal_t *journal, 982 int slot) 983 { 984 int olderr; 985 986 olderr = jbd2_journal_errno(journal); 987 if (olderr) { 988 mlog(ML_ERROR, "File system error %d recorded in " 989 "journal %u.\n", olderr, slot); 990 mlog(ML_ERROR, "File system on device %s needs checking.\n", 991 sb->s_id); 992 993 jbd2_journal_ack_err(journal); 994 jbd2_journal_clear_err(journal); 995 } 996 } 997 998 int ocfs2_journal_load(struct ocfs2_journal *journal, int local, int replayed) 999 { 1000 int status = 0; 1001 struct ocfs2_super *osb; 1002 1003 BUG_ON(!journal); 1004 1005 osb = journal->j_osb; 1006 1007 status = jbd2_journal_load(journal->j_journal); 1008 if (status < 0) { 1009 mlog(ML_ERROR, "Failed to load journal!\n"); 1010 goto done; 1011 } 1012 1013 ocfs2_clear_journal_error(osb->sb, journal->j_journal, osb->slot_num); 1014 1015 status = ocfs2_journal_toggle_dirty(osb, 1, replayed); 1016 if (status < 0) { 1017 mlog_errno(status); 1018 goto done; 1019 } 1020 1021 /* Launch the commit thread */ 1022 if (!local) { 1023 osb->commit_task = kthread_run(ocfs2_commit_thread, osb, 1024 "ocfs2cmt"); 1025 if (IS_ERR(osb->commit_task)) { 1026 status = PTR_ERR(osb->commit_task); 1027 osb->commit_task = NULL; 1028 mlog(ML_ERROR, "unable to launch ocfs2commit thread, " 1029 "error=%d", status); 1030 goto done; 1031 } 1032 } else 1033 osb->commit_task = NULL; 1034 1035 done: 1036 return status; 1037 } 1038 1039 1040 /* 'full' flag tells us whether we clear out all blocks or if we just 1041 * mark the journal clean */ 1042 int ocfs2_journal_wipe(struct ocfs2_journal *journal, int full) 1043 { 1044 int status; 1045 1046 BUG_ON(!journal); 1047 1048 status = jbd2_journal_wipe(journal->j_journal, full); 1049 if (status < 0) { 1050 mlog_errno(status); 1051 goto bail; 1052 } 1053 1054 status = ocfs2_journal_toggle_dirty(journal->j_osb, 0, 0); 1055 if (status < 0) 1056 mlog_errno(status); 1057 1058 bail: 1059 return status; 1060 } 1061 1062 static int ocfs2_recovery_completed(struct ocfs2_super *osb) 1063 { 1064 int empty; 1065 struct ocfs2_recovery_map *rm = osb->recovery_map; 1066 1067 spin_lock(&osb->osb_lock); 1068 empty = (rm->rm_used == 0); 1069 spin_unlock(&osb->osb_lock); 1070 1071 return empty; 1072 } 1073 1074 void ocfs2_wait_for_recovery(struct ocfs2_super *osb) 1075 { 1076 wait_event(osb->recovery_event, ocfs2_recovery_completed(osb)); 1077 } 1078 1079 /* 1080 * JBD Might read a cached version of another nodes journal file. We 1081 * don't want this as this file changes often and we get no 1082 * notification on those changes. The only way to be sure that we've 1083 * got the most up to date version of those blocks then is to force 1084 * read them off disk. Just searching through the buffer cache won't 1085 * work as there may be pages backing this file which are still marked 1086 * up to date. We know things can't change on this file underneath us 1087 * as we have the lock by now :) 1088 */ 1089 static int ocfs2_force_read_journal(struct inode *inode) 1090 { 1091 int status = 0; 1092 int i; 1093 u64 v_blkno, p_blkno, p_blocks, num_blocks; 1094 #define CONCURRENT_JOURNAL_FILL 32ULL 1095 struct buffer_head *bhs[CONCURRENT_JOURNAL_FILL]; 1096 1097 memset(bhs, 0, sizeof(struct buffer_head *) * CONCURRENT_JOURNAL_FILL); 1098 1099 num_blocks = ocfs2_blocks_for_bytes(inode->i_sb, inode->i_size); 1100 v_blkno = 0; 1101 while (v_blkno < num_blocks) { 1102 status = ocfs2_extent_map_get_blocks(inode, v_blkno, 1103 &p_blkno, &p_blocks, NULL); 1104 if (status < 0) { 1105 mlog_errno(status); 1106 goto bail; 1107 } 1108 1109 if (p_blocks > CONCURRENT_JOURNAL_FILL) 1110 p_blocks = CONCURRENT_JOURNAL_FILL; 1111 1112 /* We are reading journal data which should not 1113 * be put in the uptodate cache */ 1114 status = ocfs2_read_blocks_sync(OCFS2_SB(inode->i_sb), 1115 p_blkno, p_blocks, bhs); 1116 if (status < 0) { 1117 mlog_errno(status); 1118 goto bail; 1119 } 1120 1121 for(i = 0; i < p_blocks; i++) { 1122 brelse(bhs[i]); 1123 bhs[i] = NULL; 1124 } 1125 1126 v_blkno += p_blocks; 1127 } 1128 1129 bail: 1130 for(i = 0; i < CONCURRENT_JOURNAL_FILL; i++) 1131 brelse(bhs[i]); 1132 return status; 1133 } 1134 1135 struct ocfs2_la_recovery_item { 1136 struct list_head lri_list; 1137 int lri_slot; 1138 struct ocfs2_dinode *lri_la_dinode; 1139 struct ocfs2_dinode *lri_tl_dinode; 1140 struct ocfs2_quota_recovery *lri_qrec; 1141 }; 1142 1143 /* Does the second half of the recovery process. By this point, the 1144 * node is marked clean and can actually be considered recovered, 1145 * hence it's no longer in the recovery map, but there's still some 1146 * cleanup we can do which shouldn't happen within the recovery thread 1147 * as locking in that context becomes very difficult if we are to take 1148 * recovering nodes into account. 1149 * 1150 * NOTE: This function can and will sleep on recovery of other nodes 1151 * during cluster locking, just like any other ocfs2 process. 1152 */ 1153 void ocfs2_complete_recovery(struct work_struct *work) 1154 { 1155 int ret = 0; 1156 struct ocfs2_journal *journal = 1157 container_of(work, struct ocfs2_journal, j_recovery_work); 1158 struct ocfs2_super *osb = journal->j_osb; 1159 struct ocfs2_dinode *la_dinode, *tl_dinode; 1160 struct ocfs2_la_recovery_item *item, *n; 1161 struct ocfs2_quota_recovery *qrec; 1162 LIST_HEAD(tmp_la_list); 1163 1164 trace_ocfs2_complete_recovery( 1165 (unsigned long long)OCFS2_I(journal->j_inode)->ip_blkno); 1166 1167 spin_lock(&journal->j_lock); 1168 list_splice_init(&journal->j_la_cleanups, &tmp_la_list); 1169 spin_unlock(&journal->j_lock); 1170 1171 list_for_each_entry_safe(item, n, &tmp_la_list, lri_list) { 1172 list_del_init(&item->lri_list); 1173 1174 ocfs2_wait_on_quotas(osb); 1175 1176 la_dinode = item->lri_la_dinode; 1177 tl_dinode = item->lri_tl_dinode; 1178 qrec = item->lri_qrec; 1179 1180 trace_ocfs2_complete_recovery_slot(item->lri_slot, 1181 la_dinode ? le64_to_cpu(la_dinode->i_blkno) : 0, 1182 tl_dinode ? le64_to_cpu(tl_dinode->i_blkno) : 0, 1183 qrec); 1184 1185 if (la_dinode) { 1186 ret = ocfs2_complete_local_alloc_recovery(osb, 1187 la_dinode); 1188 if (ret < 0) 1189 mlog_errno(ret); 1190 1191 kfree(la_dinode); 1192 } 1193 1194 if (tl_dinode) { 1195 ret = ocfs2_complete_truncate_log_recovery(osb, 1196 tl_dinode); 1197 if (ret < 0) 1198 mlog_errno(ret); 1199 1200 kfree(tl_dinode); 1201 } 1202 1203 ret = ocfs2_recover_orphans(osb, item->lri_slot); 1204 if (ret < 0) 1205 mlog_errno(ret); 1206 1207 if (qrec) { 1208 ret = ocfs2_finish_quota_recovery(osb, qrec, 1209 item->lri_slot); 1210 if (ret < 0) 1211 mlog_errno(ret); 1212 /* Recovery info is already freed now */ 1213 } 1214 1215 kfree(item); 1216 } 1217 1218 trace_ocfs2_complete_recovery_end(ret); 1219 } 1220 1221 /* NOTE: This function always eats your references to la_dinode and 1222 * tl_dinode, either manually on error, or by passing them to 1223 * ocfs2_complete_recovery */ 1224 static void ocfs2_queue_recovery_completion(struct ocfs2_journal *journal, 1225 int slot_num, 1226 struct ocfs2_dinode *la_dinode, 1227 struct ocfs2_dinode *tl_dinode, 1228 struct ocfs2_quota_recovery *qrec) 1229 { 1230 struct ocfs2_la_recovery_item *item; 1231 1232 item = kmalloc(sizeof(struct ocfs2_la_recovery_item), GFP_NOFS); 1233 if (!item) { 1234 /* Though we wish to avoid it, we are in fact safe in 1235 * skipping local alloc cleanup as fsck.ocfs2 is more 1236 * than capable of reclaiming unused space. */ 1237 if (la_dinode) 1238 kfree(la_dinode); 1239 1240 if (tl_dinode) 1241 kfree(tl_dinode); 1242 1243 if (qrec) 1244 ocfs2_free_quota_recovery(qrec); 1245 1246 mlog_errno(-ENOMEM); 1247 return; 1248 } 1249 1250 INIT_LIST_HEAD(&item->lri_list); 1251 item->lri_la_dinode = la_dinode; 1252 item->lri_slot = slot_num; 1253 item->lri_tl_dinode = tl_dinode; 1254 item->lri_qrec = qrec; 1255 1256 spin_lock(&journal->j_lock); 1257 list_add_tail(&item->lri_list, &journal->j_la_cleanups); 1258 queue_work(ocfs2_wq, &journal->j_recovery_work); 1259 spin_unlock(&journal->j_lock); 1260 } 1261 1262 /* Called by the mount code to queue recovery the last part of 1263 * recovery for it's own and offline slot(s). */ 1264 void ocfs2_complete_mount_recovery(struct ocfs2_super *osb) 1265 { 1266 struct ocfs2_journal *journal = osb->journal; 1267 1268 if (ocfs2_is_hard_readonly(osb)) 1269 return; 1270 1271 /* No need to queue up our truncate_log as regular cleanup will catch 1272 * that */ 1273 ocfs2_queue_recovery_completion(journal, osb->slot_num, 1274 osb->local_alloc_copy, NULL, NULL); 1275 ocfs2_schedule_truncate_log_flush(osb, 0); 1276 1277 osb->local_alloc_copy = NULL; 1278 osb->dirty = 0; 1279 1280 /* queue to recover orphan slots for all offline slots */ 1281 ocfs2_replay_map_set_state(osb, REPLAY_NEEDED); 1282 ocfs2_queue_replay_slots(osb); 1283 ocfs2_free_replay_slots(osb); 1284 } 1285 1286 void ocfs2_complete_quota_recovery(struct ocfs2_super *osb) 1287 { 1288 if (osb->quota_rec) { 1289 ocfs2_queue_recovery_completion(osb->journal, 1290 osb->slot_num, 1291 NULL, 1292 NULL, 1293 osb->quota_rec); 1294 osb->quota_rec = NULL; 1295 } 1296 } 1297 1298 static int __ocfs2_recovery_thread(void *arg) 1299 { 1300 int status, node_num, slot_num; 1301 struct ocfs2_super *osb = arg; 1302 struct ocfs2_recovery_map *rm = osb->recovery_map; 1303 int *rm_quota = NULL; 1304 int rm_quota_used = 0, i; 1305 struct ocfs2_quota_recovery *qrec; 1306 1307 status = ocfs2_wait_on_mount(osb); 1308 if (status < 0) { 1309 goto bail; 1310 } 1311 1312 rm_quota = kzalloc(osb->max_slots * sizeof(int), GFP_NOFS); 1313 if (!rm_quota) { 1314 status = -ENOMEM; 1315 goto bail; 1316 } 1317 restart: 1318 status = ocfs2_super_lock(osb, 1); 1319 if (status < 0) { 1320 mlog_errno(status); 1321 goto bail; 1322 } 1323 1324 status = ocfs2_compute_replay_slots(osb); 1325 if (status < 0) 1326 mlog_errno(status); 1327 1328 /* queue recovery for our own slot */ 1329 ocfs2_queue_recovery_completion(osb->journal, osb->slot_num, NULL, 1330 NULL, NULL); 1331 1332 spin_lock(&osb->osb_lock); 1333 while (rm->rm_used) { 1334 /* It's always safe to remove entry zero, as we won't 1335 * clear it until ocfs2_recover_node() has succeeded. */ 1336 node_num = rm->rm_entries[0]; 1337 spin_unlock(&osb->osb_lock); 1338 slot_num = ocfs2_node_num_to_slot(osb, node_num); 1339 trace_ocfs2_recovery_thread_node(node_num, slot_num); 1340 if (slot_num == -ENOENT) { 1341 status = 0; 1342 goto skip_recovery; 1343 } 1344 1345 /* It is a bit subtle with quota recovery. We cannot do it 1346 * immediately because we have to obtain cluster locks from 1347 * quota files and we also don't want to just skip it because 1348 * then quota usage would be out of sync until some node takes 1349 * the slot. So we remember which nodes need quota recovery 1350 * and when everything else is done, we recover quotas. */ 1351 for (i = 0; i < rm_quota_used && rm_quota[i] != slot_num; i++); 1352 if (i == rm_quota_used) 1353 rm_quota[rm_quota_used++] = slot_num; 1354 1355 status = ocfs2_recover_node(osb, node_num, slot_num); 1356 skip_recovery: 1357 if (!status) { 1358 ocfs2_recovery_map_clear(osb, node_num); 1359 } else { 1360 mlog(ML_ERROR, 1361 "Error %d recovering node %d on device (%u,%u)!\n", 1362 status, node_num, 1363 MAJOR(osb->sb->s_dev), MINOR(osb->sb->s_dev)); 1364 mlog(ML_ERROR, "Volume requires unmount.\n"); 1365 } 1366 1367 spin_lock(&osb->osb_lock); 1368 } 1369 spin_unlock(&osb->osb_lock); 1370 trace_ocfs2_recovery_thread_end(status); 1371 1372 /* Refresh all journal recovery generations from disk */ 1373 status = ocfs2_check_journals_nolocks(osb); 1374 status = (status == -EROFS) ? 0 : status; 1375 if (status < 0) 1376 mlog_errno(status); 1377 1378 /* Now it is right time to recover quotas... We have to do this under 1379 * superblock lock so that no one can start using the slot (and crash) 1380 * before we recover it */ 1381 for (i = 0; i < rm_quota_used; i++) { 1382 qrec = ocfs2_begin_quota_recovery(osb, rm_quota[i]); 1383 if (IS_ERR(qrec)) { 1384 status = PTR_ERR(qrec); 1385 mlog_errno(status); 1386 continue; 1387 } 1388 ocfs2_queue_recovery_completion(osb->journal, rm_quota[i], 1389 NULL, NULL, qrec); 1390 } 1391 1392 ocfs2_super_unlock(osb, 1); 1393 1394 /* queue recovery for offline slots */ 1395 ocfs2_queue_replay_slots(osb); 1396 1397 bail: 1398 mutex_lock(&osb->recovery_lock); 1399 if (!status && !ocfs2_recovery_completed(osb)) { 1400 mutex_unlock(&osb->recovery_lock); 1401 goto restart; 1402 } 1403 1404 ocfs2_free_replay_slots(osb); 1405 osb->recovery_thread_task = NULL; 1406 mb(); /* sync with ocfs2_recovery_thread_running */ 1407 wake_up(&osb->recovery_event); 1408 1409 mutex_unlock(&osb->recovery_lock); 1410 1411 if (rm_quota) 1412 kfree(rm_quota); 1413 1414 /* no one is callint kthread_stop() for us so the kthread() api 1415 * requires that we call do_exit(). And it isn't exported, but 1416 * complete_and_exit() seems to be a minimal wrapper around it. */ 1417 complete_and_exit(NULL, status); 1418 return status; 1419 } 1420 1421 void ocfs2_recovery_thread(struct ocfs2_super *osb, int node_num) 1422 { 1423 mutex_lock(&osb->recovery_lock); 1424 1425 trace_ocfs2_recovery_thread(node_num, osb->node_num, 1426 osb->disable_recovery, osb->recovery_thread_task, 1427 osb->disable_recovery ? 1428 -1 : ocfs2_recovery_map_set(osb, node_num)); 1429 1430 if (osb->disable_recovery) 1431 goto out; 1432 1433 if (osb->recovery_thread_task) 1434 goto out; 1435 1436 osb->recovery_thread_task = kthread_run(__ocfs2_recovery_thread, osb, 1437 "ocfs2rec"); 1438 if (IS_ERR(osb->recovery_thread_task)) { 1439 mlog_errno((int)PTR_ERR(osb->recovery_thread_task)); 1440 osb->recovery_thread_task = NULL; 1441 } 1442 1443 out: 1444 mutex_unlock(&osb->recovery_lock); 1445 wake_up(&osb->recovery_event); 1446 } 1447 1448 static int ocfs2_read_journal_inode(struct ocfs2_super *osb, 1449 int slot_num, 1450 struct buffer_head **bh, 1451 struct inode **ret_inode) 1452 { 1453 int status = -EACCES; 1454 struct inode *inode = NULL; 1455 1456 BUG_ON(slot_num >= osb->max_slots); 1457 1458 inode = ocfs2_get_system_file_inode(osb, JOURNAL_SYSTEM_INODE, 1459 slot_num); 1460 if (!inode || is_bad_inode(inode)) { 1461 mlog_errno(status); 1462 goto bail; 1463 } 1464 SET_INODE_JOURNAL(inode); 1465 1466 status = ocfs2_read_inode_block_full(inode, bh, OCFS2_BH_IGNORE_CACHE); 1467 if (status < 0) { 1468 mlog_errno(status); 1469 goto bail; 1470 } 1471 1472 status = 0; 1473 1474 bail: 1475 if (inode) { 1476 if (status || !ret_inode) 1477 iput(inode); 1478 else 1479 *ret_inode = inode; 1480 } 1481 return status; 1482 } 1483 1484 /* Does the actual journal replay and marks the journal inode as 1485 * clean. Will only replay if the journal inode is marked dirty. */ 1486 static int ocfs2_replay_journal(struct ocfs2_super *osb, 1487 int node_num, 1488 int slot_num) 1489 { 1490 int status; 1491 int got_lock = 0; 1492 unsigned int flags; 1493 struct inode *inode = NULL; 1494 struct ocfs2_dinode *fe; 1495 journal_t *journal = NULL; 1496 struct buffer_head *bh = NULL; 1497 u32 slot_reco_gen; 1498 1499 status = ocfs2_read_journal_inode(osb, slot_num, &bh, &inode); 1500 if (status) { 1501 mlog_errno(status); 1502 goto done; 1503 } 1504 1505 fe = (struct ocfs2_dinode *)bh->b_data; 1506 slot_reco_gen = ocfs2_get_recovery_generation(fe); 1507 brelse(bh); 1508 bh = NULL; 1509 1510 /* 1511 * As the fs recovery is asynchronous, there is a small chance that 1512 * another node mounted (and recovered) the slot before the recovery 1513 * thread could get the lock. To handle that, we dirty read the journal 1514 * inode for that slot to get the recovery generation. If it is 1515 * different than what we expected, the slot has been recovered. 1516 * If not, it needs recovery. 1517 */ 1518 if (osb->slot_recovery_generations[slot_num] != slot_reco_gen) { 1519 trace_ocfs2_replay_journal_recovered(slot_num, 1520 osb->slot_recovery_generations[slot_num], slot_reco_gen); 1521 osb->slot_recovery_generations[slot_num] = slot_reco_gen; 1522 status = -EBUSY; 1523 goto done; 1524 } 1525 1526 /* Continue with recovery as the journal has not yet been recovered */ 1527 1528 status = ocfs2_inode_lock_full(inode, &bh, 1, OCFS2_META_LOCK_RECOVERY); 1529 if (status < 0) { 1530 trace_ocfs2_replay_journal_lock_err(status); 1531 if (status != -ERESTARTSYS) 1532 mlog(ML_ERROR, "Could not lock journal!\n"); 1533 goto done; 1534 } 1535 got_lock = 1; 1536 1537 fe = (struct ocfs2_dinode *) bh->b_data; 1538 1539 flags = le32_to_cpu(fe->id1.journal1.ij_flags); 1540 slot_reco_gen = ocfs2_get_recovery_generation(fe); 1541 1542 if (!(flags & OCFS2_JOURNAL_DIRTY_FL)) { 1543 trace_ocfs2_replay_journal_skip(node_num); 1544 /* Refresh recovery generation for the slot */ 1545 osb->slot_recovery_generations[slot_num] = slot_reco_gen; 1546 goto done; 1547 } 1548 1549 /* we need to run complete recovery for offline orphan slots */ 1550 ocfs2_replay_map_set_state(osb, REPLAY_NEEDED); 1551 1552 printk(KERN_NOTICE "ocfs2: Begin replay journal (node %d, slot %d) on "\ 1553 "device (%u,%u)\n", node_num, slot_num, MAJOR(osb->sb->s_dev), 1554 MINOR(osb->sb->s_dev)); 1555 1556 OCFS2_I(inode)->ip_clusters = le32_to_cpu(fe->i_clusters); 1557 1558 status = ocfs2_force_read_journal(inode); 1559 if (status < 0) { 1560 mlog_errno(status); 1561 goto done; 1562 } 1563 1564 journal = jbd2_journal_init_inode(inode); 1565 if (journal == NULL) { 1566 mlog(ML_ERROR, "Linux journal layer error\n"); 1567 status = -EIO; 1568 goto done; 1569 } 1570 1571 status = jbd2_journal_load(journal); 1572 if (status < 0) { 1573 mlog_errno(status); 1574 if (!igrab(inode)) 1575 BUG(); 1576 jbd2_journal_destroy(journal); 1577 goto done; 1578 } 1579 1580 ocfs2_clear_journal_error(osb->sb, journal, slot_num); 1581 1582 /* wipe the journal */ 1583 jbd2_journal_lock_updates(journal); 1584 status = jbd2_journal_flush(journal); 1585 jbd2_journal_unlock_updates(journal); 1586 if (status < 0) 1587 mlog_errno(status); 1588 1589 /* This will mark the node clean */ 1590 flags = le32_to_cpu(fe->id1.journal1.ij_flags); 1591 flags &= ~OCFS2_JOURNAL_DIRTY_FL; 1592 fe->id1.journal1.ij_flags = cpu_to_le32(flags); 1593 1594 /* Increment recovery generation to indicate successful recovery */ 1595 ocfs2_bump_recovery_generation(fe); 1596 osb->slot_recovery_generations[slot_num] = 1597 ocfs2_get_recovery_generation(fe); 1598 1599 ocfs2_compute_meta_ecc(osb->sb, bh->b_data, &fe->i_check); 1600 status = ocfs2_write_block(osb, bh, INODE_CACHE(inode)); 1601 if (status < 0) 1602 mlog_errno(status); 1603 1604 if (!igrab(inode)) 1605 BUG(); 1606 1607 jbd2_journal_destroy(journal); 1608 1609 printk(KERN_NOTICE "ocfs2: End replay journal (node %d, slot %d) on "\ 1610 "device (%u,%u)\n", node_num, slot_num, MAJOR(osb->sb->s_dev), 1611 MINOR(osb->sb->s_dev)); 1612 done: 1613 /* drop the lock on this nodes journal */ 1614 if (got_lock) 1615 ocfs2_inode_unlock(inode, 1); 1616 1617 if (inode) 1618 iput(inode); 1619 1620 brelse(bh); 1621 1622 return status; 1623 } 1624 1625 /* 1626 * Do the most important parts of node recovery: 1627 * - Replay it's journal 1628 * - Stamp a clean local allocator file 1629 * - Stamp a clean truncate log 1630 * - Mark the node clean 1631 * 1632 * If this function completes without error, a node in OCFS2 can be 1633 * said to have been safely recovered. As a result, failure during the 1634 * second part of a nodes recovery process (local alloc recovery) is 1635 * far less concerning. 1636 */ 1637 static int ocfs2_recover_node(struct ocfs2_super *osb, 1638 int node_num, int slot_num) 1639 { 1640 int status = 0; 1641 struct ocfs2_dinode *la_copy = NULL; 1642 struct ocfs2_dinode *tl_copy = NULL; 1643 1644 trace_ocfs2_recover_node(node_num, slot_num, osb->node_num); 1645 1646 /* Should not ever be called to recover ourselves -- in that 1647 * case we should've called ocfs2_journal_load instead. */ 1648 BUG_ON(osb->node_num == node_num); 1649 1650 status = ocfs2_replay_journal(osb, node_num, slot_num); 1651 if (status < 0) { 1652 if (status == -EBUSY) { 1653 trace_ocfs2_recover_node_skip(slot_num, node_num); 1654 status = 0; 1655 goto done; 1656 } 1657 mlog_errno(status); 1658 goto done; 1659 } 1660 1661 /* Stamp a clean local alloc file AFTER recovering the journal... */ 1662 status = ocfs2_begin_local_alloc_recovery(osb, slot_num, &la_copy); 1663 if (status < 0) { 1664 mlog_errno(status); 1665 goto done; 1666 } 1667 1668 /* An error from begin_truncate_log_recovery is not 1669 * serious enough to warrant halting the rest of 1670 * recovery. */ 1671 status = ocfs2_begin_truncate_log_recovery(osb, slot_num, &tl_copy); 1672 if (status < 0) 1673 mlog_errno(status); 1674 1675 /* Likewise, this would be a strange but ultimately not so 1676 * harmful place to get an error... */ 1677 status = ocfs2_clear_slot(osb, slot_num); 1678 if (status < 0) 1679 mlog_errno(status); 1680 1681 /* This will kfree the memory pointed to by la_copy and tl_copy */ 1682 ocfs2_queue_recovery_completion(osb->journal, slot_num, la_copy, 1683 tl_copy, NULL); 1684 1685 status = 0; 1686 done: 1687 1688 return status; 1689 } 1690 1691 /* Test node liveness by trylocking his journal. If we get the lock, 1692 * we drop it here. Return 0 if we got the lock, -EAGAIN if node is 1693 * still alive (we couldn't get the lock) and < 0 on error. */ 1694 static int ocfs2_trylock_journal(struct ocfs2_super *osb, 1695 int slot_num) 1696 { 1697 int status, flags; 1698 struct inode *inode = NULL; 1699 1700 inode = ocfs2_get_system_file_inode(osb, JOURNAL_SYSTEM_INODE, 1701 slot_num); 1702 if (inode == NULL) { 1703 mlog(ML_ERROR, "access error\n"); 1704 status = -EACCES; 1705 goto bail; 1706 } 1707 if (is_bad_inode(inode)) { 1708 mlog(ML_ERROR, "access error (bad inode)\n"); 1709 iput(inode); 1710 inode = NULL; 1711 status = -EACCES; 1712 goto bail; 1713 } 1714 SET_INODE_JOURNAL(inode); 1715 1716 flags = OCFS2_META_LOCK_RECOVERY | OCFS2_META_LOCK_NOQUEUE; 1717 status = ocfs2_inode_lock_full(inode, NULL, 1, flags); 1718 if (status < 0) { 1719 if (status != -EAGAIN) 1720 mlog_errno(status); 1721 goto bail; 1722 } 1723 1724 ocfs2_inode_unlock(inode, 1); 1725 bail: 1726 if (inode) 1727 iput(inode); 1728 1729 return status; 1730 } 1731 1732 /* Call this underneath ocfs2_super_lock. It also assumes that the 1733 * slot info struct has been updated from disk. */ 1734 int ocfs2_mark_dead_nodes(struct ocfs2_super *osb) 1735 { 1736 unsigned int node_num; 1737 int status, i; 1738 u32 gen; 1739 struct buffer_head *bh = NULL; 1740 struct ocfs2_dinode *di; 1741 1742 /* This is called with the super block cluster lock, so we 1743 * know that the slot map can't change underneath us. */ 1744 1745 for (i = 0; i < osb->max_slots; i++) { 1746 /* Read journal inode to get the recovery generation */ 1747 status = ocfs2_read_journal_inode(osb, i, &bh, NULL); 1748 if (status) { 1749 mlog_errno(status); 1750 goto bail; 1751 } 1752 di = (struct ocfs2_dinode *)bh->b_data; 1753 gen = ocfs2_get_recovery_generation(di); 1754 brelse(bh); 1755 bh = NULL; 1756 1757 spin_lock(&osb->osb_lock); 1758 osb->slot_recovery_generations[i] = gen; 1759 1760 trace_ocfs2_mark_dead_nodes(i, 1761 osb->slot_recovery_generations[i]); 1762 1763 if (i == osb->slot_num) { 1764 spin_unlock(&osb->osb_lock); 1765 continue; 1766 } 1767 1768 status = ocfs2_slot_to_node_num_locked(osb, i, &node_num); 1769 if (status == -ENOENT) { 1770 spin_unlock(&osb->osb_lock); 1771 continue; 1772 } 1773 1774 if (__ocfs2_recovery_map_test(osb, node_num)) { 1775 spin_unlock(&osb->osb_lock); 1776 continue; 1777 } 1778 spin_unlock(&osb->osb_lock); 1779 1780 /* Ok, we have a slot occupied by another node which 1781 * is not in the recovery map. We trylock his journal 1782 * file here to test if he's alive. */ 1783 status = ocfs2_trylock_journal(osb, i); 1784 if (!status) { 1785 /* Since we're called from mount, we know that 1786 * the recovery thread can't race us on 1787 * setting / checking the recovery bits. */ 1788 ocfs2_recovery_thread(osb, node_num); 1789 } else if ((status < 0) && (status != -EAGAIN)) { 1790 mlog_errno(status); 1791 goto bail; 1792 } 1793 } 1794 1795 status = 0; 1796 bail: 1797 return status; 1798 } 1799 1800 /* 1801 * Scan timer should get fired every ORPHAN_SCAN_SCHEDULE_TIMEOUT. Add some 1802 * randomness to the timeout to minimize multple nodes firing the timer at the 1803 * same time. 1804 */ 1805 static inline unsigned long ocfs2_orphan_scan_timeout(void) 1806 { 1807 unsigned long time; 1808 1809 get_random_bytes(&time, sizeof(time)); 1810 time = ORPHAN_SCAN_SCHEDULE_TIMEOUT + (time % 5000); 1811 return msecs_to_jiffies(time); 1812 } 1813 1814 /* 1815 * ocfs2_queue_orphan_scan calls ocfs2_queue_recovery_completion for 1816 * every slot, queuing a recovery of the slot on the ocfs2_wq thread. This 1817 * is done to catch any orphans that are left over in orphan directories. 1818 * 1819 * It scans all slots, even ones that are in use. It does so to handle the 1820 * case described below: 1821 * 1822 * Node 1 has an inode it was using. The dentry went away due to memory 1823 * pressure. Node 1 closes the inode, but it's on the free list. The node 1824 * has the open lock. 1825 * Node 2 unlinks the inode. It grabs the dentry lock to notify others, 1826 * but node 1 has no dentry and doesn't get the message. It trylocks the 1827 * open lock, sees that another node has a PR, and does nothing. 1828 * Later node 2 runs its orphan dir. It igets the inode, trylocks the 1829 * open lock, sees the PR still, and does nothing. 1830 * Basically, we have to trigger an orphan iput on node 1. The only way 1831 * for this to happen is if node 1 runs node 2's orphan dir. 1832 * 1833 * ocfs2_queue_orphan_scan gets called every ORPHAN_SCAN_SCHEDULE_TIMEOUT 1834 * seconds. It gets an EX lock on os_lockres and checks sequence number 1835 * stored in LVB. If the sequence number has changed, it means some other 1836 * node has done the scan. This node skips the scan and tracks the 1837 * sequence number. If the sequence number didn't change, it means a scan 1838 * hasn't happened. The node queues a scan and increments the 1839 * sequence number in the LVB. 1840 */ 1841 void ocfs2_queue_orphan_scan(struct ocfs2_super *osb) 1842 { 1843 struct ocfs2_orphan_scan *os; 1844 int status, i; 1845 u32 seqno = 0; 1846 1847 os = &osb->osb_orphan_scan; 1848 1849 if (atomic_read(&os->os_state) == ORPHAN_SCAN_INACTIVE) 1850 goto out; 1851 1852 trace_ocfs2_queue_orphan_scan_begin(os->os_count, os->os_seqno, 1853 atomic_read(&os->os_state)); 1854 1855 status = ocfs2_orphan_scan_lock(osb, &seqno); 1856 if (status < 0) { 1857 if (status != -EAGAIN) 1858 mlog_errno(status); 1859 goto out; 1860 } 1861 1862 /* Do no queue the tasks if the volume is being umounted */ 1863 if (atomic_read(&os->os_state) == ORPHAN_SCAN_INACTIVE) 1864 goto unlock; 1865 1866 if (os->os_seqno != seqno) { 1867 os->os_seqno = seqno; 1868 goto unlock; 1869 } 1870 1871 for (i = 0; i < osb->max_slots; i++) 1872 ocfs2_queue_recovery_completion(osb->journal, i, NULL, NULL, 1873 NULL); 1874 /* 1875 * We queued a recovery on orphan slots, increment the sequence 1876 * number and update LVB so other node will skip the scan for a while 1877 */ 1878 seqno++; 1879 os->os_count++; 1880 os->os_scantime = CURRENT_TIME; 1881 unlock: 1882 ocfs2_orphan_scan_unlock(osb, seqno); 1883 out: 1884 trace_ocfs2_queue_orphan_scan_end(os->os_count, os->os_seqno, 1885 atomic_read(&os->os_state)); 1886 return; 1887 } 1888 1889 /* Worker task that gets fired every ORPHAN_SCAN_SCHEDULE_TIMEOUT millsec */ 1890 void ocfs2_orphan_scan_work(struct work_struct *work) 1891 { 1892 struct ocfs2_orphan_scan *os; 1893 struct ocfs2_super *osb; 1894 1895 os = container_of(work, struct ocfs2_orphan_scan, 1896 os_orphan_scan_work.work); 1897 osb = os->os_osb; 1898 1899 mutex_lock(&os->os_lock); 1900 ocfs2_queue_orphan_scan(osb); 1901 if (atomic_read(&os->os_state) == ORPHAN_SCAN_ACTIVE) 1902 queue_delayed_work(ocfs2_wq, &os->os_orphan_scan_work, 1903 ocfs2_orphan_scan_timeout()); 1904 mutex_unlock(&os->os_lock); 1905 } 1906 1907 void ocfs2_orphan_scan_stop(struct ocfs2_super *osb) 1908 { 1909 struct ocfs2_orphan_scan *os; 1910 1911 os = &osb->osb_orphan_scan; 1912 if (atomic_read(&os->os_state) == ORPHAN_SCAN_ACTIVE) { 1913 atomic_set(&os->os_state, ORPHAN_SCAN_INACTIVE); 1914 mutex_lock(&os->os_lock); 1915 cancel_delayed_work(&os->os_orphan_scan_work); 1916 mutex_unlock(&os->os_lock); 1917 } 1918 } 1919 1920 void ocfs2_orphan_scan_init(struct ocfs2_super *osb) 1921 { 1922 struct ocfs2_orphan_scan *os; 1923 1924 os = &osb->osb_orphan_scan; 1925 os->os_osb = osb; 1926 os->os_count = 0; 1927 os->os_seqno = 0; 1928 mutex_init(&os->os_lock); 1929 INIT_DELAYED_WORK(&os->os_orphan_scan_work, ocfs2_orphan_scan_work); 1930 } 1931 1932 void ocfs2_orphan_scan_start(struct ocfs2_super *osb) 1933 { 1934 struct ocfs2_orphan_scan *os; 1935 1936 os = &osb->osb_orphan_scan; 1937 os->os_scantime = CURRENT_TIME; 1938 if (ocfs2_is_hard_readonly(osb) || ocfs2_mount_local(osb)) 1939 atomic_set(&os->os_state, ORPHAN_SCAN_INACTIVE); 1940 else { 1941 atomic_set(&os->os_state, ORPHAN_SCAN_ACTIVE); 1942 queue_delayed_work(ocfs2_wq, &os->os_orphan_scan_work, 1943 ocfs2_orphan_scan_timeout()); 1944 } 1945 } 1946 1947 struct ocfs2_orphan_filldir_priv { 1948 struct inode *head; 1949 struct ocfs2_super *osb; 1950 }; 1951 1952 static int ocfs2_orphan_filldir(void *priv, const char *name, int name_len, 1953 loff_t pos, u64 ino, unsigned type) 1954 { 1955 struct ocfs2_orphan_filldir_priv *p = priv; 1956 struct inode *iter; 1957 1958 if (name_len == 1 && !strncmp(".", name, 1)) 1959 return 0; 1960 if (name_len == 2 && !strncmp("..", name, 2)) 1961 return 0; 1962 1963 /* Skip bad inodes so that recovery can continue */ 1964 iter = ocfs2_iget(p->osb, ino, 1965 OCFS2_FI_FLAG_ORPHAN_RECOVERY, 0); 1966 if (IS_ERR(iter)) 1967 return 0; 1968 1969 trace_ocfs2_orphan_filldir((unsigned long long)OCFS2_I(iter)->ip_blkno); 1970 /* No locking is required for the next_orphan queue as there 1971 * is only ever a single process doing orphan recovery. */ 1972 OCFS2_I(iter)->ip_next_orphan = p->head; 1973 p->head = iter; 1974 1975 return 0; 1976 } 1977 1978 static int ocfs2_queue_orphans(struct ocfs2_super *osb, 1979 int slot, 1980 struct inode **head) 1981 { 1982 int status; 1983 struct inode *orphan_dir_inode = NULL; 1984 struct ocfs2_orphan_filldir_priv priv; 1985 loff_t pos = 0; 1986 1987 priv.osb = osb; 1988 priv.head = *head; 1989 1990 orphan_dir_inode = ocfs2_get_system_file_inode(osb, 1991 ORPHAN_DIR_SYSTEM_INODE, 1992 slot); 1993 if (!orphan_dir_inode) { 1994 status = -ENOENT; 1995 mlog_errno(status); 1996 return status; 1997 } 1998 1999 mutex_lock(&orphan_dir_inode->i_mutex); 2000 status = ocfs2_inode_lock(orphan_dir_inode, NULL, 0); 2001 if (status < 0) { 2002 mlog_errno(status); 2003 goto out; 2004 } 2005 2006 status = ocfs2_dir_foreach(orphan_dir_inode, &pos, &priv, 2007 ocfs2_orphan_filldir); 2008 if (status) { 2009 mlog_errno(status); 2010 goto out_cluster; 2011 } 2012 2013 *head = priv.head; 2014 2015 out_cluster: 2016 ocfs2_inode_unlock(orphan_dir_inode, 0); 2017 out: 2018 mutex_unlock(&orphan_dir_inode->i_mutex); 2019 iput(orphan_dir_inode); 2020 return status; 2021 } 2022 2023 static int ocfs2_orphan_recovery_can_continue(struct ocfs2_super *osb, 2024 int slot) 2025 { 2026 int ret; 2027 2028 spin_lock(&osb->osb_lock); 2029 ret = !osb->osb_orphan_wipes[slot]; 2030 spin_unlock(&osb->osb_lock); 2031 return ret; 2032 } 2033 2034 static void ocfs2_mark_recovering_orphan_dir(struct ocfs2_super *osb, 2035 int slot) 2036 { 2037 spin_lock(&osb->osb_lock); 2038 /* Mark ourselves such that new processes in delete_inode() 2039 * know to quit early. */ 2040 ocfs2_node_map_set_bit(osb, &osb->osb_recovering_orphan_dirs, slot); 2041 while (osb->osb_orphan_wipes[slot]) { 2042 /* If any processes are already in the middle of an 2043 * orphan wipe on this dir, then we need to wait for 2044 * them. */ 2045 spin_unlock(&osb->osb_lock); 2046 wait_event_interruptible(osb->osb_wipe_event, 2047 ocfs2_orphan_recovery_can_continue(osb, slot)); 2048 spin_lock(&osb->osb_lock); 2049 } 2050 spin_unlock(&osb->osb_lock); 2051 } 2052 2053 static void ocfs2_clear_recovering_orphan_dir(struct ocfs2_super *osb, 2054 int slot) 2055 { 2056 ocfs2_node_map_clear_bit(osb, &osb->osb_recovering_orphan_dirs, slot); 2057 } 2058 2059 /* 2060 * Orphan recovery. Each mounted node has it's own orphan dir which we 2061 * must run during recovery. Our strategy here is to build a list of 2062 * the inodes in the orphan dir and iget/iput them. The VFS does 2063 * (most) of the rest of the work. 2064 * 2065 * Orphan recovery can happen at any time, not just mount so we have a 2066 * couple of extra considerations. 2067 * 2068 * - We grab as many inodes as we can under the orphan dir lock - 2069 * doing iget() outside the orphan dir risks getting a reference on 2070 * an invalid inode. 2071 * - We must be sure not to deadlock with other processes on the 2072 * system wanting to run delete_inode(). This can happen when they go 2073 * to lock the orphan dir and the orphan recovery process attempts to 2074 * iget() inside the orphan dir lock. This can be avoided by 2075 * advertising our state to ocfs2_delete_inode(). 2076 */ 2077 static int ocfs2_recover_orphans(struct ocfs2_super *osb, 2078 int slot) 2079 { 2080 int ret = 0; 2081 struct inode *inode = NULL; 2082 struct inode *iter; 2083 struct ocfs2_inode_info *oi; 2084 2085 trace_ocfs2_recover_orphans(slot); 2086 2087 ocfs2_mark_recovering_orphan_dir(osb, slot); 2088 ret = ocfs2_queue_orphans(osb, slot, &inode); 2089 ocfs2_clear_recovering_orphan_dir(osb, slot); 2090 2091 /* Error here should be noted, but we want to continue with as 2092 * many queued inodes as we've got. */ 2093 if (ret) 2094 mlog_errno(ret); 2095 2096 while (inode) { 2097 oi = OCFS2_I(inode); 2098 trace_ocfs2_recover_orphans_iput( 2099 (unsigned long long)oi->ip_blkno); 2100 2101 iter = oi->ip_next_orphan; 2102 2103 spin_lock(&oi->ip_lock); 2104 /* The remote delete code may have set these on the 2105 * assumption that the other node would wipe them 2106 * successfully. If they are still in the node's 2107 * orphan dir, we need to reset that state. */ 2108 oi->ip_flags &= ~(OCFS2_INODE_DELETED|OCFS2_INODE_SKIP_DELETE); 2109 2110 /* Set the proper information to get us going into 2111 * ocfs2_delete_inode. */ 2112 oi->ip_flags |= OCFS2_INODE_MAYBE_ORPHANED; 2113 spin_unlock(&oi->ip_lock); 2114 2115 iput(inode); 2116 2117 inode = iter; 2118 } 2119 2120 return ret; 2121 } 2122 2123 static int __ocfs2_wait_on_mount(struct ocfs2_super *osb, int quota) 2124 { 2125 /* This check is good because ocfs2 will wait on our recovery 2126 * thread before changing it to something other than MOUNTED 2127 * or DISABLED. */ 2128 wait_event(osb->osb_mount_event, 2129 (!quota && atomic_read(&osb->vol_state) == VOLUME_MOUNTED) || 2130 atomic_read(&osb->vol_state) == VOLUME_MOUNTED_QUOTAS || 2131 atomic_read(&osb->vol_state) == VOLUME_DISABLED); 2132 2133 /* If there's an error on mount, then we may never get to the 2134 * MOUNTED flag, but this is set right before 2135 * dismount_volume() so we can trust it. */ 2136 if (atomic_read(&osb->vol_state) == VOLUME_DISABLED) { 2137 trace_ocfs2_wait_on_mount(VOLUME_DISABLED); 2138 mlog(0, "mount error, exiting!\n"); 2139 return -EBUSY; 2140 } 2141 2142 return 0; 2143 } 2144 2145 static int ocfs2_commit_thread(void *arg) 2146 { 2147 int status; 2148 struct ocfs2_super *osb = arg; 2149 struct ocfs2_journal *journal = osb->journal; 2150 2151 /* we can trust j_num_trans here because _should_stop() is only set in 2152 * shutdown and nobody other than ourselves should be able to start 2153 * transactions. committing on shutdown might take a few iterations 2154 * as final transactions put deleted inodes on the list */ 2155 while (!(kthread_should_stop() && 2156 atomic_read(&journal->j_num_trans) == 0)) { 2157 2158 wait_event_interruptible(osb->checkpoint_event, 2159 atomic_read(&journal->j_num_trans) 2160 || kthread_should_stop()); 2161 2162 status = ocfs2_commit_cache(osb); 2163 if (status < 0) 2164 mlog_errno(status); 2165 2166 if (kthread_should_stop() && atomic_read(&journal->j_num_trans)){ 2167 mlog(ML_KTHREAD, 2168 "commit_thread: %u transactions pending on " 2169 "shutdown\n", 2170 atomic_read(&journal->j_num_trans)); 2171 } 2172 } 2173 2174 return 0; 2175 } 2176 2177 /* Reads all the journal inodes without taking any cluster locks. Used 2178 * for hard readonly access to determine whether any journal requires 2179 * recovery. Also used to refresh the recovery generation numbers after 2180 * a journal has been recovered by another node. 2181 */ 2182 int ocfs2_check_journals_nolocks(struct ocfs2_super *osb) 2183 { 2184 int ret = 0; 2185 unsigned int slot; 2186 struct buffer_head *di_bh = NULL; 2187 struct ocfs2_dinode *di; 2188 int journal_dirty = 0; 2189 2190 for(slot = 0; slot < osb->max_slots; slot++) { 2191 ret = ocfs2_read_journal_inode(osb, slot, &di_bh, NULL); 2192 if (ret) { 2193 mlog_errno(ret); 2194 goto out; 2195 } 2196 2197 di = (struct ocfs2_dinode *) di_bh->b_data; 2198 2199 osb->slot_recovery_generations[slot] = 2200 ocfs2_get_recovery_generation(di); 2201 2202 if (le32_to_cpu(di->id1.journal1.ij_flags) & 2203 OCFS2_JOURNAL_DIRTY_FL) 2204 journal_dirty = 1; 2205 2206 brelse(di_bh); 2207 di_bh = NULL; 2208 } 2209 2210 out: 2211 if (journal_dirty) 2212 ret = -EROFS; 2213 return ret; 2214 } 2215