| 1 | /* |
| 2 | * Copyright (c) 2008 The DragonFly Project. All rights reserved. |
| 3 | * |
| 4 | * This code is derived from software contributed to The DragonFly Project |
| 5 | * by Matthew Dillon <dillon@backplane.com> |
| 6 | * |
| 7 | * Redistribution and use in source and binary forms, with or without |
| 8 | * modification, are permitted provided that the following conditions |
| 9 | * are met: |
| 10 | * |
| 11 | * 1. Redistributions of source code must retain the above copyright |
| 12 | * notice, this list of conditions and the following disclaimer. |
| 13 | * 2. Redistributions in binary form must reproduce the above copyright |
| 14 | * notice, this list of conditions and the following disclaimer in |
| 15 | * the documentation and/or other materials provided with the |
| 16 | * distribution. |
| 17 | * 3. Neither the name of The DragonFly Project nor the names of its |
| 18 | * contributors may be used to endorse or promote products derived |
| 19 | * from this software without specific, prior written permission. |
| 20 | * |
| 21 | * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS |
| 22 | * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT |
| 23 | * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS |
| 24 | * FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE |
| 25 | * COPYRIGHT HOLDERS OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, |
| 26 | * INCIDENTAL, SPECIAL, EXEMPLARY OR CONSEQUENTIAL DAMAGES (INCLUDING, |
| 27 | * BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; |
| 28 | * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED |
| 29 | * AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, |
| 30 | * OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT |
| 31 | * OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF |
| 32 | * SUCH DAMAGE. |
| 33 | */ |
| 34 | |
| 35 | /* |
| 36 | * UNDO ALGORITHM: |
| 37 | * |
| 38 | * The UNDO algorithm is trivial. The nominal UNDO range in the |
| 39 | * FIFO is determined by taking the first/next offset stored in |
| 40 | * the volume header. The next offset may not be correct since |
| 41 | * UNDO flushes are not required to flush the volume header, so |
| 42 | * the code also scans forward until it finds a discontinuous |
| 43 | * sequence number. |
| 44 | * |
| 45 | * The UNDOs are then scanned and executed in reverse order. These |
| 46 | * UNDOs are effectively just data restorations based on HAMMER offsets. |
| 47 | * |
| 48 | * REDO ALGORITHM: |
| 49 | * |
| 50 | * REDO records are laid down in the UNDO/REDO FIFO for nominal |
| 51 | * writes, truncations, and file extension ops. On a per-inode |
| 52 | * basis two types of REDO records are generated, REDO_WRITE |
| 53 | * and REDO_TRUNC. |
| 54 | * |
| 55 | * Essentially the recovery block will contain UNDO records backing |
| 56 | * out partial operations and REDO records to regenerate those partial |
| 57 | * operations guaranteed by the filesystem during recovery. |
| 58 | * |
| 59 | * REDO generation is optional, and can also be started and then |
| 60 | * later stopped due to excessive write()s inbetween fsyncs, or not |
| 61 | * started at all. Because of this the recovery code must determine |
| 62 | * when REDOs are valid and when they are not. Additional records are |
| 63 | * generated to help figure it out. |
| 64 | * |
| 65 | * The REDO_TERM_WRITE and REDO_TERM_TRUNC records are generated |
| 66 | * during a flush cycle indicating which records the flush cycle |
| 67 | * has synched meta-data for, and HAMMER_REDO_SYNC is generated in |
| 68 | * each flush cycle to indicate how far back in the UNDO/REDO FIFO |
| 69 | * the recovery code must go to find the earliest applicable REDO |
| 70 | * record. Applicable REDO records can be far outside the nominal |
| 71 | * UNDO recovery range, for example if a write() lays down a REDO but |
| 72 | * the related file is not flushed for several cycles. |
| 73 | * |
| 74 | * The SYNC reference is to a point prior to the nominal UNDO FIFO |
| 75 | * range, creating an extended REDO range which must be scanned. |
| 76 | * |
| 77 | * Any REDO_WRITE/REDO_TRUNC encountered within the extended range |
| 78 | * which have no matching REDO_TERM_WRITE/REDO_TERM_TRUNC records |
| 79 | * prior to the start of the nominal UNDO range are applicable. |
| 80 | * That is, any REDO_TERM_* records in the extended range but not in |
| 81 | * the nominal undo range will mask any redo operations for prior REDO |
| 82 | * records. This is necessary because once the TERM is laid down |
| 83 | * followup operations may make additional changes to the related |
| 84 | * records but not necessarily record them as REDOs (because REDOs are |
| 85 | * optional). |
| 86 | * |
| 87 | * REDO_TERM_WRITE/REDO_TERM_TRUNC records in the nominal UNDO range |
| 88 | * must be ignored since they represent meta-data flushes which are |
| 89 | * undone by the UNDOs in that nominal UNDO range by the recovery |
| 90 | * code. Only REDO_TERM_* records in the extended range but not |
| 91 | * in the nominal undo range are applicable. |
| 92 | * |
| 93 | * The REDO_SYNC record itself always exists in the nominal UNDO range |
| 94 | * (this is how the extended range is determined). For recovery |
| 95 | * purposes the most recent REDO_SYNC record is always used if several |
| 96 | * are found. |
| 97 | * |
| 98 | * CRASHES DURING UNDO/REDO |
| 99 | * |
| 100 | * A crash during the UNDO phase requires no additional effort. The |
| 101 | * UNDOs will simply be re-run again. The state of the UNDO/REDO fifo |
| 102 | * remains unchanged and has no re-crash issues. |
| 103 | * |
| 104 | * A crash during the REDO phase is more complex because the REDOs |
| 105 | * run normal filesystem ops and generate additional UNDO/REDO records. |
| 106 | * REDO is disabled during REDO recovery and any SYNC records generated |
| 107 | * by flushes during REDO recovery must continue to reference the |
| 108 | * original extended range. |
| 109 | * |
| 110 | * If multiple crashes occur and the UNDO/REDO FIFO wraps, REDO recovery |
| 111 | * may become impossible. This is detected when the start of the |
| 112 | * extended range fails to have monotonically increasing sequence |
| 113 | * numbers leading into the nominal undo range. |
| 114 | */ |
| 115 | |
| 116 | #include "hammer.h" |
| 117 | |
| 118 | /* |
| 119 | * Each rterm entry has a list of fifo offsets indicating termination |
| 120 | * points. These are stripped as the scan progresses. |
| 121 | */ |
| 122 | typedef struct hammer_rterm_entry { |
| 123 | struct hammer_rterm_entry *next; |
| 124 | hammer_off_t fifo_offset; |
| 125 | } *hammer_rterm_entry_t; |
| 126 | |
| 127 | /* |
| 128 | * rterm entries sorted in RB tree are indexed by objid, flags, and offset. |
| 129 | * TRUNC entries ignore the offset. |
| 130 | */ |
| 131 | typedef struct hammer_rterm { |
| 132 | RB_ENTRY(hammer_rterm) rb_node; |
| 133 | int64_t redo_objid; |
| 134 | u_int32_t redo_localization; |
| 135 | u_int32_t redo_flags; |
| 136 | hammer_off_t redo_offset; |
| 137 | hammer_rterm_entry_t term_list; |
| 138 | } *hammer_rterm_t; |
| 139 | |
| 140 | static int hammer_rterm_rb_cmp(hammer_rterm_t rt1, hammer_rterm_t rt2); |
| 141 | struct hammer_rterm_rb_tree; |
| 142 | RB_HEAD(hammer_rterm_rb_tree, hammer_rterm); |
| 143 | RB_PROTOTYPE(hammer_rterm_rb_tree, hammer_rterm, rb_node, hammer_rterm_rb_cmp); |
| 144 | |
| 145 | static int hammer_check_tail_signature(hammer_fifo_tail_t tail, |
| 146 | hammer_off_t end_off); |
| 147 | static int hammer_check_head_signature(hammer_fifo_head_t head, |
| 148 | hammer_off_t beg_off); |
| 149 | static void hammer_recover_copy_undo(hammer_off_t undo_offset, |
| 150 | char *src, char *dst, int bytes); |
| 151 | static hammer_fifo_any_t hammer_recover_scan_fwd(hammer_mount_t hmp, |
| 152 | hammer_volume_t root_volume, |
| 153 | hammer_off_t *scan_offsetp, |
| 154 | int *errorp, struct hammer_buffer **bufferp); |
| 155 | static hammer_fifo_any_t hammer_recover_scan_rev(hammer_mount_t hmp, |
| 156 | hammer_volume_t root_volume, |
| 157 | hammer_off_t *scan_offsetp, |
| 158 | int *errorp, struct hammer_buffer **bufferp); |
| 159 | #if 0 |
| 160 | static void hammer_recover_debug_dump(int w, char *buf, int bytes); |
| 161 | #endif |
| 162 | static int hammer_recover_undo(hammer_mount_t hmp, hammer_volume_t root_volume, |
| 163 | hammer_fifo_undo_t undo); |
| 164 | static int hammer_recover_redo_rec(hammer_mount_t hmp, |
| 165 | struct hammer_rterm_rb_tree *root, |
| 166 | hammer_off_t redo_fifo_offset, hammer_fifo_redo_t redo); |
| 167 | static int hammer_recover_redo_run(hammer_mount_t hmp, |
| 168 | struct hammer_rterm_rb_tree *root, |
| 169 | hammer_off_t redo_fifo_offset, hammer_fifo_redo_t redo); |
| 170 | static void hammer_recover_redo_exec(hammer_mount_t hmp, |
| 171 | hammer_fifo_redo_t redo); |
| 172 | |
| 173 | RB_GENERATE(hammer_rterm_rb_tree, hammer_rterm, rb_node, hammer_rterm_rb_cmp); |
| 174 | |
| 175 | /* |
| 176 | * Recover filesystem meta-data on mount. This procedure figures out the |
| 177 | * UNDO FIFO range and runs the UNDOs backwards. The FIFO pointers are not |
| 178 | * resynchronized by this procedure. |
| 179 | * |
| 180 | * This procedure is run near the beginning of the mount sequence, before |
| 181 | * any B-Tree or high-level accesses are enabled, and is responsible for |
| 182 | * restoring the meta-data to a consistent state. High level HAMMER data |
| 183 | * structures (such as the B-Tree) cannot be accessed here. |
| 184 | * |
| 185 | * NOTE: No information from the root volume has been cached in the |
| 186 | * hammer_mount structure yet, so we need to access the root volume's |
| 187 | * buffer directly. |
| 188 | * |
| 189 | * NOTE: |
| 190 | */ |
| 191 | int |
| 192 | hammer_recover_stage1(hammer_mount_t hmp, hammer_volume_t root_volume) |
| 193 | { |
| 194 | hammer_blockmap_t rootmap; |
| 195 | hammer_buffer_t buffer; |
| 196 | hammer_off_t scan_offset; |
| 197 | hammer_off_t scan_offset_save; |
| 198 | hammer_off_t bytes; |
| 199 | hammer_fifo_any_t head; |
| 200 | hammer_off_t first_offset; |
| 201 | hammer_off_t last_offset; |
| 202 | u_int32_t seqno; |
| 203 | int error; |
| 204 | int degenerate_case = 0; |
| 205 | |
| 206 | /* |
| 207 | * Examine the UNDO FIFO indices in the volume header. |
| 208 | */ |
| 209 | rootmap = &root_volume->ondisk->vol0_blockmap[HAMMER_ZONE_UNDO_INDEX]; |
| 210 | first_offset = rootmap->first_offset; |
| 211 | last_offset = rootmap->next_offset; |
| 212 | buffer = NULL; |
| 213 | error = 0; |
| 214 | |
| 215 | hmp->recover_stage2_offset = 0; |
| 216 | |
| 217 | if (first_offset > rootmap->alloc_offset || |
| 218 | last_offset > rootmap->alloc_offset) { |
| 219 | kprintf("HAMMER(%s) Illegal UNDO FIFO index range " |
| 220 | "%016jx, %016jx limit %016jx\n", |
| 221 | root_volume->ondisk->vol_name, |
| 222 | (intmax_t)first_offset, |
| 223 | (intmax_t)last_offset, |
| 224 | (intmax_t)rootmap->alloc_offset); |
| 225 | error = EIO; |
| 226 | goto done; |
| 227 | } |
| 228 | |
| 229 | /* |
| 230 | * In HAMMER version 4+ filesystems the volume header does NOT |
| 231 | * contain definitive UNDO FIFO state. In particular, the |
| 232 | * rootmap->next_offset may not be indexed completely to the |
| 233 | * end of the active UNDO FIFO. |
| 234 | */ |
| 235 | if (hmp->version >= HAMMER_VOL_VERSION_FOUR) { |
| 236 | /* |
| 237 | * To find the definitive range we must first scan backwards |
| 238 | * from first_offset to locate the first real record and |
| 239 | * extract the sequence number from it. This record is not |
| 240 | * part of the active undo space. |
| 241 | */ |
| 242 | scan_offset = first_offset; |
| 243 | seqno = 0; |
| 244 | |
| 245 | for (;;) { |
| 246 | head = hammer_recover_scan_rev(hmp, root_volume, |
| 247 | &scan_offset, |
| 248 | &error, &buffer); |
| 249 | if (error) |
| 250 | break; |
| 251 | if (head->head.hdr_type != HAMMER_HEAD_TYPE_PAD) { |
| 252 | seqno = head->head.hdr_seq; |
| 253 | break; |
| 254 | } |
| 255 | } |
| 256 | if (error) { |
| 257 | kprintf("HAMMER(%s) recovery failure " |
| 258 | "during seqno backscan\n", |
| 259 | root_volume->ondisk->vol_name); |
| 260 | goto done; |
| 261 | } |
| 262 | |
| 263 | /* |
| 264 | * Scan forwards from first_offset and (seqno+1) looking |
| 265 | * for a sequence space discontinuity. This denotes the |
| 266 | * end of the active FIFO area. |
| 267 | * |
| 268 | * NOTE: For the case where the FIFO is empty the very first |
| 269 | * record we find will be discontinuous. |
| 270 | * |
| 271 | * NOTE: Do not include trailing PADs in the scan range, |
| 272 | * and remember the returned scan_offset after a |
| 273 | * fwd iteration points to the end of the returned |
| 274 | * record. |
| 275 | */ |
| 276 | kprintf("HAMMER(%s) recovery check seqno=%08x\n", |
| 277 | root_volume->ondisk->vol_name, |
| 278 | seqno); |
| 279 | |
| 280 | scan_offset = first_offset; |
| 281 | scan_offset_save = scan_offset; |
| 282 | ++seqno; |
| 283 | hmp->recover_stage2_seqno = seqno; |
| 284 | |
| 285 | for (;;) { |
| 286 | head = hammer_recover_scan_fwd(hmp, root_volume, |
| 287 | &scan_offset, |
| 288 | &error, &buffer); |
| 289 | if (error) |
| 290 | break; |
| 291 | if (head->head.hdr_type != HAMMER_HEAD_TYPE_PAD) { |
| 292 | if (seqno != head->head.hdr_seq) { |
| 293 | scan_offset = scan_offset_save; |
| 294 | break; |
| 295 | } |
| 296 | scan_offset_save = scan_offset; |
| 297 | ++seqno; |
| 298 | } |
| 299 | |
| 300 | #if 0 |
| 301 | /* |
| 302 | * If the forward scan is grossly ahead of last_offset |
| 303 | * then something is wrong. last_offset is supposed |
| 304 | * to be flushed out |
| 305 | */ |
| 306 | if (last_offset >= scan_offset) { |
| 307 | bytes = last_offset - scan_offset; |
| 308 | } else { |
| 309 | bytes = rootmap->alloc_offset - scan_offset + |
| 310 | (last_offset & HAMMER_OFF_LONG_MASK); |
| 311 | } |
| 312 | if (bytes > |
| 313 | (rootmap->alloc_offset & HAMMER_OFF_LONG_MASK) * |
| 314 | 4 / 5) { |
| 315 | kprintf("HAMMER(%s) recovery forward scan is " |
| 316 | "grossly beyond the last_offset in " |
| 317 | "the volume header, this can't be " |
| 318 | "right.\n", |
| 319 | root_volume->ondisk->vol_name); |
| 320 | error = EIO; |
| 321 | break; |
| 322 | } |
| 323 | #endif |
| 324 | } |
| 325 | |
| 326 | /* |
| 327 | * Store the seqno. This will be the next seqno we lay down |
| 328 | * when generating new UNDOs. |
| 329 | */ |
| 330 | hmp->undo_seqno = seqno; |
| 331 | if (error) { |
| 332 | kprintf("HAMMER(%s) recovery failure " |
| 333 | "during seqno fwdscan\n", |
| 334 | root_volume->ondisk->vol_name); |
| 335 | goto done; |
| 336 | } |
| 337 | last_offset = scan_offset; |
| 338 | kprintf("HAMMER(%s) recovery range %016jx-%016jx\n" |
| 339 | "HAMMER(%s) recovery nexto %016jx endseqno=%08x\n", |
| 340 | root_volume->ondisk->vol_name, |
| 341 | (intmax_t)first_offset, |
| 342 | (intmax_t)last_offset, |
| 343 | root_volume->ondisk->vol_name, |
| 344 | (intmax_t)rootmap->next_offset, |
| 345 | seqno); |
| 346 | } |
| 347 | |
| 348 | /* |
| 349 | * Calculate the size of the active portion of the FIFO. If the |
| 350 | * FIFO is empty the filesystem is clean and no further action is |
| 351 | * needed. |
| 352 | */ |
| 353 | if (last_offset >= first_offset) { |
| 354 | bytes = last_offset - first_offset; |
| 355 | } else { |
| 356 | bytes = rootmap->alloc_offset - first_offset + |
| 357 | (last_offset & HAMMER_OFF_LONG_MASK); |
| 358 | } |
| 359 | if (bytes == 0) { |
| 360 | degenerate_case = 1; |
| 361 | error = 0; |
| 362 | goto done; |
| 363 | } |
| 364 | |
| 365 | kprintf("HAMMER(%s) recovery undo %016jx-%016jx (%jd bytes)%s\n", |
| 366 | root_volume->ondisk->vol_name, |
| 367 | (intmax_t)first_offset, |
| 368 | (intmax_t)last_offset, |
| 369 | (intmax_t)bytes, |
| 370 | (hmp->ronly ? " (RO)" : "(RW)")); |
| 371 | if (bytes > (rootmap->alloc_offset & HAMMER_OFF_LONG_MASK)) { |
| 372 | kprintf("Undo size is absurd, unable to mount\n"); |
| 373 | error = EIO; |
| 374 | goto done; |
| 375 | } |
| 376 | |
| 377 | /* |
| 378 | * Scan the UNDOs backwards. |
| 379 | */ |
| 380 | scan_offset = last_offset; |
| 381 | |
| 382 | while ((int64_t)bytes > 0) { |
| 383 | KKASSERT(scan_offset != first_offset); |
| 384 | head = hammer_recover_scan_rev(hmp, root_volume, |
| 385 | &scan_offset, &error, &buffer); |
| 386 | if (error) |
| 387 | break; |
| 388 | |
| 389 | /* |
| 390 | * Normal UNDO |
| 391 | */ |
| 392 | error = hammer_recover_undo(hmp, root_volume, &head->undo); |
| 393 | if (error) { |
| 394 | kprintf("HAMMER(%s) UNDO record at %016jx failed\n", |
| 395 | root_volume->ondisk->vol_name, |
| 396 | (intmax_t)scan_offset - head->head.hdr_size); |
| 397 | break; |
| 398 | } |
| 399 | |
| 400 | /* |
| 401 | * The first REDO_SYNC record encountered (scanning backwards) |
| 402 | * enables REDO processing. |
| 403 | */ |
| 404 | if (head->head.hdr_type == HAMMER_HEAD_TYPE_REDO && |
| 405 | head->redo.redo_flags == HAMMER_REDO_SYNC) { |
| 406 | if (hmp->flags & HAMMER_MOUNT_REDO_RECOVERY_REQ) { |
| 407 | kprintf("HAMMER(%s) Ignoring extra REDO_SYNC " |
| 408 | "records in UNDO/REDO FIFO.\n", |
| 409 | root_volume->ondisk->vol_name |
| 410 | ); |
| 411 | } else { |
| 412 | hmp->flags |= HAMMER_MOUNT_REDO_RECOVERY_REQ; |
| 413 | hmp->recover_stage2_offset = |
| 414 | head->redo.redo_offset; |
| 415 | kprintf("HAMMER(%s) Found REDO_SYNC %016jx\n", |
| 416 | root_volume->ondisk->vol_name, |
| 417 | (intmax_t)head->redo.redo_offset); |
| 418 | } |
| 419 | } |
| 420 | |
| 421 | bytes -= head->head.hdr_size; |
| 422 | |
| 423 | /* |
| 424 | * If too many dirty buffers have built up we have to flush'm |
| 425 | * out. As long as we do not flush out the volume header |
| 426 | * a crash here should not cause any problems. |
| 427 | * |
| 428 | * buffer must be released so the flush can assert that |
| 429 | * all buffers are idle. |
| 430 | */ |
| 431 | if (hammer_flusher_meta_limit(hmp)) { |
| 432 | if (buffer) { |
| 433 | hammer_rel_buffer(buffer, 0); |
| 434 | buffer = NULL; |
| 435 | } |
| 436 | if (hmp->ronly == 0) { |
| 437 | hammer_recover_flush_buffers(hmp, root_volume, |
| 438 | 0); |
| 439 | kprintf("HAMMER(%s) Continuing recovery\n", |
| 440 | root_volume->ondisk->vol_name); |
| 441 | } else { |
| 442 | kprintf("HAMMER(%s) Recovery failure: Insufficient buffer cache to hold dirty buffers on read-only mount!\n", |
| 443 | root_volume->ondisk->vol_name); |
| 444 | error = EIO; |
| 445 | break; |
| 446 | } |
| 447 | } |
| 448 | } |
| 449 | KKASSERT(error || bytes == 0); |
| 450 | done: |
| 451 | if (buffer) { |
| 452 | hammer_rel_buffer(buffer, 0); |
| 453 | buffer = NULL; |
| 454 | } |
| 455 | |
| 456 | /* |
| 457 | * After completely flushing all the recovered buffers the volume |
| 458 | * header will also be flushed. |
| 459 | */ |
| 460 | if (root_volume->io.recovered == 0) { |
| 461 | hammer_ref_volume(root_volume); |
| 462 | root_volume->io.recovered = 1; |
| 463 | } |
| 464 | |
| 465 | /* |
| 466 | * Finish up flushing (or discarding) recovered buffers. FIFO |
| 467 | * indices in the volume header are updated to the actual undo |
| 468 | * range but will not be collapsed until stage 2. |
| 469 | */ |
| 470 | if (error == 0) { |
| 471 | hammer_modify_volume(NULL, root_volume, NULL, 0); |
| 472 | rootmap = &root_volume->ondisk->vol0_blockmap[HAMMER_ZONE_UNDO_INDEX]; |
| 473 | rootmap->first_offset = first_offset; |
| 474 | rootmap->next_offset = last_offset; |
| 475 | hammer_modify_volume_done(root_volume); |
| 476 | if (hmp->ronly == 0) |
| 477 | hammer_recover_flush_buffers(hmp, root_volume, 1); |
| 478 | } else { |
| 479 | hammer_recover_flush_buffers(hmp, root_volume, -1); |
| 480 | } |
| 481 | if (degenerate_case == 0) { |
| 482 | kprintf("HAMMER(%s) recovery complete\n", |
| 483 | root_volume->ondisk->vol_name); |
| 484 | } else { |
| 485 | kprintf("HAMMER(%s) mounted clean, no recovery needed\n", |
| 486 | root_volume->ondisk->vol_name); |
| 487 | } |
| 488 | return (error); |
| 489 | } |
| 490 | |
| 491 | /* |
| 492 | * Execute redo operations |
| 493 | * |
| 494 | * This procedure is run at the end of the mount sequence, after the hammer |
| 495 | * mount structure has been completely initialized but before the filesystem |
| 496 | * goes live. It can access standard cursors, the B-Tree, flush the |
| 497 | * filesystem, and so forth. |
| 498 | * |
| 499 | * This code may only be called for read-write mounts or when a mount |
| 500 | * switches from read-only to read-write. vnodes may or may not be present. |
| 501 | * |
| 502 | * The stage1 code will have already calculated the correct FIFO range |
| 503 | * for the nominal UNDO FIFO and stored it in the rootmap. The extended |
| 504 | * range for REDO is stored in hmp->recover_stage2_offset. |
| 505 | */ |
| 506 | int |
| 507 | hammer_recover_stage2(hammer_mount_t hmp, hammer_volume_t root_volume) |
| 508 | { |
| 509 | hammer_blockmap_t rootmap; |
| 510 | hammer_buffer_t buffer; |
| 511 | hammer_off_t scan_offset; |
| 512 | hammer_off_t oscan_offset; |
| 513 | hammer_off_t bytes; |
| 514 | hammer_off_t ext_bytes; |
| 515 | hammer_fifo_any_t head; |
| 516 | hammer_off_t first_offset; |
| 517 | hammer_off_t last_offset; |
| 518 | hammer_off_t ext_offset; |
| 519 | struct hammer_rterm_rb_tree rterm_root; |
| 520 | u_int32_t seqno; |
| 521 | int error; |
| 522 | int verbose = 0; |
| 523 | int dorscan; |
| 524 | |
| 525 | /* |
| 526 | * Stage 2 can only be run on a RW mount, or when the mount is |
| 527 | * switched from RO to RW. |
| 528 | */ |
| 529 | KKASSERT(hmp->ronly == 0); |
| 530 | RB_INIT(&rterm_root); |
| 531 | |
| 532 | /* |
| 533 | * Examine the UNDO FIFO. If it is empty the filesystem is clean |
| 534 | * and no action need be taken. |
| 535 | */ |
| 536 | rootmap = &root_volume->ondisk->vol0_blockmap[HAMMER_ZONE_UNDO_INDEX]; |
| 537 | first_offset = rootmap->first_offset; |
| 538 | last_offset = rootmap->next_offset; |
| 539 | if (first_offset == last_offset) { |
| 540 | KKASSERT((hmp->flags & HAMMER_MOUNT_REDO_RECOVERY_REQ) == 0); |
| 541 | return(0); |
| 542 | } |
| 543 | |
| 544 | /* |
| 545 | * Stage2 must only be run once, and will not be run at all |
| 546 | * if Stage1 did not find a REDO_SYNC record. |
| 547 | */ |
| 548 | error = 0; |
| 549 | buffer = NULL; |
| 550 | |
| 551 | if ((hmp->flags & HAMMER_MOUNT_REDO_RECOVERY_REQ) == 0) |
| 552 | goto done; |
| 553 | hmp->flags &= ~HAMMER_MOUNT_REDO_RECOVERY_REQ; |
| 554 | hmp->flags |= HAMMER_MOUNT_REDO_RECOVERY_RUN; |
| 555 | ext_offset = hmp->recover_stage2_offset; |
| 556 | if (ext_offset == 0) { |
| 557 | kprintf("HAMMER(%s) REDO stage specified but no REDO_SYNC " |
| 558 | "offset, ignoring\n", |
| 559 | root_volume->ondisk->vol_name); |
| 560 | goto done; |
| 561 | } |
| 562 | |
| 563 | /* |
| 564 | * Calculate nominal UNDO range (this is not yet the extended |
| 565 | * range). |
| 566 | */ |
| 567 | if (last_offset >= first_offset) { |
| 568 | bytes = last_offset - first_offset; |
| 569 | } else { |
| 570 | bytes = rootmap->alloc_offset - first_offset + |
| 571 | (last_offset & HAMMER_OFF_LONG_MASK); |
| 572 | } |
| 573 | kprintf("HAMMER(%s) recovery redo %016jx-%016jx (%jd bytes)%s\n", |
| 574 | root_volume->ondisk->vol_name, |
| 575 | (intmax_t)first_offset, |
| 576 | (intmax_t)last_offset, |
| 577 | (intmax_t)bytes, |
| 578 | (hmp->ronly ? " (RO)" : "(RW)")); |
| 579 | verbose = 1; |
| 580 | if (bytes > (rootmap->alloc_offset & HAMMER_OFF_LONG_MASK)) { |
| 581 | kprintf("Undo size is absurd, unable to mount\n"); |
| 582 | error = EIO; |
| 583 | goto fatal; |
| 584 | } |
| 585 | |
| 586 | /* |
| 587 | * Scan the REDOs backwards collecting REDO_TERM_* information. |
| 588 | * This information is only collected for the extended range, |
| 589 | * non-inclusive of any TERMs in the nominal UNDO range. |
| 590 | * |
| 591 | * If the stage2 extended range is inside the nominal undo range |
| 592 | * we have nothing to scan. |
| 593 | * |
| 594 | * This must fit in memory! |
| 595 | */ |
| 596 | if (first_offset < last_offset) { |
| 597 | /* |
| 598 | * [ first_offset........last_offset ] |
| 599 | */ |
| 600 | if (ext_offset < first_offset) { |
| 601 | dorscan = 1; |
| 602 | ext_bytes = first_offset - ext_offset; |
| 603 | } else if (ext_offset > last_offset) { |
| 604 | dorscan = 1; |
| 605 | ext_bytes = (rootmap->alloc_offset - ext_offset) + |
| 606 | (first_offset & HAMMER_OFF_LONG_MASK); |
| 607 | } else { |
| 608 | ext_bytes = -(ext_offset - first_offset); |
| 609 | dorscan = 0; |
| 610 | } |
| 611 | } else { |
| 612 | /* |
| 613 | * [......last_offset first_offset.....] |
| 614 | */ |
| 615 | if (ext_offset < last_offset) { |
| 616 | ext_bytes = -((rootmap->alloc_offset - first_offset) + |
| 617 | (ext_offset & HAMMER_OFF_LONG_MASK)); |
| 618 | dorscan = 0; |
| 619 | } else if (ext_offset > first_offset) { |
| 620 | ext_bytes = -(ext_offset - first_offset); |
| 621 | dorscan = 0; |
| 622 | } else { |
| 623 | ext_bytes = first_offset - ext_offset; |
| 624 | dorscan = 1; |
| 625 | } |
| 626 | } |
| 627 | |
| 628 | if (dorscan) { |
| 629 | scan_offset = first_offset; |
| 630 | kprintf("HAMMER(%s) Find extended redo %016jx, %jd extbytes\n", |
| 631 | root_volume->ondisk->vol_name, |
| 632 | (intmax_t)ext_offset, |
| 633 | (intmax_t)ext_bytes); |
| 634 | seqno = hmp->recover_stage2_seqno - 1; |
| 635 | for (;;) { |
| 636 | head = hammer_recover_scan_rev(hmp, root_volume, |
| 637 | &scan_offset, |
| 638 | &error, &buffer); |
| 639 | if (error) |
| 640 | break; |
| 641 | if (head->head.hdr_type != HAMMER_HEAD_TYPE_PAD) { |
| 642 | if (head->head.hdr_seq != seqno) { |
| 643 | error = ERANGE; |
| 644 | break; |
| 645 | } |
| 646 | error = hammer_recover_redo_rec( |
| 647 | hmp, &rterm_root, |
| 648 | scan_offset, &head->redo); |
| 649 | --seqno; |
| 650 | } |
| 651 | if (scan_offset == ext_offset) |
| 652 | break; |
| 653 | } |
| 654 | if (error) { |
| 655 | kprintf("HAMMER(%s) Find extended redo failed %d, " |
| 656 | "unable to run REDO\n", |
| 657 | root_volume->ondisk->vol_name, |
| 658 | error); |
| 659 | goto done; |
| 660 | } |
| 661 | } else { |
| 662 | kprintf("HAMMER(%s) Embedded extended redo %016jx, " |
| 663 | "%jd extbytes\n", |
| 664 | root_volume->ondisk->vol_name, |
| 665 | (intmax_t)ext_offset, |
| 666 | (intmax_t)ext_bytes); |
| 667 | } |
| 668 | |
| 669 | /* |
| 670 | * Scan the REDO forwards through the entire extended range. |
| 671 | * Anything with a previously recorded matching TERM is discarded. |
| 672 | */ |
| 673 | scan_offset = ext_offset; |
| 674 | bytes += ext_bytes; |
| 675 | |
| 676 | /* |
| 677 | * NOTE: when doing a forward scan the returned scan_offset is |
| 678 | * for the record following the returned record, so we |
| 679 | * have to play a bit. |
| 680 | */ |
| 681 | while ((int64_t)bytes > 0) { |
| 682 | KKASSERT(scan_offset != last_offset); |
| 683 | |
| 684 | oscan_offset = scan_offset; |
| 685 | head = hammer_recover_scan_fwd(hmp, root_volume, |
| 686 | &scan_offset, &error, &buffer); |
| 687 | if (error) |
| 688 | break; |
| 689 | |
| 690 | error = hammer_recover_redo_run(hmp, &rterm_root, |
| 691 | oscan_offset, &head->redo); |
| 692 | if (error) { |
| 693 | kprintf("HAMMER(%s) UNDO record at %016jx failed\n", |
| 694 | root_volume->ondisk->vol_name, |
| 695 | (intmax_t)scan_offset - head->head.hdr_size); |
| 696 | break; |
| 697 | } |
| 698 | bytes -= head->head.hdr_size; |
| 699 | } |
| 700 | KKASSERT(error || bytes == 0); |
| 701 | |
| 702 | done: |
| 703 | if (buffer) { |
| 704 | hammer_rel_buffer(buffer, 0); |
| 705 | buffer = NULL; |
| 706 | } |
| 707 | |
| 708 | /* |
| 709 | * Cleanup rterm tree |
| 710 | */ |
| 711 | { |
| 712 | hammer_rterm_t rterm; |
| 713 | hammer_rterm_entry_t rte; |
| 714 | |
| 715 | while ((rterm = RB_ROOT(&rterm_root)) != NULL) { |
| 716 | RB_REMOVE(hammer_rterm_rb_tree, &rterm_root, rterm); |
| 717 | while ((rte = rterm->term_list) != NULL) { |
| 718 | rterm->term_list = rte->next; |
| 719 | kfree(rte, hmp->m_misc); |
| 720 | } |
| 721 | kfree(rterm, hmp->m_misc); |
| 722 | } |
| 723 | } |
| 724 | |
| 725 | /* |
| 726 | * Finish up flushing (or discarding) recovered buffers by executing |
| 727 | * a normal flush cycle. Setting HMNT_UNDO_DIRTY bypasses degenerate |
| 728 | * case tests and forces the flush in order to update the FIFO indices. |
| 729 | * |
| 730 | * If a crash occurs during the flush the entire undo/redo will be |
| 731 | * re-run during recovery on the next mount. |
| 732 | */ |
| 733 | if (error == 0) { |
| 734 | if (rootmap->first_offset != rootmap->next_offset) |
| 735 | hmp->hflags |= HMNT_UNDO_DIRTY; |
| 736 | hammer_flusher_sync(hmp); |
| 737 | } |
| 738 | fatal: |
| 739 | hmp->flags &= ~HAMMER_MOUNT_REDO_RECOVERY_RUN; |
| 740 | if (verbose) { |
| 741 | kprintf("HAMMER(%s) End redo recovery\n", |
| 742 | root_volume->ondisk->vol_name); |
| 743 | } |
| 744 | return (error); |
| 745 | } |
| 746 | |
| 747 | /* |
| 748 | * Scan backwards from *scan_offsetp, return the FIFO record prior to the |
| 749 | * record at *scan_offsetp or NULL if an error occured. |
| 750 | * |
| 751 | * On return *scan_offsetp will be the offset of the returned record. |
| 752 | */ |
| 753 | hammer_fifo_any_t |
| 754 | hammer_recover_scan_rev(hammer_mount_t hmp, hammer_volume_t root_volume, |
| 755 | hammer_off_t *scan_offsetp, |
| 756 | int *errorp, struct hammer_buffer **bufferp) |
| 757 | { |
| 758 | hammer_off_t scan_offset; |
| 759 | hammer_blockmap_t rootmap; |
| 760 | hammer_fifo_any_t head; |
| 761 | hammer_fifo_tail_t tail; |
| 762 | |
| 763 | rootmap = &root_volume->ondisk->vol0_blockmap[HAMMER_ZONE_UNDO_INDEX]; |
| 764 | scan_offset = *scan_offsetp; |
| 765 | |
| 766 | if (hammer_debug_general & 0x0080) |
| 767 | kprintf("rev scan_offset %016jx\n", (intmax_t)scan_offset); |
| 768 | if (scan_offset == HAMMER_ZONE_ENCODE(HAMMER_ZONE_UNDO_INDEX, 0)) |
| 769 | scan_offset = rootmap->alloc_offset; |
| 770 | if (scan_offset - sizeof(*tail) < |
| 771 | HAMMER_ZONE_ENCODE(HAMMER_ZONE_UNDO_INDEX, 0)) { |
| 772 | kprintf("HAMMER(%s) UNDO record at %016jx FIFO underflow\n", |
| 773 | root_volume->ondisk->vol_name, |
| 774 | (intmax_t)scan_offset); |
| 775 | *errorp = EIO; |
| 776 | return (NULL); |
| 777 | } |
| 778 | tail = hammer_bread(hmp, scan_offset - sizeof(*tail), |
| 779 | errorp, bufferp); |
| 780 | if (*errorp) { |
| 781 | kprintf("HAMMER(%s) Unable to read UNDO TAIL " |
| 782 | "at %016jx\n", |
| 783 | root_volume->ondisk->vol_name, |
| 784 | (intmax_t)scan_offset - sizeof(*tail)); |
| 785 | return (NULL); |
| 786 | } |
| 787 | |
| 788 | if (hammer_check_tail_signature(tail, scan_offset) != 0) { |
| 789 | kprintf("HAMMER(%s) Illegal UNDO TAIL signature " |
| 790 | "at %016jx\n", |
| 791 | root_volume->ondisk->vol_name, |
| 792 | (intmax_t)scan_offset - sizeof(*tail)); |
| 793 | *errorp = EIO; |
| 794 | return (NULL); |
| 795 | } |
| 796 | head = (void *)((char *)tail + sizeof(*tail) - tail->tail_size); |
| 797 | *scan_offsetp = scan_offset - head->head.hdr_size; |
| 798 | |
| 799 | return (head); |
| 800 | } |
| 801 | |
| 802 | /* |
| 803 | * Scan forwards from *scan_offsetp, return the FIFO record or NULL if |
| 804 | * an error occured. |
| 805 | * |
| 806 | * On return *scan_offsetp will be the offset of the record following |
| 807 | * the returned record. |
| 808 | */ |
| 809 | hammer_fifo_any_t |
| 810 | hammer_recover_scan_fwd(hammer_mount_t hmp, hammer_volume_t root_volume, |
| 811 | hammer_off_t *scan_offsetp, |
| 812 | int *errorp, struct hammer_buffer **bufferp) |
| 813 | { |
| 814 | hammer_off_t scan_offset; |
| 815 | hammer_blockmap_t rootmap; |
| 816 | hammer_fifo_any_t head; |
| 817 | |
| 818 | rootmap = &root_volume->ondisk->vol0_blockmap[HAMMER_ZONE_UNDO_INDEX]; |
| 819 | scan_offset = *scan_offsetp; |
| 820 | |
| 821 | if (hammer_debug_general & 0x0080) |
| 822 | kprintf("fwd scan_offset %016jx\n", (intmax_t)scan_offset); |
| 823 | if (scan_offset == rootmap->alloc_offset) |
| 824 | scan_offset = HAMMER_ZONE_ENCODE(HAMMER_ZONE_UNDO_INDEX, 0); |
| 825 | |
| 826 | head = hammer_bread(hmp, scan_offset, errorp, bufferp); |
| 827 | if (*errorp) { |
| 828 | kprintf("HAMMER(%s) Unable to read UNDO HEAD at %016jx\n", |
| 829 | root_volume->ondisk->vol_name, |
| 830 | (intmax_t)scan_offset); |
| 831 | return (NULL); |
| 832 | } |
| 833 | |
| 834 | if (hammer_check_head_signature(&head->head, scan_offset) != 0) { |
| 835 | kprintf("HAMMER(%s) Illegal UNDO TAIL signature " |
| 836 | "at %016jx\n", |
| 837 | root_volume->ondisk->vol_name, |
| 838 | (intmax_t)scan_offset); |
| 839 | *errorp = EIO; |
| 840 | return (NULL); |
| 841 | } |
| 842 | scan_offset += head->head.hdr_size; |
| 843 | if (scan_offset == rootmap->alloc_offset) |
| 844 | scan_offset = HAMMER_ZONE_ENCODE(HAMMER_ZONE_UNDO_INDEX, 0); |
| 845 | *scan_offsetp = scan_offset; |
| 846 | |
| 847 | return (head); |
| 848 | } |
| 849 | |
| 850 | /* |
| 851 | * Helper function for hammer_check_{head,tail}_signature(). Check stuff |
| 852 | * once the head and tail has been established. |
| 853 | * |
| 854 | * This function validates the entire FIFO record wrapper. |
| 855 | */ |
| 856 | static __inline |
| 857 | int |
| 858 | _hammer_check_signature(hammer_fifo_head_t head, hammer_fifo_tail_t tail, |
| 859 | hammer_off_t beg_off) |
| 860 | { |
| 861 | hammer_off_t end_off; |
| 862 | u_int32_t crc; |
| 863 | int bytes; |
| 864 | |
| 865 | /* |
| 866 | * Check signatures. The tail signature is allowed to be the |
| 867 | * head signature only for 8-byte PADs. |
| 868 | */ |
| 869 | if (head->hdr_signature != HAMMER_HEAD_SIGNATURE) { |
| 870 | kprintf("HAMMER: FIFO record bad head signature " |
| 871 | "%04x at %016jx\n", |
| 872 | head->hdr_signature, |
| 873 | (intmax_t)beg_off); |
| 874 | return(2); |
| 875 | } |
| 876 | if (head->hdr_size < HAMMER_HEAD_ALIGN || |
| 877 | (head->hdr_size & HAMMER_HEAD_ALIGN_MASK)) { |
| 878 | kprintf("HAMMER: FIFO record unaligned or bad size" |
| 879 | "%04x at %016jx\n", |
| 880 | head->hdr_size, |
| 881 | (intmax_t)beg_off); |
| 882 | return(2); |
| 883 | } |
| 884 | end_off = beg_off + head->hdr_size; |
| 885 | |
| 886 | if (head->hdr_type != HAMMER_HEAD_TYPE_PAD || |
| 887 | (size_t)(end_off - beg_off) != sizeof(*tail)) { |
| 888 | if (head->hdr_type != tail->tail_type) { |
| 889 | kprintf("HAMMER: FIFO record head/tail type mismatch " |
| 890 | "%04x %04x at %016jx\n", |
| 891 | head->hdr_type, tail->tail_type, |
| 892 | (intmax_t)beg_off); |
| 893 | return(2); |
| 894 | } |
| 895 | if (head->hdr_size != tail->tail_size) { |
| 896 | kprintf("HAMMER: FIFO record head/tail size mismatch " |
| 897 | "%04x %04x at %016jx\n", |
| 898 | head->hdr_size, tail->tail_size, |
| 899 | (intmax_t)beg_off); |
| 900 | return(2); |
| 901 | } |
| 902 | if (tail->tail_signature != HAMMER_TAIL_SIGNATURE) { |
| 903 | kprintf("HAMMER: FIFO record bad tail signature " |
| 904 | "%04x at %016jx\n", |
| 905 | tail->tail_signature, |
| 906 | (intmax_t)beg_off); |
| 907 | return(3); |
| 908 | } |
| 909 | } |
| 910 | |
| 911 | /* |
| 912 | * Non-PAD records must have a CRC and must be sized at |
| 913 | * least large enough to fit the head and tail. |
| 914 | */ |
| 915 | if (head->hdr_type != HAMMER_HEAD_TYPE_PAD) { |
| 916 | crc = crc32(head, HAMMER_FIFO_HEAD_CRCOFF) ^ |
| 917 | crc32(head + 1, head->hdr_size - sizeof(*head)); |
| 918 | if (head->hdr_crc != crc) { |
| 919 | kprintf("HAMMER: FIFO record CRC failed %08x %08x " |
| 920 | "at %016jx\n", |
| 921 | head->hdr_crc, crc, |
| 922 | (intmax_t)beg_off); |
| 923 | return(EIO); |
| 924 | } |
| 925 | if (head->hdr_size < sizeof(*head) + sizeof(*tail)) { |
| 926 | kprintf("HAMMER: FIFO record too small " |
| 927 | "%04x at %016jx\n", |
| 928 | head->hdr_size, |
| 929 | (intmax_t)beg_off); |
| 930 | return(EIO); |
| 931 | } |
| 932 | } |
| 933 | |
| 934 | /* |
| 935 | * Check the tail |
| 936 | */ |
| 937 | bytes = head->hdr_size; |
| 938 | tail = (void *)((char *)head + bytes - sizeof(*tail)); |
| 939 | if (tail->tail_size != head->hdr_size) { |
| 940 | kprintf("HAMMER: Bad tail size %04x vs %04x at %016jx\n", |
| 941 | tail->tail_size, head->hdr_size, |
| 942 | (intmax_t)beg_off); |
| 943 | return(EIO); |
| 944 | } |
| 945 | if (tail->tail_type != head->hdr_type) { |
| 946 | kprintf("HAMMER: Bad tail type %04x vs %04x at %016jx\n", |
| 947 | tail->tail_type, head->hdr_type, |
| 948 | (intmax_t)beg_off); |
| 949 | return(EIO); |
| 950 | } |
| 951 | |
| 952 | return(0); |
| 953 | } |
| 954 | |
| 955 | /* |
| 956 | * Check that the FIFO record is in-bounds given the head and the |
| 957 | * hammer offset. |
| 958 | * |
| 959 | * Also checks that the head and tail structures agree with each other, |
| 960 | * but does not check beyond the signature, type, and size. |
| 961 | */ |
| 962 | static int |
| 963 | hammer_check_head_signature(hammer_fifo_head_t head, hammer_off_t beg_off) |
| 964 | { |
| 965 | hammer_fifo_tail_t tail; |
| 966 | hammer_off_t end_off; |
| 967 | |
| 968 | /* |
| 969 | * head overlaps buffer boundary. This could be a PAD so only |
| 970 | * check the minimum PAD size here. |
| 971 | */ |
| 972 | if (((beg_off + sizeof(*tail) - 1) ^ (beg_off)) & ~HAMMER_BUFMASK64) |
| 973 | return(1); |
| 974 | |
| 975 | /* |
| 976 | * Calculate the ending offset and make sure the record does |
| 977 | * not cross a buffer boundary. |
| 978 | */ |
| 979 | end_off = beg_off + head->hdr_size; |
| 980 | if ((beg_off ^ (end_off - 1)) & ~HAMMER_BUFMASK64) |
| 981 | return(1); |
| 982 | tail = (void *)((char *)head + head->hdr_size - sizeof(*tail)); |
| 983 | return (_hammer_check_signature(head, tail, beg_off)); |
| 984 | } |
| 985 | |
| 986 | /* |
| 987 | * Check that the FIFO record is in-bounds given the tail and the |
| 988 | * hammer offset. The offset is pointing at the ending boundary of the |
| 989 | * record. |
| 990 | * |
| 991 | * Also checks that the head and tail structures agree with each other, |
| 992 | * but does not check beyond the signature, type, and size. |
| 993 | */ |
| 994 | static int |
| 995 | hammer_check_tail_signature(hammer_fifo_tail_t tail, hammer_off_t end_off) |
| 996 | { |
| 997 | hammer_fifo_head_t head; |
| 998 | hammer_off_t beg_off; |
| 999 | |
| 1000 | /* |
| 1001 | * tail overlaps buffer boundary |
| 1002 | */ |
| 1003 | if (((end_off - sizeof(*tail)) ^ (end_off - 1)) & ~HAMMER_BUFMASK64) |
| 1004 | return(1); |
| 1005 | |
| 1006 | /* |
| 1007 | * Calculate the begining offset and make sure the record does |
| 1008 | * not cross a buffer boundary. |
| 1009 | */ |
| 1010 | beg_off = end_off - tail->tail_size; |
| 1011 | if ((beg_off ^ (end_off - 1)) & ~HAMMER_BUFMASK64) |
| 1012 | return(1); |
| 1013 | head = (void *)((char *)tail + sizeof(*tail) - tail->tail_size); |
| 1014 | return (_hammer_check_signature(head, tail, beg_off)); |
| 1015 | } |
| 1016 | |
| 1017 | static int |
| 1018 | hammer_recover_undo(hammer_mount_t hmp, hammer_volume_t root_volume, |
| 1019 | hammer_fifo_undo_t undo) |
| 1020 | { |
| 1021 | hammer_volume_t volume; |
| 1022 | hammer_buffer_t buffer; |
| 1023 | hammer_off_t buf_offset; |
| 1024 | int zone; |
| 1025 | int error; |
| 1026 | int vol_no; |
| 1027 | int bytes; |
| 1028 | u_int32_t offset; |
| 1029 | |
| 1030 | /* |
| 1031 | * Only process UNDO records. Flag if we find other records to |
| 1032 | * optimize stage2 recovery. |
| 1033 | */ |
| 1034 | if (undo->head.hdr_type != HAMMER_HEAD_TYPE_UNDO) |
| 1035 | return(0); |
| 1036 | |
| 1037 | /* |
| 1038 | * Validate the UNDO record. |
| 1039 | */ |
| 1040 | bytes = undo->head.hdr_size - sizeof(*undo) - |
| 1041 | sizeof(struct hammer_fifo_tail); |
| 1042 | if (bytes < 0 || undo->undo_data_bytes < 0 || |
| 1043 | undo->undo_data_bytes > bytes) { |
| 1044 | kprintf("HAMMER: Corrupt UNDO record, undo_data_bytes %d/%d\n", |
| 1045 | undo->undo_data_bytes, bytes); |
| 1046 | return(EIO); |
| 1047 | } |
| 1048 | |
| 1049 | bytes = undo->undo_data_bytes; |
| 1050 | |
| 1051 | /* |
| 1052 | * The undo offset may only be a zone-1 or zone-2 offset. |
| 1053 | * |
| 1054 | * Currently we only support a zone-1 offset representing the |
| 1055 | * volume header. |
| 1056 | */ |
| 1057 | zone = HAMMER_ZONE_DECODE(undo->undo_offset); |
| 1058 | offset = undo->undo_offset & HAMMER_BUFMASK; |
| 1059 | |
| 1060 | if (offset + bytes > HAMMER_BUFSIZE) { |
| 1061 | kprintf("HAMMER: Corrupt UNDO record, bad offset\n"); |
| 1062 | return (EIO); |
| 1063 | } |
| 1064 | |
| 1065 | switch(zone) { |
| 1066 | case HAMMER_ZONE_RAW_VOLUME_INDEX: |
| 1067 | vol_no = HAMMER_VOL_DECODE(undo->undo_offset); |
| 1068 | volume = hammer_get_volume(hmp, vol_no, &error); |
| 1069 | if (volume == NULL) { |
| 1070 | kprintf("HAMMER: UNDO record, " |
| 1071 | "cannot access volume %d\n", vol_no); |
| 1072 | break; |
| 1073 | } |
| 1074 | hammer_modify_volume(NULL, volume, NULL, 0); |
| 1075 | hammer_recover_copy_undo(undo->undo_offset, |
| 1076 | (char *)(undo + 1), |
| 1077 | (char *)volume->ondisk + offset, |
| 1078 | bytes); |
| 1079 | hammer_modify_volume_done(volume); |
| 1080 | |
| 1081 | /* |
| 1082 | * Multiple modifications may be made to the same buffer. |
| 1083 | * Also, the volume header cannot be written out until |
| 1084 | * everything else has been flushed. This also |
| 1085 | * covers the read-only case by preventing the kernel from |
| 1086 | * flushing the buffer. |
| 1087 | */ |
| 1088 | if (volume->io.recovered == 0) |
| 1089 | volume->io.recovered = 1; |
| 1090 | else |
| 1091 | hammer_rel_volume(volume, 0); |
| 1092 | break; |
| 1093 | case HAMMER_ZONE_RAW_BUFFER_INDEX: |
| 1094 | buf_offset = undo->undo_offset & ~HAMMER_BUFMASK64; |
| 1095 | buffer = hammer_get_buffer(hmp, buf_offset, HAMMER_BUFSIZE, |
| 1096 | 0, &error); |
| 1097 | if (buffer == NULL) { |
| 1098 | kprintf("HAMMER: UNDO record, " |
| 1099 | "cannot access buffer %016jx\n", |
| 1100 | (intmax_t)undo->undo_offset); |
| 1101 | break; |
| 1102 | } |
| 1103 | hammer_modify_buffer(NULL, buffer, NULL, 0); |
| 1104 | hammer_recover_copy_undo(undo->undo_offset, |
| 1105 | (char *)(undo + 1), |
| 1106 | (char *)buffer->ondisk + offset, |
| 1107 | bytes); |
| 1108 | hammer_modify_buffer_done(buffer); |
| 1109 | |
| 1110 | /* |
| 1111 | * Multiple modifications may be made to the same buffer, |
| 1112 | * improve performance by delaying the flush. This also |
| 1113 | * covers the read-only case by preventing the kernel from |
| 1114 | * flushing the buffer. |
| 1115 | */ |
| 1116 | if (buffer->io.recovered == 0) |
| 1117 | buffer->io.recovered = 1; |
| 1118 | else |
| 1119 | hammer_rel_buffer(buffer, 0); |
| 1120 | break; |
| 1121 | default: |
| 1122 | kprintf("HAMMER: Corrupt UNDO record\n"); |
| 1123 | error = EIO; |
| 1124 | } |
| 1125 | return (error); |
| 1126 | } |
| 1127 | |
| 1128 | static void |
| 1129 | hammer_recover_copy_undo(hammer_off_t undo_offset, |
| 1130 | char *src, char *dst, int bytes) |
| 1131 | { |
| 1132 | if (hammer_debug_general & 0x0080) { |
| 1133 | kprintf("UNDO %016jx: %d\n", |
| 1134 | (intmax_t)undo_offset, bytes); |
| 1135 | } |
| 1136 | #if 0 |
| 1137 | kprintf("UNDO %016jx:", (intmax_t)undo_offset); |
| 1138 | hammer_recover_debug_dump(22, dst, bytes); |
| 1139 | kprintf("%22s", "to:"); |
| 1140 | hammer_recover_debug_dump(22, src, bytes); |
| 1141 | #endif |
| 1142 | bcopy(src, dst, bytes); |
| 1143 | } |
| 1144 | |
| 1145 | /* |
| 1146 | * Record HAMMER_REDO_TERM_WRITE and HAMMER_REDO_TERM_TRUNC operations |
| 1147 | * during the backwards scan of the extended UNDO/REDO FIFO. This scan |
| 1148 | * does not include the nominal UNDO range, just the extended range. |
| 1149 | */ |
| 1150 | int |
| 1151 | hammer_recover_redo_rec(hammer_mount_t hmp, struct hammer_rterm_rb_tree *root, |
| 1152 | hammer_off_t scan_offset, hammer_fifo_redo_t redo) |
| 1153 | { |
| 1154 | hammer_rterm_t rterm; |
| 1155 | hammer_rterm_t nrterm; |
| 1156 | hammer_rterm_entry_t rte; |
| 1157 | |
| 1158 | if (redo->head.hdr_type != HAMMER_HEAD_TYPE_REDO) |
| 1159 | return(0); |
| 1160 | if (redo->redo_flags != HAMMER_REDO_TERM_WRITE && |
| 1161 | redo->redo_flags != HAMMER_REDO_TERM_TRUNC) { |
| 1162 | return(0); |
| 1163 | } |
| 1164 | |
| 1165 | nrterm = kmalloc(sizeof(*nrterm), hmp->m_misc, M_WAITOK|M_ZERO); |
| 1166 | nrterm->redo_objid = redo->redo_objid; |
| 1167 | nrterm->redo_localization = redo->redo_localization; |
| 1168 | nrterm->redo_flags = redo->redo_flags; |
| 1169 | nrterm->redo_offset = redo->redo_offset; |
| 1170 | |
| 1171 | rterm = RB_INSERT(hammer_rterm_rb_tree, root, nrterm); |
| 1172 | if (rterm) |
| 1173 | kfree(nrterm, hmp->m_misc); |
| 1174 | else |
| 1175 | rterm = nrterm; |
| 1176 | |
| 1177 | if (bootverbose) { |
| 1178 | kprintf("record record %016jx objid %016jx " |
| 1179 | "offset %016jx flags %08x\n", |
| 1180 | (intmax_t)scan_offset, |
| 1181 | (intmax_t)redo->redo_objid, |
| 1182 | (intmax_t)redo->redo_offset, |
| 1183 | (int)redo->redo_flags); |
| 1184 | } |
| 1185 | |
| 1186 | /* |
| 1187 | * Scan in reverse order, rte prepended, so the rte list will be |
| 1188 | * in forward order. |
| 1189 | */ |
| 1190 | rte = kmalloc(sizeof(*rte), hmp->m_misc, M_WAITOK|M_ZERO); |
| 1191 | rte->fifo_offset = scan_offset; |
| 1192 | rte->next = rterm->term_list; |
| 1193 | rterm->term_list = rte; |
| 1194 | |
| 1195 | return(0); |
| 1196 | } |
| 1197 | |
| 1198 | /* |
| 1199 | * Execute HAMMER_REDO_WRITE and HAMMER_REDO_TRUNC operations during |
| 1200 | * the forwards scan of the entire extended UNDO/REDO FIFO range. |
| 1201 | * |
| 1202 | * Records matching previously recorded TERMs have already been committed |
| 1203 | * and are ignored. |
| 1204 | */ |
| 1205 | int |
| 1206 | hammer_recover_redo_run(hammer_mount_t hmp, struct hammer_rterm_rb_tree *root, |
| 1207 | hammer_off_t scan_offset, hammer_fifo_redo_t redo) |
| 1208 | { |
| 1209 | struct hammer_rterm rtval; |
| 1210 | hammer_rterm_t rterm; |
| 1211 | hammer_rterm_entry_t rte; |
| 1212 | |
| 1213 | if (redo->head.hdr_type != HAMMER_HEAD_TYPE_REDO) |
| 1214 | return(0); |
| 1215 | |
| 1216 | switch(redo->redo_flags) { |
| 1217 | case HAMMER_REDO_WRITE: |
| 1218 | case HAMMER_REDO_TRUNC: |
| 1219 | /* |
| 1220 | * We hit a REDO request. The REDO request is only executed |
| 1221 | * if there is no matching TERM. |
| 1222 | */ |
| 1223 | bzero(&rtval, sizeof(rtval)); |
| 1224 | rtval.redo_objid = redo->redo_objid; |
| 1225 | rtval.redo_localization = redo->redo_localization; |
| 1226 | rtval.redo_offset = redo->redo_offset; |
| 1227 | rtval.redo_flags = (redo->redo_flags == HAMMER_REDO_WRITE) ? |
| 1228 | HAMMER_REDO_TERM_WRITE : |
| 1229 | HAMMER_REDO_TERM_TRUNC; |
| 1230 | |
| 1231 | rterm = RB_FIND(hammer_rterm_rb_tree, root, &rtval); |
| 1232 | if (rterm) { |
| 1233 | if (bootverbose) { |
| 1234 | kprintf("ignore record %016jx objid %016jx " |
| 1235 | "offset %016jx flags %08x\n", |
| 1236 | (intmax_t)scan_offset, |
| 1237 | (intmax_t)redo->redo_objid, |
| 1238 | (intmax_t)redo->redo_offset, |
| 1239 | (int)redo->redo_flags); |
| 1240 | } |
| 1241 | break; |
| 1242 | } |
| 1243 | if (bootverbose) { |
| 1244 | kprintf("run record %016jx objid %016jx " |
| 1245 | "offset %016jx flags %08x\n", |
| 1246 | (intmax_t)scan_offset, |
| 1247 | (intmax_t)redo->redo_objid, |
| 1248 | (intmax_t)redo->redo_offset, |
| 1249 | (int)redo->redo_flags); |
| 1250 | } |
| 1251 | |
| 1252 | /* |
| 1253 | * Redo stage2 can access a live filesystem, acquire the |
| 1254 | * vnode. |
| 1255 | */ |
| 1256 | hammer_recover_redo_exec(hmp, redo); |
| 1257 | break; |
| 1258 | case HAMMER_REDO_TERM_WRITE: |
| 1259 | case HAMMER_REDO_TERM_TRUNC: |
| 1260 | /* |
| 1261 | * As we encounter TERMs in the forward scan we remove |
| 1262 | * them. Once the forward scan hits the nominal undo range |
| 1263 | * there will be no more recorded TERMs. |
| 1264 | */ |
| 1265 | bzero(&rtval, sizeof(rtval)); |
| 1266 | rtval.redo_objid = redo->redo_objid; |
| 1267 | rtval.redo_localization = redo->redo_localization; |
| 1268 | rtval.redo_flags = redo->redo_flags; |
| 1269 | rtval.redo_offset = redo->redo_offset; |
| 1270 | |
| 1271 | rterm = RB_FIND(hammer_rterm_rb_tree, root, &rtval); |
| 1272 | if (rterm) { |
| 1273 | if ((rte = rterm->term_list) != NULL) { |
| 1274 | KKASSERT(rte->fifo_offset == scan_offset); |
| 1275 | rterm->term_list = rte->next; |
| 1276 | kfree(rte, hmp->m_misc); |
| 1277 | } |
| 1278 | } |
| 1279 | break; |
| 1280 | } |
| 1281 | return(0); |
| 1282 | } |
| 1283 | |
| 1284 | static void |
| 1285 | hammer_recover_redo_exec(hammer_mount_t hmp, hammer_fifo_redo_t redo) |
| 1286 | { |
| 1287 | struct hammer_transaction trans; |
| 1288 | struct vattr va; |
| 1289 | struct hammer_inode *ip; |
| 1290 | struct vnode *vp = NULL; |
| 1291 | int error; |
| 1292 | |
| 1293 | hammer_start_transaction(&trans, hmp); |
| 1294 | |
| 1295 | ip = hammer_get_inode(&trans, NULL, redo->redo_objid, |
| 1296 | HAMMER_MAX_TID, redo->redo_localization, |
| 1297 | 0, &error); |
| 1298 | if (ip == NULL) { |
| 1299 | kprintf("unable to find objid %016jx:%08x\n", |
| 1300 | (intmax_t)redo->redo_objid, redo->redo_localization); |
| 1301 | goto done2; |
| 1302 | } |
| 1303 | error = hammer_get_vnode(ip, &vp); |
| 1304 | if (error) { |
| 1305 | kprintf("unable to acquire vnode for %016jx:%08x\n", |
| 1306 | (intmax_t)redo->redo_objid, redo->redo_localization); |
| 1307 | goto done1; |
| 1308 | } |
| 1309 | |
| 1310 | switch(redo->redo_flags) { |
| 1311 | case HAMMER_REDO_WRITE: |
| 1312 | error = VOP_OPEN(vp, FREAD|FWRITE, proc0.p_ucred, NULL); |
| 1313 | if (error) { |
| 1314 | kprintf("vn_rdwr open %016jx:%08x returned %d\n", |
| 1315 | (intmax_t)redo->redo_objid, |
| 1316 | redo->redo_localization, error); |
| 1317 | break; |
| 1318 | } |
| 1319 | vn_unlock(vp); |
| 1320 | error = vn_rdwr(UIO_WRITE, vp, (void *)(redo + 1), |
| 1321 | redo->redo_data_bytes, |
| 1322 | redo->redo_offset, UIO_SYSSPACE, |
| 1323 | 0, proc0.p_ucred, NULL); |
| 1324 | vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); |
| 1325 | if (error) { |
| 1326 | kprintf("write %016jx:%08x returned %d\n", |
| 1327 | (intmax_t)redo->redo_objid, |
| 1328 | redo->redo_localization, error); |
| 1329 | } |
| 1330 | VOP_CLOSE(vp, FREAD|FWRITE); |
| 1331 | break; |
| 1332 | case HAMMER_REDO_TRUNC: |
| 1333 | VATTR_NULL(&va); |
| 1334 | va.va_size = redo->redo_offset; |
| 1335 | error = VOP_SETATTR(vp, &va, proc0.p_ucred); |
| 1336 | if (error) { |
| 1337 | kprintf("setattr offset %016jx error %d\n", |
| 1338 | (intmax_t)redo->redo_offset, error); |
| 1339 | } |
| 1340 | break; |
| 1341 | } |
| 1342 | vput(vp); |
| 1343 | done1: |
| 1344 | hammer_rel_inode(ip, 0); |
| 1345 | done2: |
| 1346 | hammer_done_transaction(&trans); |
| 1347 | } |
| 1348 | |
| 1349 | /* |
| 1350 | * RB tree compare function. Note that REDO_TERM_TRUNC ops ignore |
| 1351 | * the offset. |
| 1352 | * |
| 1353 | * WRITE@0 TERM@0 WRITE@0 .... (no TERM@0) etc. |
| 1354 | */ |
| 1355 | static int |
| 1356 | hammer_rterm_rb_cmp(hammer_rterm_t rt1, hammer_rterm_t rt2) |
| 1357 | { |
| 1358 | if (rt1->redo_objid < rt2->redo_objid) |
| 1359 | return(-1); |
| 1360 | if (rt1->redo_objid > rt2->redo_objid) |
| 1361 | return(1); |
| 1362 | if (rt1->redo_localization < rt2->redo_localization) |
| 1363 | return(-1); |
| 1364 | if (rt1->redo_localization > rt2->redo_localization) |
| 1365 | return(1); |
| 1366 | if (rt1->redo_flags < rt2->redo_flags) |
| 1367 | return(-1); |
| 1368 | if (rt1->redo_flags > rt2->redo_flags) |
| 1369 | return(1); |
| 1370 | if (rt1->redo_flags != HAMMER_REDO_TERM_TRUNC) { |
| 1371 | if (rt1->redo_offset < rt2->redo_offset) |
| 1372 | return(-1); |
| 1373 | if (rt1->redo_offset > rt2->redo_offset) |
| 1374 | return(1); |
| 1375 | } |
| 1376 | return(0); |
| 1377 | } |
| 1378 | |
| 1379 | #if 0 |
| 1380 | |
| 1381 | static void |
| 1382 | hammer_recover_debug_dump(int w, char *buf, int bytes) |
| 1383 | { |
| 1384 | int i; |
| 1385 | |
| 1386 | for (i = 0; i < bytes; ++i) { |
| 1387 | if (i && (i & 15) == 0) |
| 1388 | kprintf("\n%*.*s", w, w, ""); |
| 1389 | kprintf(" %02x", (unsigned char)buf[i]); |
| 1390 | } |
| 1391 | kprintf("\n"); |
| 1392 | } |
| 1393 | |
| 1394 | #endif |
| 1395 | |
| 1396 | /* |
| 1397 | * Flush recovered buffers from recovery operations. The call to this |
| 1398 | * routine may be delayed if a read-only mount was made and then later |
| 1399 | * upgraded to read-write. This routine is also called when unmounting |
| 1400 | * a read-only mount to clean out recovered (dirty) buffers which we |
| 1401 | * couldn't flush (because the mount is read-only). |
| 1402 | * |
| 1403 | * The volume header is always written last. The UNDO FIFO will be forced |
| 1404 | * to zero-length by setting next_offset to first_offset. This leaves the |
| 1405 | * (now stale) UNDO information used to recover the disk available for |
| 1406 | * forensic analysis. |
| 1407 | * |
| 1408 | * final is typically 0 or 1. The volume header is only written if final |
| 1409 | * is 1. If final is -1 the recovered buffers are discarded instead of |
| 1410 | * written and root_volume can also be passed as NULL in that case. |
| 1411 | */ |
| 1412 | static int hammer_recover_flush_volume_callback(hammer_volume_t, void *); |
| 1413 | static int hammer_recover_flush_buffer_callback(hammer_buffer_t, void *); |
| 1414 | |
| 1415 | void |
| 1416 | hammer_recover_flush_buffers(hammer_mount_t hmp, hammer_volume_t root_volume, |
| 1417 | int final) |
| 1418 | { |
| 1419 | /* |
| 1420 | * Flush the buffers out asynchronously, wait for all the I/O to |
| 1421 | * complete, then do it again to destroy the buffer cache buffer |
| 1422 | * so it doesn't alias something later on. |
| 1423 | */ |
| 1424 | RB_SCAN(hammer_buf_rb_tree, &hmp->rb_bufs_root, NULL, |
| 1425 | hammer_recover_flush_buffer_callback, &final); |
| 1426 | hammer_io_wait_all(hmp, "hmrrcw", 1); |
| 1427 | RB_SCAN(hammer_buf_rb_tree, &hmp->rb_bufs_root, NULL, |
| 1428 | hammer_recover_flush_buffer_callback, &final); |
| 1429 | |
| 1430 | /* |
| 1431 | * Flush all volume headers except the root volume. If final < 0 |
| 1432 | * we discard all volume headers including the root volume. |
| 1433 | */ |
| 1434 | if (final >= 0) { |
| 1435 | RB_SCAN(hammer_vol_rb_tree, &hmp->rb_vols_root, NULL, |
| 1436 | hammer_recover_flush_volume_callback, root_volume); |
| 1437 | } else { |
| 1438 | RB_SCAN(hammer_vol_rb_tree, &hmp->rb_vols_root, NULL, |
| 1439 | hammer_recover_flush_volume_callback, NULL); |
| 1440 | } |
| 1441 | |
| 1442 | /* |
| 1443 | * Finalize the root volume header. |
| 1444 | */ |
| 1445 | if (root_volume && root_volume->io.recovered && final > 0) { |
| 1446 | hammer_io_wait_all(hmp, "hmrflx", 1); |
| 1447 | root_volume->io.recovered = 0; |
| 1448 | hammer_io_flush(&root_volume->io, 0); |
| 1449 | hammer_rel_volume(root_volume, 0); |
| 1450 | hammer_io_wait_all(hmp, "hmrfly", 1); |
| 1451 | } |
| 1452 | } |
| 1453 | |
| 1454 | /* |
| 1455 | * Callback to flush volume headers. If discarding data will be NULL and |
| 1456 | * all volume headers (including the root volume) will be discarded. |
| 1457 | * Otherwise data is the root_volume and we flush all volume headers |
| 1458 | * EXCEPT the root_volume. |
| 1459 | * |
| 1460 | * Clear any I/O error or modified condition when discarding buffers to |
| 1461 | * clean up the reference count, otherwise the buffer may have extra refs |
| 1462 | * on it. |
| 1463 | */ |
| 1464 | static |
| 1465 | int |
| 1466 | hammer_recover_flush_volume_callback(hammer_volume_t volume, void *data) |
| 1467 | { |
| 1468 | hammer_volume_t root_volume = data; |
| 1469 | |
| 1470 | if (volume->io.recovered && volume != root_volume) { |
| 1471 | volume->io.recovered = 0; |
| 1472 | if (root_volume != NULL) { |
| 1473 | hammer_io_flush(&volume->io, 0); |
| 1474 | } else { |
| 1475 | hammer_io_clear_error(&volume->io); |
| 1476 | hammer_io_clear_modify(&volume->io, 1); |
| 1477 | } |
| 1478 | hammer_rel_volume(volume, 0); |
| 1479 | } |
| 1480 | return(0); |
| 1481 | } |
| 1482 | |
| 1483 | /* |
| 1484 | * Flush or discard recovered I/O buffers. |
| 1485 | * |
| 1486 | * Clear any I/O error or modified condition when discarding buffers to |
| 1487 | * clean up the reference count, otherwise the buffer may have extra refs |
| 1488 | * on it. |
| 1489 | */ |
| 1490 | static |
| 1491 | int |
| 1492 | hammer_recover_flush_buffer_callback(hammer_buffer_t buffer, void *data) |
| 1493 | { |
| 1494 | int final = *(int *)data; |
| 1495 | int flush; |
| 1496 | |
| 1497 | if (buffer->io.recovered) { |
| 1498 | buffer->io.recovered = 0; |
| 1499 | buffer->io.reclaim = 1; |
| 1500 | if (final < 0) { |
| 1501 | hammer_io_clear_error(&buffer->io); |
| 1502 | hammer_io_clear_modify(&buffer->io, 1); |
| 1503 | } else { |
| 1504 | hammer_io_flush(&buffer->io, 0); |
| 1505 | } |
| 1506 | hammer_rel_buffer(buffer, 0); |
| 1507 | } else { |
| 1508 | flush = hammer_ref_interlock(&buffer->io.lock); |
| 1509 | if (flush) |
| 1510 | ++hammer_count_refedbufs; |
| 1511 | |
| 1512 | if (final < 0) { |
| 1513 | hammer_io_clear_error(&buffer->io); |
| 1514 | hammer_io_clear_modify(&buffer->io, 1); |
| 1515 | } |
| 1516 | KKASSERT(hammer_oneref(&buffer->io.lock)); |
| 1517 | buffer->io.reclaim = 1; |
| 1518 | hammer_rel_buffer(buffer, flush); |
| 1519 | } |
| 1520 | return(0); |
| 1521 | } |
| 1522 | |