Joern Engel | 5db53f3 | 2009-11-20 20:13:39 +0100 | [diff] [blame] | 1 | /* |
| 2 | * fs/logfs/gc.c - garbage collection code |
| 3 | * |
| 4 | * As should be obvious for Linux kernel code, license is GPLv2 |
| 5 | * |
| 6 | * Copyright (c) 2005-2008 Joern Engel <joern@logfs.org> |
| 7 | */ |
| 8 | #include "logfs.h" |
| 9 | #include <linux/sched.h> |
Tejun Heo | 5a0e3ad | 2010-03-24 17:04:11 +0900 | [diff] [blame] | 10 | #include <linux/slab.h> |
Joern Engel | 5db53f3 | 2009-11-20 20:13:39 +0100 | [diff] [blame] | 11 | |
| 12 | /* |
| 13 | * Wear leveling needs to kick in when the difference between low erase |
| 14 | * counts and high erase counts gets too big. A good value for "too big" |
| 15 | * may be somewhat below 10% of maximum erase count for the device. |
| 16 | * Why not 397, to pick a nice round number with no specific meaning? :) |
| 17 | * |
| 18 | * WL_RATELIMIT is the minimum time between two wear level events. A huge |
| 19 | * number of segments may fulfil the requirements for wear leveling at the |
| 20 | * same time. If that happens we don't want to cause a latency from hell, |
| 21 | * but just gently pick one segment every so often and minimize overhead. |
| 22 | */ |
| 23 | #define WL_DELTA 397 |
| 24 | #define WL_RATELIMIT 100 |
| 25 | #define MAX_OBJ_ALIASES 2600 |
| 26 | #define SCAN_RATIO 512 /* number of scanned segments per gc'd segment */ |
| 27 | #define LIST_SIZE 64 /* base size of candidate lists */ |
| 28 | #define SCAN_ROUNDS 128 /* maximum number of complete medium scans */ |
| 29 | #define SCAN_ROUNDS_HIGH 4 /* maximum number of higher-level scans */ |
| 30 | |
| 31 | static int no_free_segments(struct super_block *sb) |
| 32 | { |
| 33 | struct logfs_super *super = logfs_super(sb); |
| 34 | |
| 35 | return super->s_free_list.count; |
| 36 | } |
| 37 | |
| 38 | /* journal has distance -1, top-most ifile layer distance 0 */ |
| 39 | static u8 root_distance(struct super_block *sb, gc_level_t __gc_level) |
| 40 | { |
| 41 | struct logfs_super *super = logfs_super(sb); |
| 42 | u8 gc_level = (__force u8)__gc_level; |
| 43 | |
| 44 | switch (gc_level) { |
| 45 | case 0: /* fall through */ |
| 46 | case 1: /* fall through */ |
| 47 | case 2: /* fall through */ |
| 48 | case 3: |
| 49 | /* file data or indirect blocks */ |
| 50 | return super->s_ifile_levels + super->s_iblock_levels - gc_level; |
| 51 | case 6: /* fall through */ |
| 52 | case 7: /* fall through */ |
| 53 | case 8: /* fall through */ |
| 54 | case 9: |
| 55 | /* inode file data or indirect blocks */ |
| 56 | return super->s_ifile_levels - (gc_level - 6); |
| 57 | default: |
| 58 | printk(KERN_ERR"LOGFS: segment of unknown level %x found\n", |
| 59 | gc_level); |
| 60 | WARN_ON(1); |
| 61 | return super->s_ifile_levels + super->s_iblock_levels; |
| 62 | } |
| 63 | } |
| 64 | |
| 65 | static int segment_is_reserved(struct super_block *sb, u32 segno) |
| 66 | { |
| 67 | struct logfs_super *super = logfs_super(sb); |
| 68 | struct logfs_area *area; |
| 69 | void *reserved; |
| 70 | int i; |
| 71 | |
| 72 | /* Some segments are reserved. Just pretend they were all valid */ |
| 73 | reserved = btree_lookup32(&super->s_reserved_segments, segno); |
| 74 | if (reserved) |
| 75 | return 1; |
| 76 | |
| 77 | /* Currently open segments */ |
| 78 | for_each_area(i) { |
| 79 | area = super->s_area[i]; |
| 80 | if (area->a_is_open && area->a_segno == segno) |
| 81 | return 1; |
| 82 | } |
| 83 | |
| 84 | return 0; |
| 85 | } |
| 86 | |
| 87 | static void logfs_mark_segment_bad(struct super_block *sb, u32 segno) |
| 88 | { |
| 89 | BUG(); |
| 90 | } |
| 91 | |
| 92 | /* |
| 93 | * Returns the bytes consumed by valid objects in this segment. Object headers |
| 94 | * are counted, the segment header is not. |
| 95 | */ |
| 96 | static u32 logfs_valid_bytes(struct super_block *sb, u32 segno, u32 *ec, |
| 97 | gc_level_t *gc_level) |
| 98 | { |
| 99 | struct logfs_segment_entry se; |
| 100 | u32 ec_level; |
| 101 | |
| 102 | logfs_get_segment_entry(sb, segno, &se); |
| 103 | if (se.ec_level == cpu_to_be32(BADSEG) || |
| 104 | se.valid == cpu_to_be32(RESERVED)) |
| 105 | return RESERVED; |
| 106 | |
| 107 | ec_level = be32_to_cpu(se.ec_level); |
| 108 | *ec = ec_level >> 4; |
| 109 | *gc_level = GC_LEVEL(ec_level & 0xf); |
| 110 | return be32_to_cpu(se.valid); |
| 111 | } |
| 112 | |
| 113 | static void logfs_cleanse_block(struct super_block *sb, u64 ofs, u64 ino, |
| 114 | u64 bix, gc_level_t gc_level) |
| 115 | { |
| 116 | struct inode *inode; |
| 117 | int err, cookie; |
| 118 | |
| 119 | inode = logfs_safe_iget(sb, ino, &cookie); |
| 120 | err = logfs_rewrite_block(inode, bix, ofs, gc_level, 0); |
| 121 | BUG_ON(err); |
| 122 | logfs_safe_iput(inode, cookie); |
| 123 | } |
| 124 | |
Joern Engel | 6f485b41 | 2010-05-07 19:38:40 +0200 | [diff] [blame] | 125 | static u32 logfs_gc_segment(struct super_block *sb, u32 segno) |
Joern Engel | 5db53f3 | 2009-11-20 20:13:39 +0100 | [diff] [blame] | 126 | { |
| 127 | struct logfs_super *super = logfs_super(sb); |
| 128 | struct logfs_segment_header sh; |
| 129 | struct logfs_object_header oh; |
| 130 | u64 ofs, ino, bix; |
| 131 | u32 seg_ofs, logical_segno, cleaned = 0; |
| 132 | int err, len, valid; |
| 133 | gc_level_t gc_level; |
| 134 | |
| 135 | LOGFS_BUG_ON(segment_is_reserved(sb, segno), sb); |
| 136 | |
| 137 | btree_insert32(&super->s_reserved_segments, segno, (void *)1, GFP_NOFS); |
| 138 | err = wbuf_read(sb, dev_ofs(sb, segno, 0), sizeof(sh), &sh); |
| 139 | BUG_ON(err); |
| 140 | gc_level = GC_LEVEL(sh.level); |
| 141 | logical_segno = be32_to_cpu(sh.segno); |
| 142 | if (sh.crc != logfs_crc32(&sh, sizeof(sh), 4)) { |
| 143 | logfs_mark_segment_bad(sb, segno); |
| 144 | cleaned = -1; |
| 145 | goto out; |
| 146 | } |
| 147 | |
| 148 | for (seg_ofs = LOGFS_SEGMENT_HEADERSIZE; |
| 149 | seg_ofs + sizeof(oh) < super->s_segsize; ) { |
| 150 | ofs = dev_ofs(sb, logical_segno, seg_ofs); |
| 151 | err = wbuf_read(sb, dev_ofs(sb, segno, seg_ofs), sizeof(oh), |
| 152 | &oh); |
| 153 | BUG_ON(err); |
| 154 | |
| 155 | if (!memchr_inv(&oh, 0xff, sizeof(oh))) |
| 156 | break; |
| 157 | |
| 158 | if (oh.crc != logfs_crc32(&oh, sizeof(oh) - 4, 4)) { |
| 159 | logfs_mark_segment_bad(sb, segno); |
| 160 | cleaned = super->s_segsize - 1; |
| 161 | goto out; |
| 162 | } |
| 163 | |
| 164 | ino = be64_to_cpu(oh.ino); |
| 165 | bix = be64_to_cpu(oh.bix); |
| 166 | len = sizeof(oh) + be16_to_cpu(oh.len); |
| 167 | valid = logfs_is_valid_block(sb, ofs, ino, bix, gc_level); |
| 168 | if (valid == 1) { |
| 169 | logfs_cleanse_block(sb, ofs, ino, bix, gc_level); |
| 170 | cleaned += len; |
| 171 | } else if (valid == 2) { |
| 172 | /* Will be invalid upon journal commit */ |
| 173 | cleaned += len; |
| 174 | } |
| 175 | seg_ofs += len; |
| 176 | } |
| 177 | out: |
| 178 | btree_remove32(&super->s_reserved_segments, segno); |
| 179 | return cleaned; |
| 180 | } |
| 181 | |
| 182 | static struct gc_candidate *add_list(struct gc_candidate *cand, |
| 183 | struct candidate_list *list) |
| 184 | { |
| 185 | struct rb_node **p = &list->rb_tree.rb_node; |
| 186 | struct rb_node *parent = NULL; |
| 187 | struct gc_candidate *cur; |
| 188 | int comp; |
| 189 | |
| 190 | cand->list = list; |
| 191 | while (*p) { |
| 192 | parent = *p; |
| 193 | cur = rb_entry(parent, struct gc_candidate, rb_node); |
| 194 | |
| 195 | if (list->sort_by_ec) |
| 196 | comp = cand->erase_count < cur->erase_count; |
| 197 | else |
| 198 | comp = cand->valid < cur->valid; |
| 199 | |
| 200 | if (comp) |
| 201 | p = &parent->rb_left; |
| 202 | else |
| 203 | p = &parent->rb_right; |
| 204 | } |
| 205 | rb_link_node(&cand->rb_node, parent, p); |
| 206 | rb_insert_color(&cand->rb_node, &list->rb_tree); |
| 207 | |
| 208 | if (list->count <= list->maxcount) { |
| 209 | list->count++; |
| 210 | return NULL; |
| 211 | } |
| 212 | cand = rb_entry(rb_last(&list->rb_tree), struct gc_candidate, rb_node); |
| 213 | rb_erase(&cand->rb_node, &list->rb_tree); |
| 214 | cand->list = NULL; |
| 215 | return cand; |
| 216 | } |
| 217 | |
| 218 | static void remove_from_list(struct gc_candidate *cand) |
| 219 | { |
| 220 | struct candidate_list *list = cand->list; |
| 221 | |
| 222 | rb_erase(&cand->rb_node, &list->rb_tree); |
| 223 | list->count--; |
| 224 | } |
| 225 | |
| 226 | static void free_candidate(struct super_block *sb, struct gc_candidate *cand) |
| 227 | { |
| 228 | struct logfs_super *super = logfs_super(sb); |
| 229 | |
| 230 | btree_remove32(&super->s_cand_tree, cand->segno); |
| 231 | kfree(cand); |
| 232 | } |
| 233 | |
| 234 | u32 get_best_cand(struct super_block *sb, struct candidate_list *list, u32 *ec) |
| 235 | { |
| 236 | struct gc_candidate *cand; |
| 237 | u32 segno; |
| 238 | |
| 239 | BUG_ON(list->count == 0); |
| 240 | |
| 241 | cand = rb_entry(rb_first(&list->rb_tree), struct gc_candidate, rb_node); |
| 242 | remove_from_list(cand); |
| 243 | segno = cand->segno; |
| 244 | if (ec) |
| 245 | *ec = cand->erase_count; |
| 246 | free_candidate(sb, cand); |
| 247 | return segno; |
| 248 | } |
| 249 | |
| 250 | /* |
| 251 | * We have several lists to manage segments with. The reserve_list is used to |
| 252 | * deal with bad blocks. We try to keep the best (lowest ec) segments on this |
| 253 | * list. |
| 254 | * The free_list contains free segments for normal usage. It usually gets the |
| 255 | * second pick after the reserve_list. But when the free_list is running short |
| 256 | * it is more important to keep the free_list full than to keep a reserve. |
| 257 | * |
| 258 | * Segments that are not free are put onto a per-level low_list. If we have |
| 259 | * to run garbage collection, we pick a candidate from there. All segments on |
| 260 | * those lists should have at least some free space so GC will make progress. |
| 261 | * |
| 262 | * And last we have the ec_list, which is used to pick segments for wear |
| 263 | * leveling. |
| 264 | * |
| 265 | * If all appropriate lists are full, we simply free the candidate and forget |
| 266 | * about that segment for a while. We have better candidates for each purpose. |
| 267 | */ |
| 268 | static void __add_candidate(struct super_block *sb, struct gc_candidate *cand) |
| 269 | { |
| 270 | struct logfs_super *super = logfs_super(sb); |
| 271 | u32 full = super->s_segsize - LOGFS_SEGMENT_RESERVE; |
| 272 | |
| 273 | if (cand->valid == 0) { |
| 274 | /* 100% free segments */ |
| 275 | log_gc_noisy("add reserve segment %x (ec %x) at %llx\n", |
| 276 | cand->segno, cand->erase_count, |
| 277 | dev_ofs(sb, cand->segno, 0)); |
| 278 | cand = add_list(cand, &super->s_reserve_list); |
| 279 | if (cand) { |
| 280 | log_gc_noisy("add free segment %x (ec %x) at %llx\n", |
| 281 | cand->segno, cand->erase_count, |
| 282 | dev_ofs(sb, cand->segno, 0)); |
| 283 | cand = add_list(cand, &super->s_free_list); |
| 284 | } |
| 285 | } else { |
| 286 | /* good candidates for Garbage Collection */ |
| 287 | if (cand->valid < full) |
| 288 | cand = add_list(cand, &super->s_low_list[cand->dist]); |
| 289 | /* good candidates for wear leveling, |
| 290 | * segments that were recently written get ignored */ |
| 291 | if (cand) |
| 292 | cand = add_list(cand, &super->s_ec_list); |
| 293 | } |
| 294 | if (cand) |
| 295 | free_candidate(sb, cand); |
| 296 | } |
| 297 | |
| 298 | static int add_candidate(struct super_block *sb, u32 segno, u32 valid, u32 ec, |
| 299 | u8 dist) |
| 300 | { |
| 301 | struct logfs_super *super = logfs_super(sb); |
| 302 | struct gc_candidate *cand; |
| 303 | |
| 304 | cand = kmalloc(sizeof(*cand), GFP_NOFS); |
| 305 | if (!cand) |
| 306 | return -ENOMEM; |
| 307 | |
| 308 | cand->segno = segno; |
| 309 | cand->valid = valid; |
| 310 | cand->erase_count = ec; |
| 311 | cand->dist = dist; |
| 312 | |
| 313 | btree_insert32(&super->s_cand_tree, segno, cand, GFP_NOFS); |
| 314 | __add_candidate(sb, cand); |
| 315 | return 0; |
| 316 | } |
| 317 | |
| 318 | static void remove_segment_from_lists(struct super_block *sb, u32 segno) |
| 319 | { |
| 320 | struct logfs_super *super = logfs_super(sb); |
| 321 | struct gc_candidate *cand; |
| 322 | |
| 323 | cand = btree_lookup32(&super->s_cand_tree, segno); |
| 324 | if (cand) { |
| 325 | remove_from_list(cand); |
| 326 | free_candidate(sb, cand); |
| 327 | } |
| 328 | } |
| 329 | |
| 330 | static void scan_segment(struct super_block *sb, u32 segno) |
| 331 | { |
| 332 | u32 valid, ec = 0; |
| 333 | gc_level_t gc_level = 0; |
| 334 | u8 dist; |
| 335 | |
| 336 | if (segment_is_reserved(sb, segno)) |
| 337 | return; |
| 338 | |
| 339 | remove_segment_from_lists(sb, segno); |
| 340 | valid = logfs_valid_bytes(sb, segno, &ec, &gc_level); |
| 341 | if (valid == RESERVED) |
| 342 | return; |
| 343 | |
| 344 | dist = root_distance(sb, gc_level); |
| 345 | add_candidate(sb, segno, valid, ec, dist); |
| 346 | } |
| 347 | |
| 348 | static struct gc_candidate *first_in_list(struct candidate_list *list) |
| 349 | { |
| 350 | if (list->count == 0) |
| 351 | return NULL; |
| 352 | return rb_entry(rb_first(&list->rb_tree), struct gc_candidate, rb_node); |
| 353 | } |
| 354 | |
| 355 | /* |
| 356 | * Find the best segment for garbage collection. Main criterion is |
| 357 | * the segment requiring the least effort to clean. Secondary |
| 358 | * criterion is to GC on the lowest level available. |
| 359 | * |
| 360 | * So we search the least effort segment on the lowest level first, |
| 361 | * then move up and pick another segment iff is requires significantly |
| 362 | * less effort. Hence the LOGFS_MAX_OBJECTSIZE in the comparison. |
| 363 | */ |
| 364 | static struct gc_candidate *get_candidate(struct super_block *sb) |
| 365 | { |
| 366 | struct logfs_super *super = logfs_super(sb); |
| 367 | int i, max_dist; |
| 368 | struct gc_candidate *cand = NULL, *this; |
| 369 | |
Joern Engel | 934eed3 | 2011-11-20 22:29:01 +0530 | [diff] [blame] | 370 | max_dist = min(no_free_segments(sb), LOGFS_NO_AREAS - 1); |
Joern Engel | 5db53f3 | 2009-11-20 20:13:39 +0100 | [diff] [blame] | 371 | |
| 372 | for (i = max_dist; i >= 0; i--) { |
| 373 | this = first_in_list(&super->s_low_list[i]); |
| 374 | if (!this) |
| 375 | continue; |
| 376 | if (!cand) |
| 377 | cand = this; |
| 378 | if (this->valid + LOGFS_MAX_OBJECTSIZE <= cand->valid) |
| 379 | cand = this; |
| 380 | } |
| 381 | return cand; |
| 382 | } |
| 383 | |
| 384 | static int __logfs_gc_once(struct super_block *sb, struct gc_candidate *cand) |
| 385 | { |
| 386 | struct logfs_super *super = logfs_super(sb); |
| 387 | gc_level_t gc_level; |
| 388 | u32 cleaned, valid, segno, ec; |
| 389 | u8 dist; |
| 390 | |
| 391 | if (!cand) { |
| 392 | log_gc("GC attempted, but no candidate found\n"); |
| 393 | return 0; |
| 394 | } |
| 395 | |
| 396 | segno = cand->segno; |
| 397 | dist = cand->dist; |
| 398 | valid = logfs_valid_bytes(sb, segno, &ec, &gc_level); |
| 399 | free_candidate(sb, cand); |
| 400 | log_gc("GC segment #%02x at %llx, %x required, %x free, %x valid, %llx free\n", |
| 401 | segno, (u64)segno << super->s_segshift, |
| 402 | dist, no_free_segments(sb), valid, |
| 403 | super->s_free_bytes); |
Joern Engel | 6f485b41 | 2010-05-07 19:38:40 +0200 | [diff] [blame] | 404 | cleaned = logfs_gc_segment(sb, segno); |
Joern Engel | 5db53f3 | 2009-11-20 20:13:39 +0100 | [diff] [blame] | 405 | log_gc("GC segment #%02x complete - now %x valid\n", segno, |
| 406 | valid - cleaned); |
| 407 | BUG_ON(cleaned != valid); |
| 408 | return 1; |
| 409 | } |
| 410 | |
| 411 | static int logfs_gc_once(struct super_block *sb) |
| 412 | { |
| 413 | struct gc_candidate *cand; |
| 414 | |
| 415 | cand = get_candidate(sb); |
| 416 | if (cand) |
| 417 | remove_from_list(cand); |
| 418 | return __logfs_gc_once(sb, cand); |
| 419 | } |
| 420 | |
| 421 | /* returns 1 if a wrap occurs, 0 otherwise */ |
| 422 | static int logfs_scan_some(struct super_block *sb) |
| 423 | { |
| 424 | struct logfs_super *super = logfs_super(sb); |
| 425 | u32 segno; |
| 426 | int i, ret = 0; |
| 427 | |
| 428 | segno = super->s_sweeper; |
| 429 | for (i = SCAN_RATIO; i > 0; i--) { |
| 430 | segno++; |
| 431 | if (segno >= super->s_no_segs) { |
| 432 | segno = 0; |
| 433 | ret = 1; |
| 434 | /* Break out of the loop. We want to read a single |
| 435 | * block from the segment size on next invocation if |
| 436 | * SCAN_RATIO is set to match block size |
| 437 | */ |
| 438 | break; |
| 439 | } |
| 440 | |
| 441 | scan_segment(sb, segno); |
| 442 | } |
| 443 | super->s_sweeper = segno; |
| 444 | return ret; |
| 445 | } |
| 446 | |
| 447 | /* |
| 448 | * In principle, this function should loop forever, looking for GC candidates |
| 449 | * and moving data. LogFS is designed in such a way that this loop is |
| 450 | * guaranteed to terminate. |
| 451 | * |
| 452 | * Limiting the loop to some iterations serves purely to catch cases when |
| 453 | * these guarantees have failed. An actual endless loop is an obvious bug |
| 454 | * and should be reported as such. |
| 455 | */ |
| 456 | static void __logfs_gc_pass(struct super_block *sb, int target) |
| 457 | { |
| 458 | struct logfs_super *super = logfs_super(sb); |
| 459 | struct logfs_block *block; |
| 460 | int round, progress, last_progress = 0; |
| 461 | |
Joern Engel | 032d8f7 | 2010-04-13 17:46:37 +0200 | [diff] [blame] | 462 | /* |
| 463 | * Doing too many changes to the segfile at once would result |
| 464 | * in a large number of aliases. Write the journal before |
| 465 | * things get out of hand. |
| 466 | */ |
| 467 | if (super->s_shadow_tree.no_shadowed_segments >= MAX_OBJ_ALIASES) |
| 468 | logfs_write_anchor(sb); |
| 469 | |
Joern Engel | 5db53f3 | 2009-11-20 20:13:39 +0100 | [diff] [blame] | 470 | if (no_free_segments(sb) >= target && |
| 471 | super->s_no_object_aliases < MAX_OBJ_ALIASES) |
| 472 | return; |
| 473 | |
| 474 | log_gc("__logfs_gc_pass(%x)\n", target); |
| 475 | for (round = 0; round < SCAN_ROUNDS; ) { |
| 476 | if (no_free_segments(sb) >= target) |
| 477 | goto write_alias; |
| 478 | |
| 479 | /* Sync in-memory state with on-medium state in case they |
| 480 | * diverged */ |
Joern Engel | c6d3830 | 2010-03-04 21:36:19 +0100 | [diff] [blame] | 481 | logfs_write_anchor(sb); |
Joern Engel | 5db53f3 | 2009-11-20 20:13:39 +0100 | [diff] [blame] | 482 | round += logfs_scan_some(sb); |
| 483 | if (no_free_segments(sb) >= target) |
| 484 | goto write_alias; |
| 485 | progress = logfs_gc_once(sb); |
| 486 | if (progress) |
| 487 | last_progress = round; |
| 488 | else if (round - last_progress > 2) |
| 489 | break; |
| 490 | continue; |
| 491 | |
| 492 | /* |
| 493 | * The goto logic is nasty, I just don't know a better way to |
| 494 | * code it. GC is supposed to ensure two things: |
| 495 | * 1. Enough free segments are available. |
| 496 | * 2. The number of aliases is bounded. |
| 497 | * When 1. is achieved, we take a look at 2. and write back |
| 498 | * some alias-containing blocks, if necessary. However, after |
| 499 | * each such write we need to go back to 1., as writes can |
| 500 | * consume free segments. |
| 501 | */ |
| 502 | write_alias: |
| 503 | if (super->s_no_object_aliases < MAX_OBJ_ALIASES) |
| 504 | return; |
| 505 | if (list_empty(&super->s_object_alias)) { |
| 506 | /* All aliases are still in btree */ |
| 507 | return; |
| 508 | } |
| 509 | log_gc("Write back one alias\n"); |
| 510 | block = list_entry(super->s_object_alias.next, |
| 511 | struct logfs_block, alias_list); |
| 512 | block->ops->write_block(block); |
| 513 | /* |
| 514 | * To round off the nasty goto logic, we reset round here. It |
| 515 | * is a safety-net for GC not making any progress and limited |
| 516 | * to something reasonably small. If incremented it for every |
| 517 | * single alias, the loop could terminate rather quickly. |
| 518 | */ |
| 519 | round = 0; |
| 520 | } |
| 521 | LOGFS_BUG(sb); |
| 522 | } |
| 523 | |
| 524 | static int wl_ratelimit(struct super_block *sb, u64 *next_event) |
| 525 | { |
| 526 | struct logfs_super *super = logfs_super(sb); |
| 527 | |
| 528 | if (*next_event < super->s_gec) { |
| 529 | *next_event = super->s_gec + WL_RATELIMIT; |
| 530 | return 0; |
| 531 | } |
| 532 | return 1; |
| 533 | } |
| 534 | |
| 535 | static void logfs_wl_pass(struct super_block *sb) |
| 536 | { |
| 537 | struct logfs_super *super = logfs_super(sb); |
| 538 | struct gc_candidate *wl_cand, *free_cand; |
| 539 | |
| 540 | if (wl_ratelimit(sb, &super->s_wl_gec_ostore)) |
| 541 | return; |
| 542 | |
| 543 | wl_cand = first_in_list(&super->s_ec_list); |
| 544 | if (!wl_cand) |
| 545 | return; |
| 546 | free_cand = first_in_list(&super->s_free_list); |
| 547 | if (!free_cand) |
| 548 | return; |
| 549 | |
| 550 | if (wl_cand->erase_count < free_cand->erase_count + WL_DELTA) { |
| 551 | remove_from_list(wl_cand); |
| 552 | __logfs_gc_once(sb, wl_cand); |
| 553 | } |
| 554 | } |
| 555 | |
| 556 | /* |
| 557 | * The journal needs wear leveling as well. But moving the journal is an |
| 558 | * expensive operation so we try to avoid it as much as possible. And if we |
| 559 | * have to do it, we move the whole journal, not individual segments. |
| 560 | * |
| 561 | * Ratelimiting is not strictly necessary here, it mainly serves to avoid the |
| 562 | * calculations. First we check whether moving the journal would be a |
| 563 | * significant improvement. That means that a) the current journal segments |
| 564 | * have more wear than the future journal segments and b) the current journal |
| 565 | * segments have more wear than normal ostore segments. |
| 566 | * Rationale for b) is that we don't have to move the journal if it is aging |
| 567 | * less than the ostore, even if the reserve segments age even less (they are |
| 568 | * excluded from wear leveling, after all). |
| 569 | * Next we check that the superblocks have less wear than the journal. Since |
| 570 | * moving the journal requires writing the superblocks, we have to protect the |
| 571 | * superblocks even more than the journal. |
| 572 | * |
| 573 | * Also we double the acceptable wear difference, compared to ostore wear |
| 574 | * leveling. Journal data is read and rewritten rapidly, comparatively. So |
| 575 | * soft errors have much less time to accumulate and we allow the journal to |
| 576 | * be a bit worse than the ostore. |
| 577 | */ |
| 578 | static void logfs_journal_wl_pass(struct super_block *sb) |
| 579 | { |
| 580 | struct logfs_super *super = logfs_super(sb); |
| 581 | struct gc_candidate *cand; |
| 582 | u32 min_journal_ec = -1, max_reserve_ec = 0; |
| 583 | int i; |
| 584 | |
| 585 | if (wl_ratelimit(sb, &super->s_wl_gec_journal)) |
| 586 | return; |
| 587 | |
| 588 | if (super->s_reserve_list.count < super->s_no_journal_segs) { |
| 589 | /* Reserve is not full enough to move complete journal */ |
| 590 | return; |
| 591 | } |
| 592 | |
| 593 | journal_for_each(i) |
| 594 | if (super->s_journal_seg[i]) |
| 595 | min_journal_ec = min(min_journal_ec, |
| 596 | super->s_journal_ec[i]); |
| 597 | cand = rb_entry(rb_first(&super->s_free_list.rb_tree), |
| 598 | struct gc_candidate, rb_node); |
| 599 | max_reserve_ec = cand->erase_count; |
| 600 | for (i = 0; i < 2; i++) { |
| 601 | struct logfs_segment_entry se; |
| 602 | u32 segno = seg_no(sb, super->s_sb_ofs[i]); |
| 603 | u32 ec; |
| 604 | |
| 605 | logfs_get_segment_entry(sb, segno, &se); |
| 606 | ec = be32_to_cpu(se.ec_level) >> 4; |
| 607 | max_reserve_ec = max(max_reserve_ec, ec); |
| 608 | } |
| 609 | |
| 610 | if (min_journal_ec > max_reserve_ec + 2 * WL_DELTA) { |
| 611 | do_logfs_journal_wl_pass(sb); |
| 612 | } |
| 613 | } |
| 614 | |
| 615 | void logfs_gc_pass(struct super_block *sb) |
| 616 | { |
| 617 | struct logfs_super *super = logfs_super(sb); |
| 618 | |
| 619 | //BUG_ON(mutex_trylock(&logfs_super(sb)->s_w_mutex)); |
| 620 | /* Write journal before free space is getting saturated with dirty |
| 621 | * objects. |
| 622 | */ |
| 623 | if (super->s_dirty_used_bytes + super->s_dirty_free_bytes |
| 624 | + LOGFS_MAX_OBJECTSIZE >= super->s_free_bytes) |
Joern Engel | c6d3830 | 2010-03-04 21:36:19 +0100 | [diff] [blame] | 625 | logfs_write_anchor(sb); |
| 626 | __logfs_gc_pass(sb, super->s_total_levels); |
Joern Engel | 5db53f3 | 2009-11-20 20:13:39 +0100 | [diff] [blame] | 627 | logfs_wl_pass(sb); |
| 628 | logfs_journal_wl_pass(sb); |
| 629 | } |
| 630 | |
| 631 | static int check_area(struct super_block *sb, int i) |
| 632 | { |
| 633 | struct logfs_super *super = logfs_super(sb); |
| 634 | struct logfs_area *area = super->s_area[i]; |
Joern Engel | 6f485b41 | 2010-05-07 19:38:40 +0200 | [diff] [blame] | 635 | gc_level_t gc_level; |
| 636 | u32 cleaned, valid, ec; |
Joern Engel | 5db53f3 | 2009-11-20 20:13:39 +0100 | [diff] [blame] | 637 | u32 segno = area->a_segno; |
Joern Engel | 6f485b41 | 2010-05-07 19:38:40 +0200 | [diff] [blame] | 638 | u64 ofs = dev_ofs(sb, area->a_segno, area->a_written_bytes); |
Joern Engel | 5db53f3 | 2009-11-20 20:13:39 +0100 | [diff] [blame] | 639 | |
| 640 | if (!area->a_is_open) |
| 641 | return 0; |
| 642 | |
Joern Engel | 6f485b41 | 2010-05-07 19:38:40 +0200 | [diff] [blame] | 643 | if (super->s_devops->can_write_buf(sb, ofs) == 0) |
| 644 | return 0; |
Joern Engel | 5db53f3 | 2009-11-20 20:13:39 +0100 | [diff] [blame] | 645 | |
Joern Engel | 6f485b41 | 2010-05-07 19:38:40 +0200 | [diff] [blame] | 646 | printk(KERN_INFO"LogFS: Possibly incomplete write at %llx\n", ofs); |
| 647 | /* |
| 648 | * The device cannot write back the write buffer. Most likely the |
| 649 | * wbuf was already written out and the system crashed at some point |
| 650 | * before the journal commit happened. In that case we wouldn't have |
| 651 | * to do anything. But if the crash happened before the wbuf was |
| 652 | * written out correctly, we must GC this segment. So assume the |
| 653 | * worst and always do the GC run. |
| 654 | */ |
| 655 | area->a_is_open = 0; |
| 656 | valid = logfs_valid_bytes(sb, segno, &ec, &gc_level); |
| 657 | cleaned = logfs_gc_segment(sb, segno); |
| 658 | if (cleaned != valid) |
| 659 | return -EIO; |
Joern Engel | 5db53f3 | 2009-11-20 20:13:39 +0100 | [diff] [blame] | 660 | return 0; |
| 661 | } |
| 662 | |
| 663 | int logfs_check_areas(struct super_block *sb) |
| 664 | { |
| 665 | int i, err; |
| 666 | |
| 667 | for_each_area(i) { |
| 668 | err = check_area(sb, i); |
| 669 | if (err) |
| 670 | return err; |
| 671 | } |
| 672 | return 0; |
| 673 | } |
| 674 | |
| 675 | static void logfs_init_candlist(struct candidate_list *list, int maxcount, |
| 676 | int sort_by_ec) |
| 677 | { |
| 678 | list->count = 0; |
| 679 | list->maxcount = maxcount; |
| 680 | list->sort_by_ec = sort_by_ec; |
| 681 | list->rb_tree = RB_ROOT; |
| 682 | } |
| 683 | |
| 684 | int logfs_init_gc(struct super_block *sb) |
| 685 | { |
| 686 | struct logfs_super *super = logfs_super(sb); |
| 687 | int i; |
| 688 | |
| 689 | btree_init_mempool32(&super->s_cand_tree, super->s_btree_pool); |
| 690 | logfs_init_candlist(&super->s_free_list, LIST_SIZE + SCAN_RATIO, 1); |
| 691 | logfs_init_candlist(&super->s_reserve_list, |
| 692 | super->s_bad_seg_reserve, 1); |
| 693 | for_each_area(i) |
| 694 | logfs_init_candlist(&super->s_low_list[i], LIST_SIZE, 0); |
| 695 | logfs_init_candlist(&super->s_ec_list, LIST_SIZE, 1); |
| 696 | return 0; |
| 697 | } |
| 698 | |
| 699 | static void logfs_cleanup_list(struct super_block *sb, |
| 700 | struct candidate_list *list) |
| 701 | { |
| 702 | struct gc_candidate *cand; |
| 703 | |
| 704 | while (list->count) { |
| 705 | cand = rb_entry(list->rb_tree.rb_node, struct gc_candidate, |
| 706 | rb_node); |
| 707 | remove_from_list(cand); |
| 708 | free_candidate(sb, cand); |
| 709 | } |
| 710 | BUG_ON(list->rb_tree.rb_node); |
| 711 | } |
| 712 | |
| 713 | void logfs_cleanup_gc(struct super_block *sb) |
| 714 | { |
| 715 | struct logfs_super *super = logfs_super(sb); |
| 716 | int i; |
| 717 | |
| 718 | if (!super->s_free_list.count) |
| 719 | return; |
| 720 | |
| 721 | /* |
| 722 | * FIXME: The btree may still contain a single empty node. So we |
| 723 | * call the grim visitor to clean up that mess. Btree code should |
| 724 | * do it for us, really. |
| 725 | */ |
| 726 | btree_grim_visitor32(&super->s_cand_tree, 0, NULL); |
| 727 | logfs_cleanup_list(sb, &super->s_free_list); |
| 728 | logfs_cleanup_list(sb, &super->s_reserve_list); |
| 729 | for_each_area(i) |
| 730 | logfs_cleanup_list(sb, &super->s_low_list[i]); |
| 731 | logfs_cleanup_list(sb, &super->s_ec_list); |
| 732 | } |