fs/bcachefs/btree_gc.c

   1 // SPDX-License-Identifier: GPL-2.0
   2 /*
   3  * Copyright (C) 2010 Kent Overstreet <kent.overstreet@gmail.com>
   4  * Copyright (C) 2014 Datera Inc.
   5  */
   6
   7 #include "bcachefs.h"
   8 #include "alloc_background.h"
   9 #include "alloc_foreground.h"
  10 #include "backpointers.h"
  11 #include "bkey_methods.h"
  12 #include "bkey_buf.h"
  13 #include "btree_journal_iter.h"
  14 #include "btree_key_cache.h"
  15 #include "btree_locking.h"
  16 #include "btree_node_scan.h"
  17 #include "btree_update_interior.h"
  18 #include "btree_io.h"
  19 #include "btree_gc.h"
  20 #include "buckets.h"
  21 #include "clock.h"
  22 #include "debug.h"
  23 #include "ec.h"
  24 #include "error.h"
  25 #include "extents.h"
  26 #include "journal.h"
  27 #include "keylist.h"
  28 #include "move.h"
  29 #include "recovery_passes.h"
  30 #include "reflink.h"
  31 #include "replicas.h"
  32 #include "super-io.h"
  33 #include "trace.h"
  34
  35 #include <linux/slab.h>
  36 #include <linux/bitops.h>
  37 #include <linux/freezer.h>
  38 #include <linux/kthread.h>
  39 #include <linux/preempt.h>
  40 #include <linux/rcupdate.h>
  41 #include <linux/sched/task.h>
  42
  43 #define DROP_THIS_NODE          10
  44 #define DROP_PREV_NODE          11
  45 #define DID_FILL_FROM_SCAN      12
  46
  47 static struct bkey_s unsafe_bkey_s_c_to_s(struct bkey_s_c k)
  48 {
  49         return (struct bkey_s) {{{
  50                 (struct bkey *) k.k,
  51                 (struct bch_val *) k.v
  52         }}};
  53 }
  54
  55 static bool should_restart_for_topology_repair(struct bch_fs *c)
  56 {
  57         return c->opts.fix_errors != FSCK_FIX_no &&
  58                 !(c->recovery_passes_complete & BIT_ULL(BCH_RECOVERY_PASS_check_topology));
  59 }
  60
  61 static inline void __gc_pos_set(struct bch_fs *c, struct gc_pos new_pos)
  62 {
  63         preempt_disable();
  64         write_seqcount_begin(&c->gc_pos_lock);
  65         c->gc_pos = new_pos;
  66         write_seqcount_end(&c->gc_pos_lock);
  67         preempt_enable();
  68 }
  69
  70 static inline void gc_pos_set(struct bch_fs *c, struct gc_pos new_pos)
  71 {
  72         BUG_ON(gc_pos_cmp(new_pos, c->gc_pos) <= 0);
  73         __gc_pos_set(c, new_pos);
  74 }
  75
  76 static void btree_ptr_to_v2(struct btree *b, struct bkey_i_btree_ptr_v2 *dst)
  77 {
  78         switch (b->key.k.type) {
  79         case KEY_TYPE_btree_ptr: {
  80                 struct bkey_i_btree_ptr *src = bkey_i_to_btree_ptr(&b->key);
  81
  82                 dst->k.p                = src->k.p;
  83                 dst->v.mem_ptr          = 0;
  84                 dst->v.seq              = b->data->keys.seq;
  85                 dst->v.sectors_written  = 0;
  86                 dst->v.flags            = 0;
  87                 dst->v.min_key          = b->data->min_key;
  88                 set_bkey_val_bytes(&dst->k, sizeof(dst->v) + bkey_val_bytes(&src->k));
  89                 memcpy(dst->v.start, src->v.start, bkey_val_bytes(&src->k));
  90                 break;
  91         }
  92         case KEY_TYPE_btree_ptr_v2:
  93                 bkey_copy(&dst->k_i, &b->key);
  94                 break;
  95         default:
  96                 BUG();
  97         }
  98 }
  99
 100 static void bch2_btree_node_update_key_early(struct btree_trans *trans,
 101                                              enum btree_id btree, unsigned level,
 102                                              struct bkey_s_c old, struct bkey_i *new)
 103 {
 104         struct bch_fs *c = trans->c;
 105         struct btree *b;
 106         struct bkey_buf tmp;
 107         int ret;
 108
 109         bch2_bkey_buf_init(&tmp);
 110         bch2_bkey_buf_reassemble(&tmp, c, old);
 111
 112         b = bch2_btree_node_get_noiter(trans, tmp.k, btree, level, true);
 113         if (!IS_ERR_OR_NULL(b)) {
 114                 mutex_lock(&c->btree_cache.lock);
 115
 116                 bch2_btree_node_hash_remove(&c->btree_cache, b);
 117
 118                 bkey_copy(&b->key, new);
 119                 ret = __bch2_btree_node_hash_insert(&c->btree_cache, b);
 120                 BUG_ON(ret);
 121
 122                 mutex_unlock(&c->btree_cache.lock);
 123                 six_unlock_read(&b->c.lock);
 124         }
 125
 126         bch2_bkey_buf_exit(&tmp, c);
 127 }
 128
 129 static int set_node_min(struct bch_fs *c, struct btree *b, struct bpos new_min)
 130 {
 131         struct bkey_i_btree_ptr_v2 *new;
 132         int ret;
 133
 134         if (c->opts.verbose) {
 135                 struct printbuf buf = PRINTBUF;
 136
 137                 bch2_bkey_val_to_text(&buf, c, bkey_i_to_s_c(&b->key));
 138                 prt_str(&buf, " -> ");
 139                 bch2_bpos_to_text(&buf, new_min);
 140
 141                 bch_info(c, "%s(): %s", __func__, buf.buf);
 142                 printbuf_exit(&buf);
 143         }
 144
 145         new = kmalloc_array(BKEY_BTREE_PTR_U64s_MAX, sizeof(u64), GFP_KERNEL);
 146         if (!new)
 147                 return -BCH_ERR_ENOMEM_gc_repair_key;
 148
 149         btree_ptr_to_v2(b, new);
 150         b->data->min_key        = new_min;
 151         new->v.min_key          = new_min;
 152         SET_BTREE_PTR_RANGE_UPDATED(&new->v, true);
 153
 154         ret = bch2_journal_key_insert_take(c, b->c.btree_id, b->c.level + 1, &new->k_i);
 155         if (ret) {
 156                 kfree(new);
 157                 return ret;
 158         }
 159
 160         bch2_btree_node_drop_keys_outside_node(b);
 161         bkey_copy(&b->key, &new->k_i);
 162         return 0;
 163 }
 164
 165 static int set_node_max(struct bch_fs *c, struct btree *b, struct bpos new_max)
 166 {
 167         struct bkey_i_btree_ptr_v2 *new;
 168         int ret;
 169
 170         if (c->opts.verbose) {
 171                 struct printbuf buf = PRINTBUF;
 172
 173                 bch2_bkey_val_to_text(&buf, c, bkey_i_to_s_c(&b->key));
 174                 prt_str(&buf, " -> ");
 175                 bch2_bpos_to_text(&buf, new_max);
 176
 177                 bch_info(c, "%s(): %s", __func__, buf.buf);
 178                 printbuf_exit(&buf);
 179         }
 180
 181         ret = bch2_journal_key_delete(c, b->c.btree_id, b->c.level + 1, b->key.k.p);
 182         if (ret)
 183                 return ret;
 184
 185         new = kmalloc_array(BKEY_BTREE_PTR_U64s_MAX, sizeof(u64), GFP_KERNEL);
 186         if (!new)
 187                 return -BCH_ERR_ENOMEM_gc_repair_key;
 188
 189         btree_ptr_to_v2(b, new);
 190         b->data->max_key        = new_max;
 191         new->k.p                = new_max;
 192         SET_BTREE_PTR_RANGE_UPDATED(&new->v, true);
 193
 194         ret = bch2_journal_key_insert_take(c, b->c.btree_id, b->c.level + 1, &new->k_i);
 195         if (ret) {
 196                 kfree(new);
 197                 return ret;
 198         }
 199
 200         bch2_btree_node_drop_keys_outside_node(b);
 201
 202         mutex_lock(&c->btree_cache.lock);
 203         bch2_btree_node_hash_remove(&c->btree_cache, b);
 204
 205         bkey_copy(&b->key, &new->k_i);
 206         ret = __bch2_btree_node_hash_insert(&c->btree_cache, b);
 207         BUG_ON(ret);
 208         mutex_unlock(&c->btree_cache.lock);
 209         return 0;
 210 }
 211
 212 static int btree_check_node_boundaries(struct bch_fs *c, struct btree *b,
 213                                        struct btree *prev, struct btree *cur,
 214                                        struct bpos *pulled_from_scan)
 215 {
 216         struct bpos expected_start = !prev
 217                 ? b->data->min_key
 218                 : bpos_successor(prev->key.k.p);
 219         struct printbuf buf = PRINTBUF;
 220         int ret = 0;
 221
 222         BUG_ON(b->key.k.type == KEY_TYPE_btree_ptr_v2 &&
 223                !bpos_eq(bkey_i_to_btree_ptr_v2(&b->key)->v.min_key,
 224                         b->data->min_key));
 225
 226         if (bpos_eq(expected_start, cur->data->min_key))
 227                 return 0;
 228
 229         prt_printf(&buf, "  at btree %s level %u:\n  parent: ",
 230                    bch2_btree_id_str(b->c.btree_id), b->c.level);
 231         bch2_bkey_val_to_text(&buf, c, bkey_i_to_s_c(&b->key));
 232
 233         if (prev) {
 234                 prt_printf(&buf, "\n  prev: ");
 235                 bch2_bkey_val_to_text(&buf, c, bkey_i_to_s_c(&prev->key));
 236         }
 237
 238         prt_str(&buf, "\n  next: ");
 239         bch2_bkey_val_to_text(&buf, c, bkey_i_to_s_c(&cur->key));
 240
 241         if (bpos_lt(expected_start, cur->data->min_key)) {                              /* gap */
 242                 if (b->c.level == 1 &&
 243                     bpos_lt(*pulled_from_scan, cur->data->min_key)) {
 244                         ret = bch2_get_scanned_nodes(c, b->c.btree_id, 0,
 245                                                      expected_start,
 246                                                      bpos_predecessor(cur->data->min_key));
 247                         if (ret)
 248                                 goto err;
 249
 250                         *pulled_from_scan = cur->data->min_key;
 251                         ret = DID_FILL_FROM_SCAN;
 252                 } else {
 253                         if (mustfix_fsck_err(c, btree_node_topology_bad_min_key,
 254                                              "btree node with incorrect min_key%s", buf.buf))
 255                                 ret = set_node_min(c, cur, expected_start);
 256                 }
 257         } else {                                                                        /* overlap */
 258                 if (prev && BTREE_NODE_SEQ(cur->data) > BTREE_NODE_SEQ(prev->data)) {   /* cur overwrites prev */
 259                         if (bpos_ge(prev->data->min_key, cur->data->min_key)) {         /* fully? */
 260                                 if (mustfix_fsck_err(c, btree_node_topology_overwritten_by_next_node,
 261                                                      "btree node overwritten by next node%s", buf.buf))
 262                                         ret = DROP_PREV_NODE;
 263                         } else {
 264                                 if (mustfix_fsck_err(c, btree_node_topology_bad_max_key,
 265                                                      "btree node with incorrect max_key%s", buf.buf))
 266                                         ret = set_node_max(c, prev,
 267                                                            bpos_predecessor(cur->data->min_key));
 268                         }
 269                 } else {
 270                         if (bpos_ge(expected_start, cur->data->max_key)) {              /* fully? */
 271                                 if (mustfix_fsck_err(c, btree_node_topology_overwritten_by_prev_node,
 272                                                      "btree node overwritten by prev node%s", buf.buf))
 273                                         ret = DROP_THIS_NODE;
 274                         } else {
 275                                 if (mustfix_fsck_err(c, btree_node_topology_bad_min_key,
 276                                                      "btree node with incorrect min_key%s", buf.buf))
 277                                         ret = set_node_min(c, cur, expected_start);
 278                         }
 279                 }
 280         }
 281 err:
 282 fsck_err:
 283         printbuf_exit(&buf);
 284         return ret;
 285 }
 286
 287 static int btree_repair_node_end(struct bch_fs *c, struct btree *b,
 288                                  struct btree *child, struct bpos *pulled_from_scan)
 289 {
 290         struct printbuf buf = PRINTBUF;
 291         int ret = 0;
 292
 293         if (bpos_eq(child->key.k.p, b->key.k.p))
 294                 return 0;
 295
 296         prt_printf(&buf, "at btree %s level %u:\n  parent: ",
 297                    bch2_btree_id_str(b->c.btree_id), b->c.level);
 298         bch2_bkey_val_to_text(&buf, c, bkey_i_to_s_c(&b->key));
 299
 300         prt_str(&buf, "\n  child: ");
 301         bch2_bkey_val_to_text(&buf, c, bkey_i_to_s_c(&child->key));
 302
 303         if (mustfix_fsck_err(c, btree_node_topology_bad_max_key,
 304                              "btree node with incorrect max_key%s", buf.buf)) {
 305                 if (b->c.level == 1 &&
 306                     bpos_lt(*pulled_from_scan, b->key.k.p)) {
 307                         ret = bch2_get_scanned_nodes(c, b->c.btree_id, 0,
 308                                                 bpos_successor(child->key.k.p), b->key.k.p);
 309                         if (ret)
 310                                 goto err;
 311
 312                         *pulled_from_scan = b->key.k.p;
 313                         ret = DID_FILL_FROM_SCAN;
 314                 } else {
 315                         ret = set_node_max(c, child, b->key.k.p);
 316                 }
 317         }
 318 err:
 319 fsck_err:
 320         printbuf_exit(&buf);
 321         return ret;
 322 }
 323
 324 static int bch2_btree_repair_topology_recurse(struct btree_trans *trans, struct btree *b,
 325                                               struct bpos *pulled_from_scan)
 326 {
 327         struct bch_fs *c = trans->c;
 328         struct btree_and_journal_iter iter;
 329         struct bkey_s_c k;
 330         struct bkey_buf prev_k, cur_k;
 331         struct btree *prev = NULL, *cur = NULL;
 332         bool have_child, new_pass = false;
 333         struct printbuf buf = PRINTBUF;
 334         int ret = 0;
 335
 336         if (!b->c.level)
 337                 return 0;
 338
 339         bch2_bkey_buf_init(&prev_k);
 340         bch2_bkey_buf_init(&cur_k);
 341 again:
 342         cur = prev = NULL;
 343         have_child = new_pass = false;
 344         bch2_btree_and_journal_iter_init_node_iter(trans, &iter, b);
 345         iter.prefetch = true;
 346
 347         while ((k = bch2_btree_and_journal_iter_peek(&iter)).k) {
 348                 BUG_ON(bpos_lt(k.k->p, b->data->min_key));
 349                 BUG_ON(bpos_gt(k.k->p, b->data->max_key));
 350
 351                 bch2_btree_and_journal_iter_advance(&iter);
 352                 bch2_bkey_buf_reassemble(&cur_k, c, k);
 353
 354                 cur = bch2_btree_node_get_noiter(trans, cur_k.k,
 355                                         b->c.btree_id, b->c.level - 1,
 356                                         false);
 357                 ret = PTR_ERR_OR_ZERO(cur);
 358
 359                 printbuf_reset(&buf);
 360                 bch2_bkey_val_to_text(&buf, c, bkey_i_to_s_c(cur_k.k));
 361
 362                 if (mustfix_fsck_err_on(bch2_err_matches(ret, EIO), c,
 363                                 btree_node_unreadable,
 364                                 "Topology repair: unreadable btree node at btree %s level %u:\n"
 365                                 "  %s",
 366                                 bch2_btree_id_str(b->c.btree_id),
 367                                 b->c.level - 1,
 368                                 buf.buf)) {
 369                         bch2_btree_node_evict(trans, cur_k.k);
 370                         cur = NULL;
 371                         ret = bch2_journal_key_delete(c, b->c.btree_id,
 372                                                       b->c.level, cur_k.k->k.p);
 373                         if (ret)
 374                                 break;
 375
 376                         if (!btree_id_is_alloc(b->c.btree_id)) {
 377                                 ret = bch2_run_explicit_recovery_pass(c, BCH_RECOVERY_PASS_scan_for_btree_nodes);
 378                                 if (ret)
 379                                         break;
 380                         }
 381                         continue;
 382                 }
 383
 384                 bch_err_msg(c, ret, "getting btree node");
 385                 if (ret)
 386                         break;
 387
 388                 if (bch2_btree_node_is_stale(c, cur)) {
 389                         bch_info(c, "btree node %s older than nodes found by scanning", buf.buf);
 390                         six_unlock_read(&cur->c.lock);
 391                         bch2_btree_node_evict(trans, cur_k.k);
 392                         ret = bch2_journal_key_delete(c, b->c.btree_id,
 393                                                       b->c.level, cur_k.k->k.p);
 394                         cur = NULL;
 395                         if (ret)
 396                                 break;
 397                         continue;
 398                 }
 399
 400                 ret = btree_check_node_boundaries(c, b, prev, cur, pulled_from_scan);
 401                 if (ret == DID_FILL_FROM_SCAN) {
 402                         new_pass = true;
 403                         ret = 0;
 404                 }
 405
 406                 if (ret == DROP_THIS_NODE) {
 407                         six_unlock_read(&cur->c.lock);
 408                         bch2_btree_node_evict(trans, cur_k.k);
 409                         ret = bch2_journal_key_delete(c, b->c.btree_id,
 410                                                       b->c.level, cur_k.k->k.p);
 411                         cur = NULL;
 412                         if (ret)
 413                                 break;
 414                         continue;
 415                 }
 416
 417                 if (prev)
 418                         six_unlock_read(&prev->c.lock);
 419                 prev = NULL;
 420
 421                 if (ret == DROP_PREV_NODE) {
 422                         bch_info(c, "dropped prev node");
 423                         bch2_btree_node_evict(trans, prev_k.k);
 424                         ret = bch2_journal_key_delete(c, b->c.btree_id,
 425                                                       b->c.level, prev_k.k->k.p);
 426                         if (ret)
 427                                 break;
 428
 429                         bch2_btree_and_journal_iter_exit(&iter);
 430                         goto again;
 431                 } else if (ret)
 432                         break;
 433
 434                 prev = cur;
 435                 cur = NULL;
 436                 bch2_bkey_buf_copy(&prev_k, c, cur_k.k);
 437         }
 438
 439         if (!ret && !IS_ERR_OR_NULL(prev)) {
 440                 BUG_ON(cur);
 441                 ret = btree_repair_node_end(c, b, prev, pulled_from_scan);
 442                 if (ret == DID_FILL_FROM_SCAN) {
 443                         new_pass = true;
 444                         ret = 0;
 445                 }
 446         }
 447
 448         if (!IS_ERR_OR_NULL(prev))
 449                 six_unlock_read(&prev->c.lock);
 450         prev = NULL;
 451         if (!IS_ERR_OR_NULL(cur))
 452                 six_unlock_read(&cur->c.lock);
 453         cur = NULL;
 454
 455         if (ret)
 456                 goto err;
 457
 458         bch2_btree_and_journal_iter_exit(&iter);
 459
 460         if (new_pass)
 461                 goto again;
 462
 463         bch2_btree_and_journal_iter_init_node_iter(trans, &iter, b);
 464         iter.prefetch = true;
 465
 466         while ((k = bch2_btree_and_journal_iter_peek(&iter)).k) {
 467                 bch2_bkey_buf_reassemble(&cur_k, c, k);
 468                 bch2_btree_and_journal_iter_advance(&iter);
 469
 470                 cur = bch2_btree_node_get_noiter(trans, cur_k.k,
 471                                         b->c.btree_id, b->c.level - 1,
 472                                         false);
 473                 ret = PTR_ERR_OR_ZERO(cur);
 474
 475                 bch_err_msg(c, ret, "getting btree node");
 476                 if (ret)
 477                         goto err;
 478
 479                 ret = bch2_btree_repair_topology_recurse(trans, cur, pulled_from_scan);
 480                 six_unlock_read(&cur->c.lock);
 481                 cur = NULL;
 482
 483                 if (ret == DROP_THIS_NODE) {
 484                         bch2_btree_node_evict(trans, cur_k.k);
 485                         ret = bch2_journal_key_delete(c, b->c.btree_id,
 486                                                       b->c.level, cur_k.k->k.p);
 487                         new_pass = true;
 488                 }
 489
 490                 if (ret)
 491                         goto err;
 492
 493                 have_child = true;
 494         }
 495
 496         printbuf_reset(&buf);
 497         bch2_bkey_val_to_text(&buf, c, bkey_i_to_s_c(&b->key));
 498
 499         if (mustfix_fsck_err_on(!have_child, c,
 500                         btree_node_topology_interior_node_empty,
 501                         "empty interior btree node at btree %s level %u\n"
 502                         "  %s",
 503                         bch2_btree_id_str(b->c.btree_id),
 504                         b->c.level, buf.buf))
 505                 ret = DROP_THIS_NODE;
 506 err:
 507 fsck_err:
 508         if (!IS_ERR_OR_NULL(prev))
 509                 six_unlock_read(&prev->c.lock);
 510         if (!IS_ERR_OR_NULL(cur))
 511                 six_unlock_read(&cur->c.lock);
 512
 513         bch2_btree_and_journal_iter_exit(&iter);
 514
 515         if (!ret && new_pass)
 516                 goto again;
 517
 518         BUG_ON(!ret && bch2_btree_node_check_topology(trans, b));
 519
 520         bch2_bkey_buf_exit(&prev_k, c);
 521         bch2_bkey_buf_exit(&cur_k, c);
 522         printbuf_exit(&buf);
 523         return ret;
 524 }
 525
 526 int bch2_check_topology(struct bch_fs *c)
 527 {
 528         struct btree_trans *trans = bch2_trans_get(c);
 529         struct bpos pulled_from_scan = POS_MIN;
 530         int ret = 0;
 531
 532         for (unsigned i = 0; i < btree_id_nr_alive(c) && !ret; i++) {
 533                 struct btree_root *r = bch2_btree_id_root(c, i);
 534                 bool reconstructed_root = false;
 535
 536                 if (r->error) {
 537                         ret = bch2_run_explicit_recovery_pass(c, BCH_RECOVERY_PASS_scan_for_btree_nodes);
 538                         if (ret)
 539                                 break;
 540 reconstruct_root:
 541                         bch_info(c, "btree root %s unreadable, must recover from scan", bch2_btree_id_str(i));
 542
 543                         r->alive = false;
 544                         r->error = 0;
 545
 546                         if (!bch2_btree_has_scanned_nodes(c, i)) {
 547                                 mustfix_fsck_err(c, btree_root_unreadable_and_scan_found_nothing,
 548                                                  "no nodes found for btree %s, continue?", bch2_btree_id_str(i));
 549                                 bch2_btree_root_alloc_fake(c, i, 0);
 550                         } else {
 551                                 bch2_btree_root_alloc_fake(c, i, 1);
 552                                 bch2_shoot_down_journal_keys(c, i, 1, BTREE_MAX_DEPTH, POS_MIN, SPOS_MAX);
 553                                 ret = bch2_get_scanned_nodes(c, i, 0, POS_MIN, SPOS_MAX);
 554                                 if (ret)
 555                                         break;
 556                         }
 557
 558                         reconstructed_root = true;
 559                 }
 560
 561                 struct btree *b = r->b;
 562
 563                 btree_node_lock_nopath_nofail(trans, &b->c, SIX_LOCK_read);
 564                 ret = bch2_btree_repair_topology_recurse(trans, b, &pulled_from_scan);
 565                 six_unlock_read(&b->c.lock);
 566
 567                 if (ret == DROP_THIS_NODE) {
 568                         bch2_btree_node_hash_remove(&c->btree_cache, b);
 569                         mutex_lock(&c->btree_cache.lock);
 570                         list_move(&b->list, &c->btree_cache.freeable);
 571                         mutex_unlock(&c->btree_cache.lock);
 572
 573                         r->b = NULL;
 574
 575                         if (!reconstructed_root)
 576                                 goto reconstruct_root;
 577
 578                         bch_err(c, "empty btree root %s", bch2_btree_id_str(i));
 579                         bch2_btree_root_alloc_fake(c, i, 0);
 580                         r->alive = false;
 581                         ret = 0;
 582                 }
 583         }
 584 fsck_err:
 585         bch2_trans_put(trans);
 586         return ret;
 587 }
 588
 589 static int bch2_check_fix_ptrs(struct btree_trans *trans, enum btree_id btree_id,
 590                                unsigned level, bool is_root,
 591                                struct bkey_s_c *k)
 592 {
 593         struct bch_fs *c = trans->c;
 594         struct bkey_ptrs_c ptrs_c = bch2_bkey_ptrs_c(*k);
 595         const union bch_extent_entry *entry_c;
 596         struct extent_ptr_decoded p = { 0 };
 597         bool do_update = false;
 598         struct printbuf buf = PRINTBUF;
 599         int ret = 0;
 600
 601         /*
 602          * XXX
 603          * use check_bucket_ref here
 604          */
 605         bkey_for_each_ptr_decode(k->k, ptrs_c, p, entry_c) {
 606                 struct bch_dev *ca = bch_dev_bkey_exists(c, p.ptr.dev);
 607                 struct bucket *g = PTR_GC_BUCKET(ca, &p.ptr);
 608                 enum bch_data_type data_type = bch2_bkey_ptr_data_type(*k, p, entry_c);
 609
 610                 if (fsck_err_on(!g->gen_valid,
 611                                 c, ptr_to_missing_alloc_key,
 612                                 "bucket %u:%zu data type %s ptr gen %u missing in alloc btree\n"
 613                                 "while marking %s",
 614                                 p.ptr.dev, PTR_BUCKET_NR(ca, &p.ptr),
 615                                 bch2_data_type_str(ptr_data_type(k->k, &p.ptr)),
 616                                 p.ptr.gen,
 617                                 (printbuf_reset(&buf),
 618                                  bch2_bkey_val_to_text(&buf, c, *k), buf.buf))) {
 619                         if (!p.ptr.cached) {
 620                                 g->gen_valid            = true;
 621                                 g->gen                  = p.ptr.gen;
 622                         } else {
 623                                 do_update = true;
 624                         }
 625                 }
 626
 627                 if (fsck_err_on(gen_cmp(p.ptr.gen, g->gen) > 0,
 628                                 c, ptr_gen_newer_than_bucket_gen,
 629                                 "bucket %u:%zu data type %s ptr gen in the future: %u > %u\n"
 630                                 "while marking %s",
 631                                 p.ptr.dev, PTR_BUCKET_NR(ca, &p.ptr),
 632                                 bch2_data_type_str(ptr_data_type(k->k, &p.ptr)),
 633                                 p.ptr.gen, g->gen,
 634                                 (printbuf_reset(&buf),
 635                                  bch2_bkey_val_to_text(&buf, c, *k), buf.buf))) {
 636                         if (!p.ptr.cached) {
 637                                 g->gen_valid            = true;
 638                                 g->gen                  = p.ptr.gen;
 639                                 g->data_type            = 0;
 640                                 g->dirty_sectors        = 0;
 641                                 g->cached_sectors       = 0;
 642                                 set_bit(BCH_FS_need_another_gc, &c->flags);
 643                         } else {
 644                                 do_update = true;
 645                         }
 646                 }
 647
 648                 if (fsck_err_on(gen_cmp(g->gen, p.ptr.gen) > BUCKET_GC_GEN_MAX,
 649                                 c, ptr_gen_newer_than_bucket_gen,
 650                                 "bucket %u:%zu gen %u data type %s: ptr gen %u too stale\n"
 651                                 "while marking %s",
 652                                 p.ptr.dev, PTR_BUCKET_NR(ca, &p.ptr), g->gen,
 653                                 bch2_data_type_str(ptr_data_type(k->k, &p.ptr)),
 654                                 p.ptr.gen,
 655                                 (printbuf_reset(&buf),
 656                                  bch2_bkey_val_to_text(&buf, c, *k), buf.buf)))
 657                         do_update = true;
 658
 659                 if (fsck_err_on(!p.ptr.cached && gen_cmp(p.ptr.gen, g->gen) < 0,
 660                                 c, stale_dirty_ptr,
 661                                 "bucket %u:%zu data type %s stale dirty ptr: %u < %u\n"
 662                                 "while marking %s",
 663                                 p.ptr.dev, PTR_BUCKET_NR(ca, &p.ptr),
 664                                 bch2_data_type_str(ptr_data_type(k->k, &p.ptr)),
 665                                 p.ptr.gen, g->gen,
 666                                 (printbuf_reset(&buf),
 667                                  bch2_bkey_val_to_text(&buf, c, *k), buf.buf)))
 668                         do_update = true;
 669
 670                 if (data_type != BCH_DATA_btree && p.ptr.gen != g->gen)
 671                         continue;
 672
 673                 if (fsck_err_on(bucket_data_type(g->data_type) &&
 674                                 bucket_data_type(g->data_type) !=
 675                                 bucket_data_type(data_type), c,
 676                                 ptr_bucket_data_type_mismatch,
 677                                 "bucket %u:%zu different types of data in same bucket: %s, %s\n"
 678                                 "while marking %s",
 679                                 p.ptr.dev, PTR_BUCKET_NR(ca, &p.ptr),
 680                                 bch2_data_type_str(g->data_type),
 681                                 bch2_data_type_str(data_type),
 682                                 (printbuf_reset(&buf),
 683                                  bch2_bkey_val_to_text(&buf, c, *k), buf.buf))) {
 684                         if (data_type == BCH_DATA_btree) {
 685                                 g->data_type    = data_type;
 686                                 set_bit(BCH_FS_need_another_gc, &c->flags);
 687                         } else {
 688                                 do_update = true;
 689                         }
 690                 }
 691
 692                 if (p.has_ec) {
 693                         struct gc_stripe *m = genradix_ptr(&c->gc_stripes, p.ec.idx);
 694
 695                         if (fsck_err_on(!m || !m->alive, c,
 696                                         ptr_to_missing_stripe,
 697                                         "pointer to nonexistent stripe %llu\n"
 698                                         "while marking %s",
 699                                         (u64) p.ec.idx,
 700                                         (printbuf_reset(&buf),
 701                                          bch2_bkey_val_to_text(&buf, c, *k), buf.buf)))
 702                                 do_update = true;
 703
 704                         if (fsck_err_on(m && m->alive && !bch2_ptr_matches_stripe_m(m, p), c,
 705                                         ptr_to_incorrect_stripe,
 706                                         "pointer does not match stripe %llu\n"
 707                                         "while marking %s",
 708                                         (u64) p.ec.idx,
 709                                         (printbuf_reset(&buf),
 710                                          bch2_bkey_val_to_text(&buf, c, *k), buf.buf)))
 711                                 do_update = true;
 712                 }
 713         }
 714
 715         if (do_update) {
 716                 if (is_root) {
 717                         bch_err(c, "cannot update btree roots yet");
 718                         ret = -EINVAL;
 719                         goto err;
 720                 }
 721
 722                 struct bkey_i *new = kmalloc(bkey_bytes(k->k), GFP_KERNEL);
 723                 if (!new) {
 724                         ret = -BCH_ERR_ENOMEM_gc_repair_key;
 725                         bch_err_msg(c, ret, "allocating new key");
 726                         goto err;
 727                 }
 728
 729                 bkey_reassemble(new, *k);
 730
 731                 if (level) {
 732                         /*
 733                          * We don't want to drop btree node pointers - if the
 734                          * btree node isn't there anymore, the read path will
 735                          * sort it out:
 736                          */
 737                         struct bkey_ptrs ptrs = bch2_bkey_ptrs(bkey_i_to_s(new));
 738                         bkey_for_each_ptr(ptrs, ptr) {
 739                                 struct bch_dev *ca = bch_dev_bkey_exists(c, ptr->dev);
 740                                 struct bucket *g = PTR_GC_BUCKET(ca, ptr);
 741
 742                                 ptr->gen = g->gen;
 743                         }
 744                 } else {
 745                         struct bkey_ptrs ptrs;
 746                         union bch_extent_entry *entry;
 747 restart_drop_ptrs:
 748                         ptrs = bch2_bkey_ptrs(bkey_i_to_s(new));
 749                         bkey_for_each_ptr_decode(bkey_i_to_s(new).k, ptrs, p, entry) {
 750                                 struct bch_dev *ca = bch_dev_bkey_exists(c, p.ptr.dev);
 751                                 struct bucket *g = PTR_GC_BUCKET(ca, &p.ptr);
 752                                 enum bch_data_type data_type = bch2_bkey_ptr_data_type(bkey_i_to_s_c(new), p, entry);
 753
 754                                 if ((p.ptr.cached &&
 755                                      (!g->gen_valid || gen_cmp(p.ptr.gen, g->gen) > 0)) ||
 756                                     (!p.ptr.cached &&
 757                                      gen_cmp(p.ptr.gen, g->gen) < 0) ||
 758                                     gen_cmp(g->gen, p.ptr.gen) > BUCKET_GC_GEN_MAX ||
 759                                     (g->data_type &&
 760                                      g->data_type != data_type)) {
 761                                         bch2_bkey_drop_ptr(bkey_i_to_s(new), &entry->ptr);
 762                                         goto restart_drop_ptrs;
 763                                 }
 764                         }
 765 again:
 766                         ptrs = bch2_bkey_ptrs(bkey_i_to_s(new));
 767                         bkey_extent_entry_for_each(ptrs, entry) {
 768                                 if (extent_entry_type(entry) == BCH_EXTENT_ENTRY_stripe_ptr) {
 769                                         struct gc_stripe *m = genradix_ptr(&c->gc_stripes,
 770                                                                         entry->stripe_ptr.idx);
 771                                         union bch_extent_entry *next_ptr;
 772
 773                                         bkey_extent_entry_for_each_from(ptrs, next_ptr, entry)
 774                                                 if (extent_entry_type(next_ptr) == BCH_EXTENT_ENTRY_ptr)
 775                                                         goto found;
 776                                         next_ptr = NULL;
 777 found:
 778                                         if (!next_ptr) {
 779                                                 bch_err(c, "aieee, found stripe ptr with no data ptr");
 780                                                 continue;
 781                                         }
 782
 783                                         if (!m || !m->alive ||
 784                                             !__bch2_ptr_matches_stripe(&m->ptrs[entry->stripe_ptr.block],
 785                                                                        &next_ptr->ptr,
 786                                                                        m->sectors)) {
 787                                                 bch2_bkey_extent_entry_drop(new, entry);
 788                                                 goto again;
 789                                         }
 790                                 }
 791                         }
 792                 }
 793
 794                 if (level)
 795                         bch2_btree_node_update_key_early(trans, btree_id, level - 1, *k, new);
 796
 797                 if (0) {
 798                         printbuf_reset(&buf);
 799                         bch2_bkey_val_to_text(&buf, c, *k);
 800                         bch_info(c, "updated %s", buf.buf);
 801
 802                         printbuf_reset(&buf);
 803                         bch2_bkey_val_to_text(&buf, c, bkey_i_to_s_c(new));
 804                         bch_info(c, "new key %s", buf.buf);
 805                 }
 806
 807                 ret = bch2_journal_key_insert_take(c, btree_id, level, new);
 808                 if (ret) {
 809                         kfree(new);
 810                         goto err;
 811                 }
 812
 813                 *k = bkey_i_to_s_c(new);
 814         }
 815 err:
 816 fsck_err:
 817         printbuf_exit(&buf);
 818         return ret;
 819 }
 820
 821 /* marking of btree keys/nodes: */
 822
 823 static int bch2_gc_mark_key(struct btree_trans *trans, enum btree_id btree_id,
 824                             unsigned level, bool is_root,
 825                             struct bkey_s_c *k,
 826                             bool initial)
 827 {
 828         struct bch_fs *c = trans->c;
 829         struct bkey deleted = KEY(0, 0, 0);
 830         struct bkey_s_c old = (struct bkey_s_c) { &deleted, NULL };
 831         struct printbuf buf = PRINTBUF;
 832         int ret = 0;
 833
 834         deleted.p = k->k->p;
 835
 836         if (initial) {
 837                 BUG_ON(bch2_journal_seq_verify &&
 838                        k->k->version.lo > atomic64_read(&c->journal.seq));
 839
 840                 if (fsck_err_on(k->k->version.lo > atomic64_read(&c->key_version), c,
 841                                 bkey_version_in_future,
 842                                 "key version number higher than recorded: %llu > %llu",
 843                                 k->k->version.lo,
 844                                 atomic64_read(&c->key_version)))
 845                         atomic64_set(&c->key_version, k->k->version.lo);
 846         }
 847
 848         ret = bch2_check_fix_ptrs(trans, btree_id, level, is_root, k);
 849         if (ret)
 850                 goto err;
 851
 852         if (mustfix_fsck_err_on(level && !bch2_dev_btree_bitmap_marked(c, *k),
 853                                 c, btree_bitmap_not_marked,
 854                                 "btree ptr not marked in member info btree allocated bitmap\n  %s",
 855                                 (bch2_bkey_val_to_text(&buf, c, *k),
 856                                  buf.buf))) {
 857                 mutex_lock(&c->sb_lock);
 858                 bch2_dev_btree_bitmap_mark(c, *k);
 859                 bch2_write_super(c);
 860                 mutex_unlock(&c->sb_lock);
 861         }
 862
 863         ret = commit_do(trans, NULL, NULL, 0,
 864                         bch2_key_trigger(trans, btree_id, level, old,
 865                                          unsafe_bkey_s_c_to_s(*k), BTREE_TRIGGER_GC));
 866 fsck_err:
 867 err:
 868         printbuf_exit(&buf);
 869         bch_err_fn(c, ret);
 870         return ret;
 871 }
 872
 873 static int btree_gc_mark_node(struct btree_trans *trans, struct btree *b, bool initial)
 874 {
 875         struct btree_node_iter iter;
 876         struct bkey unpacked;
 877         struct bkey_s_c k;
 878         int ret = 0;
 879
 880         ret = bch2_btree_node_check_topology(trans, b);
 881         if (ret)
 882                 return ret;
 883
 884         if (!btree_node_type_needs_gc(btree_node_type(b)))
 885                 return 0;
 886
 887         bch2_btree_node_iter_init_from_start(&iter, b);
 888
 889         while ((k = bch2_btree_node_iter_peek_unpack(&iter, b, &unpacked)).k) {
 890                 ret = bch2_gc_mark_key(trans, b->c.btree_id, b->c.level, false,
 891                                        &k, initial);
 892                 if (ret)
 893                         return ret;
 894
 895                 bch2_btree_node_iter_advance(&iter, b);
 896         }
 897
 898         return 0;
 899 }
 900
 901 static int bch2_gc_btree(struct btree_trans *trans, enum btree_id btree_id,
 902                          bool initial, bool metadata_only)
 903 {
 904         struct bch_fs *c = trans->c;
 905         struct btree_iter iter;
 906         struct btree *b;
 907         unsigned depth = metadata_only ? 1 : 0;
 908         int ret = 0;
 909
 910         gc_pos_set(c, gc_pos_btree(btree_id, POS_MIN, 0));
 911
 912         __for_each_btree_node(trans, iter, btree_id, POS_MIN,
 913                               0, depth, BTREE_ITER_PREFETCH, b, ret) {
 914                 bch2_verify_btree_nr_keys(b);
 915
 916                 gc_pos_set(c, gc_pos_btree_node(b));
 917
 918                 ret = btree_gc_mark_node(trans, b, initial);
 919                 if (ret)
 920                         break;
 921         }
 922         bch2_trans_iter_exit(trans, &iter);
 923
 924         if (ret)
 925                 return ret;
 926
 927         mutex_lock(&c->btree_root_lock);
 928         b = bch2_btree_id_root(c, btree_id)->b;
 929         if (!btree_node_fake(b)) {
 930                 struct bkey_s_c k = bkey_i_to_s_c(&b->key);
 931
 932                 ret = bch2_gc_mark_key(trans, b->c.btree_id, b->c.level + 1,
 933                                        true, &k, initial);
 934         }
 935         gc_pos_set(c, gc_pos_btree_root(b->c.btree_id));
 936         mutex_unlock(&c->btree_root_lock);
 937
 938         return ret;
 939 }
 940
 941 static int bch2_gc_btree_init_recurse(struct btree_trans *trans, struct btree *b,
 942                                       unsigned target_depth)
 943 {
 944         struct bch_fs *c = trans->c;
 945         struct btree_and_journal_iter iter;
 946         struct bkey_s_c k;
 947         struct bkey_buf cur;
 948         struct printbuf buf = PRINTBUF;
 949         int ret = 0;
 950
 951         ret = bch2_btree_node_check_topology(trans, b);
 952         if (ret)
 953                 return ret;
 954
 955         bch2_btree_and_journal_iter_init_node_iter(trans, &iter, b);
 956         bch2_bkey_buf_init(&cur);
 957
 958         while ((k = bch2_btree_and_journal_iter_peek(&iter)).k) {
 959                 BUG_ON(bpos_lt(k.k->p, b->data->min_key));
 960                 BUG_ON(bpos_gt(k.k->p, b->data->max_key));
 961
 962                 ret = bch2_gc_mark_key(trans, b->c.btree_id, b->c.level,
 963                                        false, &k, true);
 964                 if (ret)
 965                         goto fsck_err;
 966
 967                 bch2_btree_and_journal_iter_advance(&iter);
 968         }
 969
 970         if (b->c.level > target_depth) {
 971                 bch2_btree_and_journal_iter_exit(&iter);
 972                 bch2_btree_and_journal_iter_init_node_iter(trans, &iter, b);
 973                 iter.prefetch = true;
 974
 975                 while ((k = bch2_btree_and_journal_iter_peek(&iter)).k) {
 976                         struct btree *child;
 977
 978                         bch2_bkey_buf_reassemble(&cur, c, k);
 979                         bch2_btree_and_journal_iter_advance(&iter);
 980
 981                         child = bch2_btree_node_get_noiter(trans, cur.k,
 982                                                 b->c.btree_id, b->c.level - 1,
 983                                                 false);
 984                         ret = PTR_ERR_OR_ZERO(child);
 985
 986                         if (bch2_err_matches(ret, EIO)) {
 987                                 bch2_topology_error(c);
 988
 989                                 if (__fsck_err(c,
 990                                           FSCK_CAN_FIX|
 991                                           FSCK_CAN_IGNORE|
 992                                           FSCK_NO_RATELIMIT,
 993                                           btree_node_read_error,
 994                                           "Unreadable btree node at btree %s level %u:\n"
 995                                           "  %s",
 996                                           bch2_btree_id_str(b->c.btree_id),
 997                                           b->c.level - 1,
 998                                           (printbuf_reset(&buf),
 999                                            bch2_bkey_val_to_text(&buf, c, bkey_i_to_s_c(cur.k)), buf.buf)) &&
1000                                     should_restart_for_topology_repair(c)) {
1001                                         bch_info(c, "Halting mark and sweep to start topology repair pass");
1002                                         ret = bch2_run_explicit_recovery_pass(c, BCH_RECOVERY_PASS_check_topology);
1003                                         goto fsck_err;
1004                                 } else {
1005                                         /* Continue marking when opted to not
1006                                          * fix the error: */
1007                                         ret = 0;
1008                                         set_bit(BCH_FS_initial_gc_unfixed, &c->flags);
1009                                         continue;
1010                                 }
1011                         } else if (ret) {
1012                                 bch_err_msg(c, ret, "getting btree node");
1013                                 break;
1014                         }
1015
1016                         ret = bch2_gc_btree_init_recurse(trans, child,
1017                                                          target_depth);
1018                         six_unlock_read(&child->c.lock);
1019
1020                         if (ret)
1021                                 break;
1022                 }
1023         }
1024 fsck_err:
1025         bch2_bkey_buf_exit(&cur, c);
1026         bch2_btree_and_journal_iter_exit(&iter);
1027         printbuf_exit(&buf);
1028         return ret;
1029 }
1030
1031 static int bch2_gc_btree_init(struct btree_trans *trans,
1032                               enum btree_id btree_id,
1033                               bool metadata_only)
1034 {
1035         struct bch_fs *c = trans->c;
1036         struct btree *b;
1037         unsigned target_depth = metadata_only ? 1 : 0;
1038         struct printbuf buf = PRINTBUF;
1039         int ret = 0;
1040
1041         b = bch2_btree_id_root(c, btree_id)->b;
1042
1043         six_lock_read(&b->c.lock, NULL, NULL);
1044         printbuf_reset(&buf);
1045         bch2_bpos_to_text(&buf, b->data->min_key);
1046         if (mustfix_fsck_err_on(!bpos_eq(b->data->min_key, POS_MIN), c,
1047                                 btree_root_bad_min_key,
1048                         "btree root with incorrect min_key: %s", buf.buf)) {
1049                 bch_err(c, "repair unimplemented");
1050                 ret = -BCH_ERR_fsck_repair_unimplemented;
1051                 goto fsck_err;
1052         }
1053
1054         printbuf_reset(&buf);
1055         bch2_bpos_to_text(&buf, b->data->max_key);
1056         if (mustfix_fsck_err_on(!bpos_eq(b->data->max_key, SPOS_MAX), c,
1057                                 btree_root_bad_max_key,
1058                         "btree root with incorrect max_key: %s", buf.buf)) {
1059                 bch_err(c, "repair unimplemented");
1060                 ret = -BCH_ERR_fsck_repair_unimplemented;
1061                 goto fsck_err;
1062         }
1063
1064         if (b->c.level >= target_depth)
1065                 ret = bch2_gc_btree_init_recurse(trans, b, target_depth);
1066
1067         if (!ret) {
1068                 struct bkey_s_c k = bkey_i_to_s_c(&b->key);
1069
1070                 ret = bch2_gc_mark_key(trans, b->c.btree_id, b->c.level + 1, true,
1071                                        &k, true);
1072         }
1073 fsck_err:
1074         six_unlock_read(&b->c.lock);
1075
1076         bch_err_fn(c, ret);
1077         printbuf_exit(&buf);
1078         return ret;
1079 }
1080
1081 static inline int btree_id_gc_phase_cmp(enum btree_id l, enum btree_id r)
1082 {
1083         return  (int) btree_id_to_gc_phase(l) -
1084                 (int) btree_id_to_gc_phase(r);
1085 }
1086
1087 static int bch2_gc_btrees(struct bch_fs *c, bool initial, bool metadata_only)
1088 {
1089         struct btree_trans *trans = bch2_trans_get(c);
1090         enum btree_id ids[BTREE_ID_NR];
1091         unsigned i;
1092         int ret = 0;
1093
1094         for (i = 0; i < BTREE_ID_NR; i++)
1095                 ids[i] = i;
1096         bubble_sort(ids, BTREE_ID_NR, btree_id_gc_phase_cmp);
1097
1098         for (i = 0; i < BTREE_ID_NR && !ret; i++)
1099                 ret = initial
1100                         ? bch2_gc_btree_init(trans, ids[i], metadata_only)
1101                         : bch2_gc_btree(trans, ids[i], initial, metadata_only);
1102
1103         for (i = BTREE_ID_NR; i < btree_id_nr_alive(c) && !ret; i++) {
1104                 if (!bch2_btree_id_root(c, i)->alive)
1105                         continue;
1106
1107                 ret = initial
1108                         ? bch2_gc_btree_init(trans, i, metadata_only)
1109                         : bch2_gc_btree(trans, i, initial, metadata_only);
1110         }
1111
1112         bch2_trans_put(trans);
1113         bch_err_fn(c, ret);
1114         return ret;
1115 }
1116
1117 static void mark_metadata_sectors(struct bch_fs *c, struct bch_dev *ca,
1118                                   u64 start, u64 end,
1119                                   enum bch_data_type type,
1120                                   unsigned flags)
1121 {
1122         u64 b = sector_to_bucket(ca, start);
1123
1124         do {
1125                 unsigned sectors =
1126                         min_t(u64, bucket_to_sector(ca, b + 1), end) - start;
1127
1128                 bch2_mark_metadata_bucket(c, ca, b, type, sectors,
1129                                           gc_phase(GC_PHASE_SB), flags);
1130                 b++;
1131                 start += sectors;
1132         } while (start < end);
1133 }
1134
1135 static void bch2_mark_dev_superblock(struct bch_fs *c, struct bch_dev *ca,
1136                                      unsigned flags)
1137 {
1138         struct bch_sb_layout *layout = &ca->disk_sb.sb->layout;
1139         unsigned i;
1140         u64 b;
1141
1142         for (i = 0; i < layout->nr_superblocks; i++) {
1143                 u64 offset = le64_to_cpu(layout->sb_offset[i]);
1144
1145                 if (offset == BCH_SB_SECTOR)
1146                         mark_metadata_sectors(c, ca, 0, BCH_SB_SECTOR,
1147                                               BCH_DATA_sb, flags);
1148
1149                 mark_metadata_sectors(c, ca, offset,
1150                                       offset + (1 << layout->sb_max_size_bits),
1151                                       BCH_DATA_sb, flags);
1152         }
1153
1154         for (i = 0; i < ca->journal.nr; i++) {
1155                 b = ca->journal.buckets[i];
1156                 bch2_mark_metadata_bucket(c, ca, b, BCH_DATA_journal,
1157                                           ca->mi.bucket_size,
1158                                           gc_phase(GC_PHASE_SB), flags);
1159         }
1160 }
1161
1162 static void bch2_mark_superblocks(struct bch_fs *c)
1163 {
1164         mutex_lock(&c->sb_lock);
1165         gc_pos_set(c, gc_phase(GC_PHASE_SB));
1166
1167         for_each_online_member(c, ca)
1168                 bch2_mark_dev_superblock(c, ca, BTREE_TRIGGER_GC);
1169         mutex_unlock(&c->sb_lock);
1170 }
1171
1172 #if 0
1173 /* Also see bch2_pending_btree_node_free_insert_done() */
1174 static void bch2_mark_pending_btree_node_frees(struct bch_fs *c)
1175 {
1176         struct btree_update *as;
1177         struct pending_btree_node_free *d;
1178
1179         mutex_lock(&c->btree_interior_update_lock);
1180         gc_pos_set(c, gc_phase(GC_PHASE_PENDING_DELETE));
1181
1182         for_each_pending_btree_node_free(c, as, d)
1183                 if (d->index_update_done)
1184                         bch2_mark_key(c, bkey_i_to_s_c(&d->key), BTREE_TRIGGER_GC);
1185
1186         mutex_unlock(&c->btree_interior_update_lock);
1187 }
1188 #endif
1189
1190 static void bch2_gc_free(struct bch_fs *c)
1191 {
1192         genradix_free(&c->reflink_gc_table);
1193         genradix_free(&c->gc_stripes);
1194
1195         for_each_member_device(c, ca) {
1196                 kvfree(rcu_dereference_protected(ca->buckets_gc, 1));
1197                 ca->buckets_gc = NULL;
1198
1199                 free_percpu(ca->usage_gc);
1200                 ca->usage_gc = NULL;
1201         }
1202
1203         free_percpu(c->usage_gc);
1204         c->usage_gc = NULL;
1205 }
1206
1207 static int bch2_gc_done(struct bch_fs *c,
1208                         bool initial, bool metadata_only)
1209 {
1210         struct bch_dev *ca = NULL;
1211         struct printbuf buf = PRINTBUF;
1212         bool verify = !metadata_only &&
1213                 !c->opts.reconstruct_alloc &&
1214                 (!initial || (c->sb.compat & (1ULL << BCH_COMPAT_alloc_info)));
1215         unsigned i;
1216         int ret = 0;
1217
1218         percpu_down_write(&c->mark_lock);
1219
1220 #define copy_field(_err, _f, _msg, ...)                                 \
1221         if (dst->_f != src->_f &&                                       \
1222             (!verify ||                                                 \
1223              fsck_err(c, _err, _msg ": got %llu, should be %llu"        \
1224                       , ##__VA_ARGS__, dst->_f, src->_f)))              \
1225                 dst->_f = src->_f
1226 #define copy_dev_field(_err, _f, _msg, ...)                             \
1227         copy_field(_err, _f, "dev %u has wrong " _msg, ca->dev_idx, ##__VA_ARGS__)
1228 #define copy_fs_field(_err, _f, _msg, ...)                              \
1229         copy_field(_err, _f, "fs has wrong " _msg, ##__VA_ARGS__)
1230
1231         for (i = 0; i < ARRAY_SIZE(c->usage); i++)
1232                 bch2_fs_usage_acc_to_base(c, i);
1233
1234         __for_each_member_device(c, ca) {
1235                 struct bch_dev_usage *dst = ca->usage_base;
1236                 struct bch_dev_usage *src = (void *)
1237                         bch2_acc_percpu_u64s((u64 __percpu *) ca->usage_gc,
1238                                              dev_usage_u64s());
1239
1240                 for (i = 0; i < BCH_DATA_NR; i++) {
1241                         copy_dev_field(dev_usage_buckets_wrong,
1242                                        d[i].buckets,    "%s buckets", bch2_data_type_str(i));
1243                         copy_dev_field(dev_usage_sectors_wrong,
1244                                        d[i].sectors,    "%s sectors", bch2_data_type_str(i));
1245                         copy_dev_field(dev_usage_fragmented_wrong,
1246                                        d[i].fragmented, "%s fragmented", bch2_data_type_str(i));
1247                 }
1248         }
1249
1250         {
1251                 unsigned nr = fs_usage_u64s(c);
1252                 struct bch_fs_usage *dst = c->usage_base;
1253                 struct bch_fs_usage *src = (void *)
1254                         bch2_acc_percpu_u64s((u64 __percpu *) c->usage_gc, nr);
1255
1256                 copy_fs_field(fs_usage_hidden_wrong,
1257                               b.hidden,         "hidden");
1258                 copy_fs_field(fs_usage_btree_wrong,
1259                               b.btree,          "btree");
1260
1261                 if (!metadata_only) {
1262                         copy_fs_field(fs_usage_data_wrong,
1263                                       b.data,   "data");
1264                         copy_fs_field(fs_usage_cached_wrong,
1265                                       b.cached, "cached");
1266                         copy_fs_field(fs_usage_reserved_wrong,
1267                                       b.reserved,       "reserved");
1268                         copy_fs_field(fs_usage_nr_inodes_wrong,
1269                                       b.nr_inodes,"nr_inodes");
1270
1271                         for (i = 0; i < BCH_REPLICAS_MAX; i++)
1272                                 copy_fs_field(fs_usage_persistent_reserved_wrong,
1273                                               persistent_reserved[i],
1274                                               "persistent_reserved[%i]", i);
1275                 }
1276
1277                 for (i = 0; i < c->replicas.nr; i++) {
1278                         struct bch_replicas_entry_v1 *e =
1279                                 cpu_replicas_entry(&c->replicas, i);
1280
1281                         if (metadata_only &&
1282                             (e->data_type == BCH_DATA_user ||
1283                              e->data_type == BCH_DATA_cached))
1284                                 continue;
1285
1286                         printbuf_reset(&buf);
1287                         bch2_replicas_entry_to_text(&buf, e);
1288
1289                         copy_fs_field(fs_usage_replicas_wrong,
1290                                       replicas[i], "%s", buf.buf);
1291                 }
1292         }
1293
1294 #undef copy_fs_field
1295 #undef copy_dev_field
1296 #undef copy_stripe_field
1297 #undef copy_field
1298 fsck_err:
1299         if (ca)
1300                 percpu_ref_put(&ca->ref);
1301         bch_err_fn(c, ret);
1302
1303         percpu_up_write(&c->mark_lock);
1304         printbuf_exit(&buf);
1305         return ret;
1306 }
1307
1308 static int bch2_gc_start(struct bch_fs *c)
1309 {
1310         BUG_ON(c->usage_gc);
1311
1312         c->usage_gc = __alloc_percpu_gfp(fs_usage_u64s(c) * sizeof(u64),
1313                                          sizeof(u64), GFP_KERNEL);
1314         if (!c->usage_gc) {
1315                 bch_err(c, "error allocating c->usage_gc");
1316                 return -BCH_ERR_ENOMEM_gc_start;
1317         }
1318
1319         for_each_member_device(c, ca) {
1320                 BUG_ON(ca->usage_gc);
1321
1322                 ca->usage_gc = alloc_percpu(struct bch_dev_usage);
1323                 if (!ca->usage_gc) {
1324                         bch_err(c, "error allocating ca->usage_gc");
1325                         percpu_ref_put(&ca->ref);
1326                         return -BCH_ERR_ENOMEM_gc_start;
1327                 }
1328
1329                 this_cpu_write(ca->usage_gc->d[BCH_DATA_free].buckets,
1330                                ca->mi.nbuckets - ca->mi.first_bucket);
1331         }
1332
1333         return 0;
1334 }
1335
1336 static int bch2_gc_reset(struct bch_fs *c)
1337 {
1338         for_each_member_device(c, ca) {
1339                 free_percpu(ca->usage_gc);
1340                 ca->usage_gc = NULL;
1341         }
1342
1343         free_percpu(c->usage_gc);
1344         c->usage_gc = NULL;
1345
1346         return bch2_gc_start(c);
1347 }
1348
1349 /* returns true if not equal */
1350 static inline bool bch2_alloc_v4_cmp(struct bch_alloc_v4 l,
1351                                      struct bch_alloc_v4 r)
1352 {
1353         return  l.gen != r.gen                          ||
1354                 l.oldest_gen != r.oldest_gen            ||
1355                 l.data_type != r.data_type              ||
1356                 l.dirty_sectors != r.dirty_sectors      ||
1357                 l.cached_sectors != r.cached_sectors     ||
1358                 l.stripe_redundancy != r.stripe_redundancy ||
1359                 l.stripe != r.stripe;
1360 }
1361
1362 static int bch2_alloc_write_key(struct btree_trans *trans,
1363                                 struct btree_iter *iter,
1364                                 struct bkey_s_c k,
1365                                 bool metadata_only)
1366 {
1367         struct bch_fs *c = trans->c;
1368         struct bch_dev *ca = bch_dev_bkey_exists(c, iter->pos.inode);
1369         struct bucket old_gc, gc, *b;
1370         struct bkey_i_alloc_v4 *a;
1371         struct bch_alloc_v4 old_convert, new;
1372         const struct bch_alloc_v4 *old;
1373         int ret;
1374
1375         old = bch2_alloc_to_v4(k, &old_convert);
1376         new = *old;
1377
1378         percpu_down_read(&c->mark_lock);
1379         b = gc_bucket(ca, iter->pos.offset);
1380         old_gc = *b;
1381
1382         if ((old->data_type == BCH_DATA_sb ||
1383              old->data_type == BCH_DATA_journal) &&
1384             !bch2_dev_is_online(ca)) {
1385                 b->data_type = old->data_type;
1386                 b->dirty_sectors = old->dirty_sectors;
1387         }
1388
1389         /*
1390          * b->data_type doesn't yet include need_discard & need_gc_gen states -
1391          * fix that here:
1392          */
1393         b->data_type = __alloc_data_type(b->dirty_sectors,
1394                                          b->cached_sectors,
1395                                          b->stripe,
1396                                          *old,
1397                                          b->data_type);
1398         gc = *b;
1399
1400         if (gc.data_type != old_gc.data_type ||
1401             gc.dirty_sectors != old_gc.dirty_sectors)
1402                 bch2_dev_usage_update_m(c, ca, &old_gc, &gc);
1403         percpu_up_read(&c->mark_lock);
1404
1405         if (metadata_only &&
1406             gc.data_type != BCH_DATA_sb &&
1407             gc.data_type != BCH_DATA_journal &&
1408             gc.data_type != BCH_DATA_btree)
1409                 return 0;
1410
1411         if (gen_after(old->gen, gc.gen))
1412                 return 0;
1413
1414         if (fsck_err_on(new.data_type != gc.data_type, c,
1415                         alloc_key_data_type_wrong,
1416                         "bucket %llu:%llu gen %u has wrong data_type"
1417                         ": got %s, should be %s",
1418                         iter->pos.inode, iter->pos.offset,
1419                         gc.gen,
1420                         bch2_data_type_str(new.data_type),
1421                         bch2_data_type_str(gc.data_type)))
1422                 new.data_type = gc.data_type;
1423
1424 #define copy_bucket_field(_errtype, _f)                                 \
1425         if (fsck_err_on(new._f != gc._f, c, _errtype,                   \
1426                         "bucket %llu:%llu gen %u data type %s has wrong " #_f   \
1427                         ": got %u, should be %u",                       \
1428                         iter->pos.inode, iter->pos.offset,              \
1429                         gc.gen,                                         \
1430                         bch2_data_type_str(gc.data_type),               \
1431                         new._f, gc._f))                                 \
1432                 new._f = gc._f;                                         \
1433
1434         copy_bucket_field(alloc_key_gen_wrong,
1435                           gen);
1436         copy_bucket_field(alloc_key_dirty_sectors_wrong,
1437                           dirty_sectors);
1438         copy_bucket_field(alloc_key_cached_sectors_wrong,
1439                           cached_sectors);
1440         copy_bucket_field(alloc_key_stripe_wrong,
1441                           stripe);
1442         copy_bucket_field(alloc_key_stripe_redundancy_wrong,
1443                           stripe_redundancy);
1444 #undef copy_bucket_field
1445
1446         if (!bch2_alloc_v4_cmp(*old, new))
1447                 return 0;
1448
1449         a = bch2_alloc_to_v4_mut(trans, k);
1450         ret = PTR_ERR_OR_ZERO(a);
1451         if (ret)
1452                 return ret;
1453
1454         a->v = new;
1455
1456         /*
1457          * The trigger normally makes sure this is set, but we're not running
1458          * triggers:
1459          */
1460         if (a->v.data_type == BCH_DATA_cached && !a->v.io_time[READ])
1461                 a->v.io_time[READ] = max_t(u64, 1, atomic64_read(&c->io_clock[READ].now));
1462
1463         ret = bch2_trans_update(trans, iter, &a->k_i, BTREE_TRIGGER_NORUN);
1464 fsck_err:
1465         return ret;
1466 }
1467
1468 static int bch2_gc_alloc_done(struct bch_fs *c, bool metadata_only)
1469 {
1470         int ret = 0;
1471
1472         for_each_member_device(c, ca) {
1473                 ret = bch2_trans_run(c,
1474                         for_each_btree_key_upto_commit(trans, iter, BTREE_ID_alloc,
1475                                         POS(ca->dev_idx, ca->mi.first_bucket),
1476                                         POS(ca->dev_idx, ca->mi.nbuckets - 1),
1477                                         BTREE_ITER_SLOTS|BTREE_ITER_PREFETCH, k,
1478                                         NULL, NULL, BCH_TRANS_COMMIT_lazy_rw,
1479                                 bch2_alloc_write_key(trans, &iter, k, metadata_only)));
1480                 if (ret) {
1481                         percpu_ref_put(&ca->ref);
1482                         break;
1483                 }
1484         }
1485
1486         bch_err_fn(c, ret);
1487         return ret;
1488 }
1489
1490 static int bch2_gc_alloc_start(struct bch_fs *c, bool metadata_only)
1491 {
1492         for_each_member_device(c, ca) {
1493                 struct bucket_array *buckets = kvmalloc(sizeof(struct bucket_array) +
1494                                 ca->mi.nbuckets * sizeof(struct bucket),
1495                                 GFP_KERNEL|__GFP_ZERO);
1496                 if (!buckets) {
1497                         percpu_ref_put(&ca->ref);
1498                         bch_err(c, "error allocating ca->buckets[gc]");
1499                         return -BCH_ERR_ENOMEM_gc_alloc_start;
1500                 }
1501
1502                 buckets->first_bucket   = ca->mi.first_bucket;
1503                 buckets->nbuckets       = ca->mi.nbuckets;
1504                 rcu_assign_pointer(ca->buckets_gc, buckets);
1505         }
1506
1507         int ret = bch2_trans_run(c,
1508                 for_each_btree_key(trans, iter, BTREE_ID_alloc, POS_MIN,
1509                                          BTREE_ITER_PREFETCH, k, ({
1510                         struct bch_dev *ca = bch_dev_bkey_exists(c, k.k->p.inode);
1511                         struct bucket *g = gc_bucket(ca, k.k->p.offset);
1512
1513                         struct bch_alloc_v4 a_convert;
1514                         const struct bch_alloc_v4 *a = bch2_alloc_to_v4(k, &a_convert);
1515
1516                         g->gen_valid    = 1;
1517                         g->gen          = a->gen;
1518
1519                         if (metadata_only &&
1520                             (a->data_type == BCH_DATA_user ||
1521                              a->data_type == BCH_DATA_cached ||
1522                              a->data_type == BCH_DATA_parity)) {
1523                                 g->data_type            = a->data_type;
1524                                 g->dirty_sectors        = a->dirty_sectors;
1525                                 g->cached_sectors       = a->cached_sectors;
1526                                 g->stripe               = a->stripe;
1527                                 g->stripe_redundancy    = a->stripe_redundancy;
1528                         }
1529
1530                         0;
1531                 })));
1532         bch_err_fn(c, ret);
1533         return ret;
1534 }
1535
1536 static void bch2_gc_alloc_reset(struct bch_fs *c, bool metadata_only)
1537 {
1538         for_each_member_device(c, ca) {
1539                 struct bucket_array *buckets = gc_bucket_array(ca);
1540                 struct bucket *g;
1541
1542                 for_each_bucket(g, buckets) {
1543                         if (metadata_only &&
1544                             (g->data_type == BCH_DATA_user ||
1545                              g->data_type == BCH_DATA_cached ||
1546                              g->data_type == BCH_DATA_parity))
1547                                 continue;
1548                         g->data_type = 0;
1549                         g->dirty_sectors = 0;
1550                         g->cached_sectors = 0;
1551                 }
1552         }
1553 }
1554
1555 static int bch2_gc_write_reflink_key(struct btree_trans *trans,
1556                                      struct btree_iter *iter,
1557                                      struct bkey_s_c k,
1558                                      size_t *idx)
1559 {
1560         struct bch_fs *c = trans->c;
1561         const __le64 *refcount = bkey_refcount_c(k);
1562         struct printbuf buf = PRINTBUF;
1563         struct reflink_gc *r;
1564         int ret = 0;
1565
1566         if (!refcount)
1567                 return 0;
1568
1569         while ((r = genradix_ptr(&c->reflink_gc_table, *idx)) &&
1570                r->offset < k.k->p.offset)
1571                 ++*idx;
1572
1573         if (!r ||
1574             r->offset != k.k->p.offset ||
1575             r->size != k.k->size) {
1576                 bch_err(c, "unexpected inconsistency walking reflink table at gc finish");
1577                 return -EINVAL;
1578         }
1579
1580         if (fsck_err_on(r->refcount != le64_to_cpu(*refcount), c,
1581                         reflink_v_refcount_wrong,
1582                         "reflink key has wrong refcount:\n"
1583                         "  %s\n"
1584                         "  should be %u",
1585                         (bch2_bkey_val_to_text(&buf, c, k), buf.buf),
1586                         r->refcount)) {
1587                 struct bkey_i *new = bch2_bkey_make_mut_noupdate(trans, k);
1588                 ret = PTR_ERR_OR_ZERO(new);
1589                 if (ret)
1590                         goto out;
1591
1592                 if (!r->refcount)
1593                         new->k.type = KEY_TYPE_deleted;
1594                 else
1595                         *bkey_refcount(bkey_i_to_s(new)) = cpu_to_le64(r->refcount);
1596                 ret = bch2_trans_update(trans, iter, new, 0);
1597         }
1598 out:
1599 fsck_err:
1600         printbuf_exit(&buf);
1601         return ret;
1602 }
1603
1604 static int bch2_gc_reflink_done(struct bch_fs *c, bool metadata_only)
1605 {
1606         size_t idx = 0;
1607
1608         if (metadata_only)
1609                 return 0;
1610
1611         int ret = bch2_trans_run(c,
1612                 for_each_btree_key_commit(trans, iter,
1613                                 BTREE_ID_reflink, POS_MIN,
1614                                 BTREE_ITER_PREFETCH, k,
1615                                 NULL, NULL, BCH_TRANS_COMMIT_no_enospc,
1616                         bch2_gc_write_reflink_key(trans, &iter, k, &idx)));
1617         c->reflink_gc_nr = 0;
1618         return ret;
1619 }
1620
1621 static int bch2_gc_reflink_start(struct bch_fs *c,
1622                                  bool metadata_only)
1623 {
1624
1625         if (metadata_only)
1626                 return 0;
1627
1628         c->reflink_gc_nr = 0;
1629
1630         int ret = bch2_trans_run(c,
1631                 for_each_btree_key(trans, iter, BTREE_ID_reflink, POS_MIN,
1632                                    BTREE_ITER_PREFETCH, k, ({
1633                         const __le64 *refcount = bkey_refcount_c(k);
1634
1635                         if (!refcount)
1636                                 continue;
1637
1638                         struct reflink_gc *r = genradix_ptr_alloc(&c->reflink_gc_table,
1639                                                         c->reflink_gc_nr++, GFP_KERNEL);
1640                         if (!r) {
1641                                 ret = -BCH_ERR_ENOMEM_gc_reflink_start;
1642                                 break;
1643                         }
1644
1645                         r->offset       = k.k->p.offset;
1646                         r->size         = k.k->size;
1647                         r->refcount     = 0;
1648                         0;
1649                 })));
1650
1651         bch_err_fn(c, ret);
1652         return ret;
1653 }
1654
1655 static void bch2_gc_reflink_reset(struct bch_fs *c, bool metadata_only)
1656 {
1657         struct genradix_iter iter;
1658         struct reflink_gc *r;
1659
1660         genradix_for_each(&c->reflink_gc_table, iter, r)
1661                 r->refcount = 0;
1662 }
1663
1664 static int bch2_gc_write_stripes_key(struct btree_trans *trans,
1665                                      struct btree_iter *iter,
1666                                      struct bkey_s_c k)
1667 {
1668         struct bch_fs *c = trans->c;
1669         struct printbuf buf = PRINTBUF;
1670         const struct bch_stripe *s;
1671         struct gc_stripe *m;
1672         bool bad = false;
1673         unsigned i;
1674         int ret = 0;
1675
1676         if (k.k->type != KEY_TYPE_stripe)
1677                 return 0;
1678
1679         s = bkey_s_c_to_stripe(k).v;
1680         m = genradix_ptr(&c->gc_stripes, k.k->p.offset);
1681
1682         for (i = 0; i < s->nr_blocks; i++) {
1683                 u32 old = stripe_blockcount_get(s, i);
1684                 u32 new = (m ? m->block_sectors[i] : 0);
1685
1686                 if (old != new) {
1687                         prt_printf(&buf, "stripe block %u has wrong sector count: got %u, should be %u\n",
1688                                    i, old, new);
1689                         bad = true;
1690                 }
1691         }
1692
1693         if (bad)
1694                 bch2_bkey_val_to_text(&buf, c, k);
1695
1696         if (fsck_err_on(bad, c, stripe_sector_count_wrong,
1697                         "%s", buf.buf)) {
1698                 struct bkey_i_stripe *new;
1699
1700                 new = bch2_trans_kmalloc(trans, bkey_bytes(k.k));
1701                 ret = PTR_ERR_OR_ZERO(new);
1702                 if (ret)
1703                         return ret;
1704
1705                 bkey_reassemble(&new->k_i, k);
1706
1707                 for (i = 0; i < new->v.nr_blocks; i++)
1708                         stripe_blockcount_set(&new->v, i, m ? m->block_sectors[i] : 0);
1709
1710                 ret = bch2_trans_update(trans, iter, &new->k_i, 0);
1711         }
1712 fsck_err:
1713         printbuf_exit(&buf);
1714         return ret;
1715 }
1716
1717 static int bch2_gc_stripes_done(struct bch_fs *c, bool metadata_only)
1718 {
1719         if (metadata_only)
1720                 return 0;
1721
1722         return bch2_trans_run(c,
1723                 for_each_btree_key_commit(trans, iter,
1724                                 BTREE_ID_stripes, POS_MIN,
1725                                 BTREE_ITER_PREFETCH, k,
1726                                 NULL, NULL, BCH_TRANS_COMMIT_no_enospc,
1727                         bch2_gc_write_stripes_key(trans, &iter, k)));
1728 }
1729
1730 static void bch2_gc_stripes_reset(struct bch_fs *c, bool metadata_only)
1731 {
1732         genradix_free(&c->gc_stripes);
1733 }
1734
1735 /**
1736  * bch2_gc - walk _all_ references to buckets, and recompute them:
1737  *
1738  * @c:                  filesystem object
1739  * @initial:            are we in recovery?
1740  * @metadata_only:      are we just checking metadata references, or everything?
1741  *
1742  * Returns: 0 on success, or standard errcode on failure
1743  *
1744  * Order matters here:
1745  *  - Concurrent GC relies on the fact that we have a total ordering for
1746  *    everything that GC walks - see  gc_will_visit_node(),
1747  *    gc_will_visit_root()
1748  *
1749  *  - also, references move around in the course of index updates and
1750  *    various other crap: everything needs to agree on the ordering
1751  *    references are allowed to move around in - e.g., we're allowed to
1752  *    start with a reference owned by an open_bucket (the allocator) and
1753  *    move it to the btree, but not the reverse.
1754  *
1755  *    This is necessary to ensure that gc doesn't miss references that
1756  *    move around - if references move backwards in the ordering GC
1757  *    uses, GC could skip past them
1758  */
1759 int bch2_gc(struct bch_fs *c, bool initial, bool metadata_only)
1760 {
1761         unsigned iter = 0;
1762         int ret;
1763
1764         lockdep_assert_held(&c->state_lock);
1765
1766         down_write(&c->gc_lock);
1767
1768         bch2_btree_interior_updates_flush(c);
1769
1770         ret   = bch2_gc_start(c) ?:
1771                 bch2_gc_alloc_start(c, metadata_only) ?:
1772                 bch2_gc_reflink_start(c, metadata_only);
1773         if (ret)
1774                 goto out;
1775 again:
1776         gc_pos_set(c, gc_phase(GC_PHASE_START));
1777
1778         bch2_mark_superblocks(c);
1779
1780         ret = bch2_gc_btrees(c, initial, metadata_only);
1781
1782         if (ret)
1783                 goto out;
1784
1785 #if 0
1786         bch2_mark_pending_btree_node_frees(c);
1787 #endif
1788         c->gc_count++;
1789
1790         if (test_bit(BCH_FS_need_another_gc, &c->flags) ||
1791             (!iter && bch2_test_restart_gc)) {
1792                 if (iter++ > 2) {
1793                         bch_info(c, "Unable to fix bucket gens, looping");
1794                         ret = -EINVAL;
1795                         goto out;
1796                 }
1797
1798                 /*
1799                  * XXX: make sure gens we fixed got saved
1800                  */
1801                 bch_info(c, "Second GC pass needed, restarting:");
1802                 clear_bit(BCH_FS_need_another_gc, &c->flags);
1803                 __gc_pos_set(c, gc_phase(GC_PHASE_NOT_RUNNING));
1804
1805                 bch2_gc_stripes_reset(c, metadata_only);
1806                 bch2_gc_alloc_reset(c, metadata_only);
1807                 bch2_gc_reflink_reset(c, metadata_only);
1808                 ret = bch2_gc_reset(c);
1809                 if (ret)
1810                         goto out;
1811
1812                 /* flush fsck errors, reset counters */
1813                 bch2_flush_fsck_errs(c);
1814                 goto again;
1815         }
1816 out:
1817         if (!ret) {
1818                 bch2_journal_block(&c->journal);
1819
1820                 ret   = bch2_gc_alloc_done(c, metadata_only) ?:
1821                         bch2_gc_done(c, initial, metadata_only) ?:
1822                         bch2_gc_stripes_done(c, metadata_only) ?:
1823                         bch2_gc_reflink_done(c, metadata_only);
1824
1825                 bch2_journal_unblock(&c->journal);
1826         }
1827
1828         percpu_down_write(&c->mark_lock);
1829         /* Indicates that gc is no longer in progress: */
1830         __gc_pos_set(c, gc_phase(GC_PHASE_NOT_RUNNING));
1831
1832         bch2_gc_free(c);
1833         percpu_up_write(&c->mark_lock);
1834
1835         up_write(&c->gc_lock);
1836
1837         /*
1838          * At startup, allocations can happen directly instead of via the
1839          * allocator thread - issue wakeup in case they blocked on gc_lock:
1840          */
1841         closure_wake_up(&c->freelist_wait);
1842         bch_err_fn(c, ret);
1843         return ret;
1844 }
1845
1846 static int gc_btree_gens_key(struct btree_trans *trans,
1847                              struct btree_iter *iter,
1848                              struct bkey_s_c k)
1849 {
1850         struct bch_fs *c = trans->c;
1851         struct bkey_ptrs_c ptrs = bch2_bkey_ptrs_c(k);
1852         struct bkey_i *u;
1853         int ret;
1854
1855         percpu_down_read(&c->mark_lock);
1856         bkey_for_each_ptr(ptrs, ptr) {
1857                 struct bch_dev *ca = bch_dev_bkey_exists(c, ptr->dev);
1858
1859                 if (ptr_stale(ca, ptr) > 16) {
1860                         percpu_up_read(&c->mark_lock);
1861                         goto update;
1862                 }
1863         }
1864
1865         bkey_for_each_ptr(ptrs, ptr) {
1866                 struct bch_dev *ca = bch_dev_bkey_exists(c, ptr->dev);
1867                 u8 *gen = &ca->oldest_gen[PTR_BUCKET_NR(ca, ptr)];
1868
1869                 if (gen_after(*gen, ptr->gen))
1870                         *gen = ptr->gen;
1871         }
1872         percpu_up_read(&c->mark_lock);
1873         return 0;
1874 update:
1875         u = bch2_bkey_make_mut(trans, iter, &k, 0);
1876         ret = PTR_ERR_OR_ZERO(u);
1877         if (ret)
1878                 return ret;
1879
1880         bch2_extent_normalize(c, bkey_i_to_s(u));
1881         return 0;
1882 }
1883
1884 static int bch2_alloc_write_oldest_gen(struct btree_trans *trans, struct btree_iter *iter,
1885                                        struct bkey_s_c k)
1886 {
1887         struct bch_dev *ca = bch_dev_bkey_exists(trans->c, iter->pos.inode);
1888         struct bch_alloc_v4 a_convert;
1889         const struct bch_alloc_v4 *a = bch2_alloc_to_v4(k, &a_convert);
1890         struct bkey_i_alloc_v4 *a_mut;
1891         int ret;
1892
1893         if (a->oldest_gen == ca->oldest_gen[iter->pos.offset])
1894                 return 0;
1895
1896         a_mut = bch2_alloc_to_v4_mut(trans, k);
1897         ret = PTR_ERR_OR_ZERO(a_mut);
1898         if (ret)
1899                 return ret;
1900
1901         a_mut->v.oldest_gen = ca->oldest_gen[iter->pos.offset];
1902         a_mut->v.data_type = alloc_data_type(a_mut->v, a_mut->v.data_type);
1903
1904         return bch2_trans_update(trans, iter, &a_mut->k_i, 0);
1905 }
1906
1907 int bch2_gc_gens(struct bch_fs *c)
1908 {
1909         u64 b, start_time = local_clock();
1910         int ret;
1911
1912         /*
1913          * Ideally we would be using state_lock and not gc_lock here, but that
1914          * introduces a deadlock in the RO path - we currently take the state
1915          * lock at the start of going RO, thus the gc thread may get stuck:
1916          */
1917         if (!mutex_trylock(&c->gc_gens_lock))
1918                 return 0;
1919
1920         trace_and_count(c, gc_gens_start, c);
1921         down_read(&c->gc_lock);
1922
1923         for_each_member_device(c, ca) {
1924                 struct bucket_gens *gens = bucket_gens(ca);
1925
1926                 BUG_ON(ca->oldest_gen);
1927
1928                 ca->oldest_gen = kvmalloc(gens->nbuckets, GFP_KERNEL);
1929                 if (!ca->oldest_gen) {
1930                         percpu_ref_put(&ca->ref);
1931                         ret = -BCH_ERR_ENOMEM_gc_gens;
1932                         goto err;
1933                 }
1934
1935                 for (b = gens->first_bucket;
1936                      b < gens->nbuckets; b++)
1937                         ca->oldest_gen[b] = gens->b[b];
1938         }
1939
1940         for (unsigned i = 0; i < BTREE_ID_NR; i++)
1941                 if (btree_type_has_ptrs(i)) {
1942                         c->gc_gens_btree = i;
1943                         c->gc_gens_pos = POS_MIN;
1944
1945                         ret = bch2_trans_run(c,
1946                                 for_each_btree_key_commit(trans, iter, i,
1947                                                 POS_MIN,
1948                                                 BTREE_ITER_PREFETCH|BTREE_ITER_ALL_SNAPSHOTS,
1949                                                 k,
1950                                                 NULL, NULL,
1951                                                 BCH_TRANS_COMMIT_no_enospc,
1952                                         gc_btree_gens_key(trans, &iter, k)));
1953                         if (ret)
1954                                 goto err;
1955                 }
1956
1957         ret = bch2_trans_run(c,
1958                 for_each_btree_key_commit(trans, iter, BTREE_ID_alloc,
1959                                 POS_MIN,
1960                                 BTREE_ITER_PREFETCH,
1961                                 k,
1962                                 NULL, NULL,
1963                                 BCH_TRANS_COMMIT_no_enospc,
1964                         bch2_alloc_write_oldest_gen(trans, &iter, k)));
1965         if (ret)
1966                 goto err;
1967
1968         c->gc_gens_btree        = 0;
1969         c->gc_gens_pos          = POS_MIN;
1970
1971         c->gc_count++;
1972
1973         bch2_time_stats_update(&c->times[BCH_TIME_btree_gc], start_time);
1974         trace_and_count(c, gc_gens_end, c);
1975 err:
1976         for_each_member_device(c, ca) {
1977                 kvfree(ca->oldest_gen);
1978                 ca->oldest_gen = NULL;
1979         }
1980
1981         up_read(&c->gc_lock);
1982         mutex_unlock(&c->gc_gens_lock);
1983         if (!bch2_err_matches(ret, EROFS))
1984                 bch_err_fn(c, ret);
1985         return ret;
1986 }
1987
1988 static int bch2_gc_thread(void *arg)
1989 {
1990         struct bch_fs *c = arg;
1991         struct io_clock *clock = &c->io_clock[WRITE];
1992         unsigned long last = atomic64_read(&clock->now);
1993         unsigned last_kick = atomic_read(&c->kick_gc);
1994
1995         set_freezable();
1996
1997         while (1) {
1998                 while (1) {
1999                         set_current_state(TASK_INTERRUPTIBLE);
2000
2001                         if (kthread_should_stop()) {
2002                                 __set_current_state(TASK_RUNNING);
2003                                 return 0;
2004                         }
2005
2006                         if (atomic_read(&c->kick_gc) != last_kick)
2007                                 break;
2008
2009                         if (c->btree_gc_periodic) {
2010                                 unsigned long next = last + c->capacity / 16;
2011
2012                                 if (atomic64_read(&clock->now) >= next)
2013                                         break;
2014
2015                                 bch2_io_clock_schedule_timeout(clock, next);
2016                         } else {
2017                                 schedule();
2018                         }
2019
2020                         try_to_freeze();
2021                 }
2022                 __set_current_state(TASK_RUNNING);
2023
2024                 last = atomic64_read(&clock->now);
2025                 last_kick = atomic_read(&c->kick_gc);
2026
2027                 /*
2028                  * Full gc is currently incompatible with btree key cache:
2029                  */
2030 #if 0
2031                 ret = bch2_gc(c, false, false);
2032 #else
2033                 bch2_gc_gens(c);
2034 #endif
2035                 debug_check_no_locks_held();
2036         }
2037
2038         return 0;
2039 }
2040
2041 void bch2_gc_thread_stop(struct bch_fs *c)
2042 {
2043         struct task_struct *p;
2044
2045         p = c->gc_thread;
2046         c->gc_thread = NULL;
2047
2048         if (p) {
2049                 kthread_stop(p);
2050                 put_task_struct(p);
2051         }
2052 }
2053
2054 int bch2_gc_thread_start(struct bch_fs *c)
2055 {
2056         struct task_struct *p;
2057
2058         if (c->gc_thread)
2059                 return 0;
2060
2061         p = kthread_create(bch2_gc_thread, c, "bch-gc/%s", c->name);
2062         if (IS_ERR(p)) {
2063                 bch_err_fn(c, PTR_ERR(p));
2064                 return PTR_ERR(p);
2065         }
2066
2067         get_task_struct(p);
2068         c->gc_thread = p;
2069         wake_up_process(p);
2070         return 0;
2071 }