diff options
Diffstat (limited to 'fs/bcachefs/buckets.c')
| -rw-r--r-- | fs/bcachefs/buckets.c | 1672 |
1 files changed, 818 insertions, 854 deletions
diff --git a/fs/bcachefs/buckets.c b/fs/bcachefs/buckets.c index 96edf2c34d43..f25903c10e8a 100644 --- a/fs/bcachefs/buckets.c +++ b/fs/bcachefs/buckets.c @@ -13,11 +13,14 @@ #include "btree_update.h" #include "buckets.h" #include "buckets_waiting_for_journal.h" +#include "disk_accounting.h" #include "ec.h" #include "error.h" #include "inode.h" #include "movinggc.h" +#include "rebalance.h" #include "recovery.h" +#include "recovery_passes.h" #include "reflink.h" #include "replicas.h" #include "subvolume.h" @@ -25,197 +28,17 @@ #include <linux/preempt.h> -static inline void fs_usage_data_type_to_base(struct bch_fs_usage_base *fs_usage, - enum bch_data_type data_type, - s64 sectors) -{ - switch (data_type) { - case BCH_DATA_btree: - fs_usage->btree += sectors; - break; - case BCH_DATA_user: - case BCH_DATA_parity: - fs_usage->data += sectors; - break; - case BCH_DATA_cached: - fs_usage->cached += sectors; - break; - default: - break; - } -} - -void bch2_fs_usage_initialize(struct bch_fs *c) -{ - percpu_down_write(&c->mark_lock); - struct bch_fs_usage *usage = c->usage_base; - - for (unsigned i = 0; i < ARRAY_SIZE(c->usage); i++) - bch2_fs_usage_acc_to_base(c, i); - - for (unsigned i = 0; i < BCH_REPLICAS_MAX; i++) - usage->b.reserved += usage->persistent_reserved[i]; - - for (unsigned i = 0; i < c->replicas.nr; i++) { - struct bch_replicas_entry_v1 *e = - cpu_replicas_entry(&c->replicas, i); - - fs_usage_data_type_to_base(&usage->b, e->data_type, usage->replicas[i]); - } - - for_each_member_device(c, ca) { - struct bch_dev_usage dev = bch2_dev_usage_read(ca); - - usage->b.hidden += (dev.d[BCH_DATA_sb].buckets + - dev.d[BCH_DATA_journal].buckets) * - ca->mi.bucket_size; - } - - percpu_up_write(&c->mark_lock); -} - -static inline struct bch_dev_usage *dev_usage_ptr(struct bch_dev *ca, - unsigned journal_seq, - bool gc) -{ - BUG_ON(!gc && !journal_seq); - - return this_cpu_ptr(gc - ? ca->usage_gc - : ca->usage[journal_seq & JOURNAL_BUF_MASK]); -} - void bch2_dev_usage_read_fast(struct bch_dev *ca, struct bch_dev_usage *usage) { - struct bch_fs *c = ca->fs; - unsigned seq, i, u64s = dev_usage_u64s(); - - do { - seq = read_seqcount_begin(&c->usage_lock); - memcpy(usage, ca->usage_base, u64s * sizeof(u64)); - for (i = 0; i < ARRAY_SIZE(ca->usage); i++) - acc_u64s_percpu((u64 *) usage, (u64 __percpu *) ca->usage[i], u64s); - } while (read_seqcount_retry(&c->usage_lock, seq)); -} - -u64 bch2_fs_usage_read_one(struct bch_fs *c, u64 *v) -{ - ssize_t offset = v - (u64 *) c->usage_base; - unsigned i, seq; - u64 ret; - - BUG_ON(offset < 0 || offset >= fs_usage_u64s(c)); - percpu_rwsem_assert_held(&c->mark_lock); - - do { - seq = read_seqcount_begin(&c->usage_lock); - ret = *v; - - for (i = 0; i < ARRAY_SIZE(c->usage); i++) - ret += percpu_u64_get((u64 __percpu *) c->usage[i] + offset); - } while (read_seqcount_retry(&c->usage_lock, seq)); - - return ret; -} - -struct bch_fs_usage_online *bch2_fs_usage_read(struct bch_fs *c) -{ - struct bch_fs_usage_online *ret; - unsigned nr_replicas = READ_ONCE(c->replicas.nr); - unsigned seq, i; -retry: - ret = kmalloc(__fs_usage_online_u64s(nr_replicas) * sizeof(u64), GFP_KERNEL); - if (unlikely(!ret)) - return NULL; - - percpu_down_read(&c->mark_lock); - - if (nr_replicas != c->replicas.nr) { - nr_replicas = c->replicas.nr; - percpu_up_read(&c->mark_lock); - kfree(ret); - goto retry; - } - - ret->online_reserved = percpu_u64_get(c->online_reserved); - - do { - seq = read_seqcount_begin(&c->usage_lock); - unsafe_memcpy(&ret->u, c->usage_base, - __fs_usage_u64s(nr_replicas) * sizeof(u64), - "embedded variable length struct"); - for (i = 0; i < ARRAY_SIZE(c->usage); i++) - acc_u64s_percpu((u64 *) &ret->u, (u64 __percpu *) c->usage[i], - __fs_usage_u64s(nr_replicas)); - } while (read_seqcount_retry(&c->usage_lock, seq)); - - return ret; -} - -void bch2_fs_usage_acc_to_base(struct bch_fs *c, unsigned idx) -{ - unsigned u64s = fs_usage_u64s(c); - - BUG_ON(idx >= ARRAY_SIZE(c->usage)); - - preempt_disable(); - write_seqcount_begin(&c->usage_lock); - - acc_u64s_percpu((u64 *) c->usage_base, - (u64 __percpu *) c->usage[idx], u64s); - percpu_memset(c->usage[idx], 0, u64s * sizeof(u64)); - - rcu_read_lock(); - for_each_member_device_rcu(c, ca, NULL) { - u64s = dev_usage_u64s(); - - acc_u64s_percpu((u64 *) ca->usage_base, - (u64 __percpu *) ca->usage[idx], u64s); - percpu_memset(ca->usage[idx], 0, u64s * sizeof(u64)); - } - rcu_read_unlock(); - - write_seqcount_end(&c->usage_lock); - preempt_enable(); + for (unsigned i = 0; i < BCH_DATA_NR; i++) + usage->buckets[i] = percpu_u64_get(&ca->usage->d[i].buckets); } -void bch2_fs_usage_to_text(struct printbuf *out, - struct bch_fs *c, - struct bch_fs_usage_online *fs_usage) +void bch2_dev_usage_full_read_fast(struct bch_dev *ca, struct bch_dev_usage_full *usage) { - unsigned i; - - prt_printf(out, "capacity:\t\t\t%llu\n", c->capacity); - - prt_printf(out, "hidden:\t\t\t\t%llu\n", - fs_usage->u.b.hidden); - prt_printf(out, "data:\t\t\t\t%llu\n", - fs_usage->u.b.data); - prt_printf(out, "cached:\t\t\t\t%llu\n", - fs_usage->u.b.cached); - prt_printf(out, "reserved:\t\t\t%llu\n", - fs_usage->u.b.reserved); - prt_printf(out, "nr_inodes:\t\t\t%llu\n", - fs_usage->u.b.nr_inodes); - prt_printf(out, "online reserved:\t\t%llu\n", - fs_usage->online_reserved); - - for (i = 0; - i < ARRAY_SIZE(fs_usage->u.persistent_reserved); - i++) { - prt_printf(out, "%u replicas:\n", i + 1); - prt_printf(out, "\treserved:\t\t%llu\n", - fs_usage->u.persistent_reserved[i]); - } - - for (i = 0; i < c->replicas.nr; i++) { - struct bch_replicas_entry_v1 *e = - cpu_replicas_entry(&c->replicas, i); - - prt_printf(out, "\t"); - bch2_replicas_entry_to_text(out, e); - prt_printf(out, ":\t%llu\n", fs_usage->u.replicas[i]); - } + memset(usage, 0, sizeof(*usage)); + acc_u64s_percpu((u64 *) usage, (u64 __percpu *) ca->usage, + sizeof(struct bch_dev_usage_full) / sizeof(u64)); } static u64 reserve_factor(u64 r) @@ -223,16 +46,6 @@ static u64 reserve_factor(u64 r) return r + (round_up(r, (1 << RESERVE_FACTOR)) >> RESERVE_FACTOR); } -u64 bch2_fs_sectors_used(struct bch_fs *c, struct bch_fs_usage_online *fs_usage) -{ - return min(fs_usage->u.b.hidden + - fs_usage->u.b.btree + - fs_usage->u.b.data + - reserve_factor(fs_usage->u.b.reserved + - fs_usage->online_reserved), - c->capacity); -} - static struct bch_fs_usage_short __bch2_fs_usage_read_short(struct bch_fs *c) { @@ -240,17 +53,17 @@ __bch2_fs_usage_read_short(struct bch_fs *c) u64 data, reserved; ret.capacity = c->capacity - - bch2_fs_usage_read_one(c, &c->usage_base->b.hidden); + percpu_u64_get(&c->usage->hidden); - data = bch2_fs_usage_read_one(c, &c->usage_base->b.data) + - bch2_fs_usage_read_one(c, &c->usage_base->b.btree); - reserved = bch2_fs_usage_read_one(c, &c->usage_base->b.reserved) + + data = percpu_u64_get(&c->usage->data) + + percpu_u64_get(&c->usage->btree); + reserved = percpu_u64_get(&c->usage->reserved) + percpu_u64_get(c->online_reserved); ret.used = min(ret.capacity, data + reserve_factor(reserved)); ret.free = ret.capacity - ret.used; - ret.nr_inodes = bch2_fs_usage_read_one(c, &c->usage_base->b.nr_inodes); + ret.nr_inodes = percpu_u64_get(&c->usage->nr_inodes); return ret; } @@ -267,431 +80,467 @@ bch2_fs_usage_read_short(struct bch_fs *c) return ret; } -void bch2_dev_usage_init(struct bch_dev *ca) +void bch2_dev_usage_to_text(struct printbuf *out, + struct bch_dev *ca, + struct bch_dev_usage_full *usage) { - ca->usage_base->d[BCH_DATA_free].buckets = ca->mi.nbuckets - ca->mi.first_bucket; -} + if (out->nr_tabstops < 5) { + printbuf_tabstops_reset(out); + printbuf_tabstop_push(out, 12); + printbuf_tabstop_push(out, 16); + printbuf_tabstop_push(out, 16); + printbuf_tabstop_push(out, 16); + printbuf_tabstop_push(out, 16); + } -void bch2_dev_usage_to_text(struct printbuf *out, struct bch_dev_usage *usage) -{ - prt_tab(out); - prt_str(out, "buckets"); - prt_tab_rjust(out); - prt_str(out, "sectors"); - prt_tab_rjust(out); - prt_str(out, "fragmented"); - prt_tab_rjust(out); - prt_newline(out); + prt_printf(out, "\tbuckets\rsectors\rfragmented\r\n"); for (unsigned i = 0; i < BCH_DATA_NR; i++) { bch2_prt_data_type(out, i); - prt_tab(out); - prt_u64(out, usage->d[i].buckets); - prt_tab_rjust(out); - prt_u64(out, usage->d[i].sectors); - prt_tab_rjust(out); - prt_u64(out, usage->d[i].fragmented); - prt_tab_rjust(out); - prt_newline(out); + prt_printf(out, "\t%llu\r%llu\r%llu\r\n", + usage->d[i].buckets, + usage->d[i].sectors, + usage->d[i].fragmented); } + + prt_printf(out, "capacity\t%llu\r\n", ca->mi.nbuckets); } -void bch2_dev_usage_update(struct bch_fs *c, struct bch_dev *ca, - const struct bch_alloc_v4 *old, - const struct bch_alloc_v4 *new, - u64 journal_seq, bool gc) +static int bch2_check_fix_ptr(struct btree_trans *trans, + struct bkey_s_c k, + struct extent_ptr_decoded p, + const union bch_extent_entry *entry, + bool *do_update) { - struct bch_fs_usage *fs_usage; - struct bch_dev_usage *u; - - preempt_disable(); - fs_usage = fs_usage_ptr(c, journal_seq, gc); + struct bch_fs *c = trans->c; + struct printbuf buf = PRINTBUF; + int ret = 0; - if (data_type_is_hidden(old->data_type)) - fs_usage->b.hidden -= ca->mi.bucket_size; - if (data_type_is_hidden(new->data_type)) - fs_usage->b.hidden += ca->mi.bucket_size; + struct bch_dev *ca = bch2_dev_tryget(c, p.ptr.dev); + if (!ca) { + if (fsck_err_on(p.ptr.dev != BCH_SB_MEMBER_INVALID, + trans, ptr_to_invalid_device, + "pointer to missing device %u\n" + "while marking %s", + p.ptr.dev, + (printbuf_reset(&buf), + bch2_bkey_val_to_text(&buf, c, k), buf.buf))) + *do_update = true; + return 0; + } - u = dev_usage_ptr(ca, journal_seq, gc); + struct bucket *g = PTR_GC_BUCKET(ca, &p.ptr); + if (!g) { + if (fsck_err(trans, ptr_to_invalid_device, + "pointer to invalid bucket on device %u\n" + "while marking %s", + p.ptr.dev, + (printbuf_reset(&buf), + bch2_bkey_val_to_text(&buf, c, k), buf.buf))) + *do_update = true; + goto out; + } - u->d[old->data_type].buckets--; - u->d[new->data_type].buckets++; + enum bch_data_type data_type = bch2_bkey_ptr_data_type(k, p, entry); - u->d[old->data_type].sectors -= bch2_bucket_sectors_dirty(*old); - u->d[new->data_type].sectors += bch2_bucket_sectors_dirty(*new); + if (fsck_err_on(!g->gen_valid, + trans, ptr_to_missing_alloc_key, + "bucket %u:%zu data type %s ptr gen %u missing in alloc btree\n" + "while marking %s", + p.ptr.dev, PTR_BUCKET_NR(ca, &p.ptr), + bch2_data_type_str(ptr_data_type(k.k, &p.ptr)), + p.ptr.gen, + (printbuf_reset(&buf), + bch2_bkey_val_to_text(&buf, c, k), buf.buf))) { + if (!p.ptr.cached) { + g->gen_valid = true; + g->gen = p.ptr.gen; + } else { + /* this pointer will be dropped */ + *do_update = true; + goto out; + } + } - u->d[BCH_DATA_cached].sectors += new->cached_sectors; - u->d[BCH_DATA_cached].sectors -= old->cached_sectors; + /* g->gen_valid == true */ - u->d[old->data_type].fragmented -= bch2_bucket_sectors_fragmented(ca, *old); - u->d[new->data_type].fragmented += bch2_bucket_sectors_fragmented(ca, *new); + if (fsck_err_on(gen_cmp(p.ptr.gen, g->gen) > 0, + trans, ptr_gen_newer_than_bucket_gen, + "bucket %u:%zu data type %s ptr gen in the future: %u > %u\n" + "while marking %s", + p.ptr.dev, PTR_BUCKET_NR(ca, &p.ptr), + bch2_data_type_str(ptr_data_type(k.k, &p.ptr)), + p.ptr.gen, g->gen, + (printbuf_reset(&buf), + bch2_bkey_val_to_text(&buf, c, k), buf.buf))) { + if (!p.ptr.cached && + (g->data_type != BCH_DATA_btree || + data_type == BCH_DATA_btree)) { + g->data_type = data_type; + g->stripe_sectors = 0; + g->dirty_sectors = 0; + g->cached_sectors = 0; + } - preempt_enable(); -} + *do_update = true; + } -static inline struct bch_alloc_v4 bucket_m_to_alloc(struct bucket b) -{ - return (struct bch_alloc_v4) { - .gen = b.gen, - .data_type = b.data_type, - .dirty_sectors = b.dirty_sectors, - .cached_sectors = b.cached_sectors, - .stripe = b.stripe, - }; -} + if (fsck_err_on(gen_cmp(g->gen, p.ptr.gen) > BUCKET_GC_GEN_MAX, + trans, ptr_gen_newer_than_bucket_gen, + "bucket %u:%zu gen %u data type %s: ptr gen %u too stale\n" + "while marking %s", + p.ptr.dev, PTR_BUCKET_NR(ca, &p.ptr), g->gen, + bch2_data_type_str(ptr_data_type(k.k, &p.ptr)), + p.ptr.gen, + (printbuf_reset(&buf), + bch2_bkey_val_to_text(&buf, c, k), buf.buf))) + *do_update = true; -void bch2_dev_usage_update_m(struct bch_fs *c, struct bch_dev *ca, - struct bucket *old, struct bucket *new) -{ - struct bch_alloc_v4 old_a = bucket_m_to_alloc(*old); - struct bch_alloc_v4 new_a = bucket_m_to_alloc(*new); + if (fsck_err_on(!p.ptr.cached && gen_cmp(p.ptr.gen, g->gen) < 0, + trans, stale_dirty_ptr, + "bucket %u:%zu data type %s stale dirty ptr: %u < %u\n" + "while marking %s", + p.ptr.dev, PTR_BUCKET_NR(ca, &p.ptr), + bch2_data_type_str(ptr_data_type(k.k, &p.ptr)), + p.ptr.gen, g->gen, + (printbuf_reset(&buf), + bch2_bkey_val_to_text(&buf, c, k), buf.buf))) + *do_update = true; - bch2_dev_usage_update(c, ca, &old_a, &new_a, 0, true); -} + if (data_type != BCH_DATA_btree && p.ptr.gen != g->gen) + goto out; -static inline int __update_replicas(struct bch_fs *c, - struct bch_fs_usage *fs_usage, - struct bch_replicas_entry_v1 *r, - s64 sectors) -{ - int idx = bch2_replicas_entry_idx(c, r); + if (fsck_err_on(bucket_data_type_mismatch(g->data_type, data_type), + trans, ptr_bucket_data_type_mismatch, + "bucket %u:%zu gen %u different types of data in same bucket: %s, %s\n" + "while marking %s", + p.ptr.dev, PTR_BUCKET_NR(ca, &p.ptr), g->gen, + bch2_data_type_str(g->data_type), + bch2_data_type_str(data_type), + (printbuf_reset(&buf), + bch2_bkey_val_to_text(&buf, c, k), buf.buf))) { + if (!p.ptr.cached && + data_type == BCH_DATA_btree) { + switch (g->data_type) { + case BCH_DATA_sb: + bch_err(c, "btree and superblock in the same bucket - cannot repair"); + ret = bch_err_throw(c, fsck_repair_unimplemented); + goto out; + case BCH_DATA_journal: + ret = bch2_dev_journal_bucket_delete(ca, PTR_BUCKET_NR(ca, &p.ptr)); + bch_err_msg(c, ret, "error deleting journal bucket %zu", + PTR_BUCKET_NR(ca, &p.ptr)); + if (ret) + goto out; + break; + } - if (idx < 0) - return -1; + g->data_type = data_type; + g->stripe_sectors = 0; + g->dirty_sectors = 0; + g->cached_sectors = 0; + } else { + *do_update = true; + } + } - fs_usage_data_type_to_base(&fs_usage->b, r->data_type, sectors); - fs_usage->replicas[idx] += sectors; - return 0; + if (p.has_ec) { + struct gc_stripe *m = genradix_ptr(&c->gc_stripes, p.ec.idx); + + if (fsck_err_on(!m || !m->alive, + trans, ptr_to_missing_stripe, + "pointer to nonexistent stripe %llu\n" + "while marking %s", + (u64) p.ec.idx, + (printbuf_reset(&buf), + bch2_bkey_val_to_text(&buf, c, k), buf.buf))) + *do_update = true; + + if (fsck_err_on(m && m->alive && !bch2_ptr_matches_stripe_m(m, p), + trans, ptr_to_incorrect_stripe, + "pointer does not match stripe %llu\n" + "while marking %s", + (u64) p.ec.idx, + (printbuf_reset(&buf), + bch2_bkey_val_to_text(&buf, c, k), buf.buf))) + *do_update = true; + } +out: +fsck_err: + bch2_dev_put(ca); + printbuf_exit(&buf); + return ret; } -int bch2_update_replicas(struct bch_fs *c, struct bkey_s_c k, - struct bch_replicas_entry_v1 *r, s64 sectors, - unsigned journal_seq, bool gc) +int bch2_check_fix_ptrs(struct btree_trans *trans, + enum btree_id btree, unsigned level, struct bkey_s_c k, + enum btree_iter_update_trigger_flags flags) { - struct bch_fs_usage *fs_usage; - int idx, ret = 0; + struct bch_fs *c = trans->c; + struct bkey_ptrs_c ptrs_c = bch2_bkey_ptrs_c(k); + const union bch_extent_entry *entry_c; + struct extent_ptr_decoded p = { 0 }; + bool do_update = false; struct printbuf buf = PRINTBUF; + int ret = 0; - percpu_down_read(&c->mark_lock); - - idx = bch2_replicas_entry_idx(c, r); - if (idx < 0 && - fsck_err(c, ptr_to_missing_replicas_entry, - "no replicas entry\n while marking %s", - (bch2_bkey_val_to_text(&buf, c, k), buf.buf))) { - percpu_up_read(&c->mark_lock); - ret = bch2_mark_replicas(c, r); - percpu_down_read(&c->mark_lock); + /* We don't yet do btree key updates correctly for when we're RW */ + BUG_ON(test_bit(BCH_FS_rw, &c->flags)); + bkey_for_each_ptr_decode(k.k, ptrs_c, p, entry_c) { + ret = bch2_check_fix_ptr(trans, k, p, entry_c, &do_update); if (ret) goto err; - idx = bch2_replicas_entry_idx(c, r); - } - if (idx < 0) { - ret = -1; - goto err; } - preempt_disable(); - fs_usage = fs_usage_ptr(c, journal_seq, gc); - fs_usage_data_type_to_base(&fs_usage->b, r->data_type, sectors); - fs_usage->replicas[idx] += sectors; - preempt_enable(); -err: -fsck_err: - percpu_up_read(&c->mark_lock); - printbuf_exit(&buf); - return ret; -} - -static inline int update_cached_sectors(struct bch_fs *c, - struct bkey_s_c k, - unsigned dev, s64 sectors, - unsigned journal_seq, bool gc) -{ - struct bch_replicas_padded r; - - bch2_replicas_entry_cached(&r.e, dev); + if (do_update) { + struct bkey_i *new = bch2_bkey_make_mut_noupdate(trans, k); + ret = PTR_ERR_OR_ZERO(new); + if (ret) + goto err; - return bch2_update_replicas(c, k, &r.e, sectors, journal_seq, gc); -} + scoped_guard(rcu) + bch2_bkey_drop_ptrs(bkey_i_to_s(new), ptr, !bch2_dev_exists(c, ptr->dev)); -static int __replicas_deltas_realloc(struct btree_trans *trans, unsigned more, - gfp_t gfp) -{ - struct replicas_delta_list *d = trans->fs_usage_deltas; - unsigned new_size = d ? (d->size + more) * 2 : 128; - unsigned alloc_size = sizeof(*d) + new_size; + if (level) { + /* + * We don't want to drop btree node pointers - if the + * btree node isn't there anymore, the read path will + * sort it out: + */ + struct bkey_ptrs ptrs = bch2_bkey_ptrs(bkey_i_to_s(new)); + scoped_guard(rcu) + bkey_for_each_ptr(ptrs, ptr) { + struct bch_dev *ca = bch2_dev_rcu(c, ptr->dev); + ptr->gen = PTR_GC_BUCKET(ca, ptr)->gen; + } + } else { + struct bkey_ptrs ptrs; + union bch_extent_entry *entry; + + rcu_read_lock(); +restart_drop_ptrs: + ptrs = bch2_bkey_ptrs(bkey_i_to_s(new)); + bkey_for_each_ptr_decode(bkey_i_to_s(new).k, ptrs, p, entry) { + struct bch_dev *ca = bch2_dev_rcu(c, p.ptr.dev); + struct bucket *g = PTR_GC_BUCKET(ca, &p.ptr); + enum bch_data_type data_type = bch2_bkey_ptr_data_type(bkey_i_to_s_c(new), p, entry); + + if ((p.ptr.cached && + (!g->gen_valid || gen_cmp(p.ptr.gen, g->gen) > 0)) || + (!p.ptr.cached && + gen_cmp(p.ptr.gen, g->gen) < 0) || + gen_cmp(g->gen, p.ptr.gen) > BUCKET_GC_GEN_MAX || + (g->data_type && + g->data_type != data_type)) { + bch2_bkey_drop_ptr(bkey_i_to_s(new), &entry->ptr); + goto restart_drop_ptrs; + } + } + rcu_read_unlock(); +again: + ptrs = bch2_bkey_ptrs(bkey_i_to_s(new)); + bkey_extent_entry_for_each(ptrs, entry) { + if (extent_entry_type(entry) == BCH_EXTENT_ENTRY_stripe_ptr) { + struct gc_stripe *m = genradix_ptr(&c->gc_stripes, + entry->stripe_ptr.idx); + union bch_extent_entry *next_ptr; + + bkey_extent_entry_for_each_from(ptrs, next_ptr, entry) + if (extent_entry_type(next_ptr) == BCH_EXTENT_ENTRY_ptr) + goto found; + next_ptr = NULL; +found: + if (!next_ptr) { + bch_err(c, "aieee, found stripe ptr with no data ptr"); + continue; + } + + if (!m || !m->alive || + !__bch2_ptr_matches_stripe(&m->ptrs[entry->stripe_ptr.block], + &next_ptr->ptr, + m->sectors)) { + bch2_bkey_extent_entry_drop(new, entry); + goto again; + } + } + } + } - WARN_ON_ONCE(alloc_size > REPLICAS_DELTA_LIST_MAX); + if (0) { + printbuf_reset(&buf); + bch2_bkey_val_to_text(&buf, c, k); + bch_info(c, "updated %s", buf.buf); - if (!d || d->used + more > d->size) { - d = krealloc(d, alloc_size, gfp|__GFP_ZERO); + printbuf_reset(&buf); + bch2_bkey_val_to_text(&buf, c, bkey_i_to_s_c(new)); + bch_info(c, "new key %s", buf.buf); + } - if (unlikely(!d)) { - if (alloc_size > REPLICAS_DELTA_LIST_MAX) - return -ENOMEM; + if (!(flags & BTREE_TRIGGER_is_root)) { + struct btree_iter iter; + bch2_trans_node_iter_init(trans, &iter, btree, new->k.p, 0, level, + BTREE_ITER_intent|BTREE_ITER_all_snapshots); + ret = bch2_btree_iter_traverse(trans, &iter) ?: + bch2_trans_update(trans, &iter, new, + BTREE_UPDATE_internal_snapshot_node| + BTREE_TRIGGER_norun); + bch2_trans_iter_exit(trans, &iter); + if (ret) + goto err; - d = mempool_alloc(&trans->c->replicas_delta_pool, gfp); - if (!d) - return -ENOMEM; + if (level) + bch2_btree_node_update_key_early(trans, btree, level - 1, k, new); + } else { + struct jset_entry *e = bch2_trans_jset_entry_alloc(trans, + jset_u64s(new->k.u64s)); + ret = PTR_ERR_OR_ZERO(e); + if (ret) + goto err; - memset(d, 0, REPLICAS_DELTA_LIST_MAX); + journal_entry_set(e, + BCH_JSET_ENTRY_btree_root, + btree, level - 1, + new, new->k.u64s); - if (trans->fs_usage_deltas) - memcpy(d, trans->fs_usage_deltas, - trans->fs_usage_deltas->size + sizeof(*d)); + /* + * no locking, we're single threaded and not rw yet, see + * the big assertino above that we repeat here: + */ + BUG_ON(test_bit(BCH_FS_rw, &c->flags)); - new_size = REPLICAS_DELTA_LIST_MAX - sizeof(*d); - kfree(trans->fs_usage_deltas); + struct btree *b = bch2_btree_id_root(c, btree)->b; + bkey_copy(&b->key, new); } - - d->size = new_size; - trans->fs_usage_deltas = d; } - - return 0; -} - -int bch2_replicas_deltas_realloc(struct btree_trans *trans, unsigned more) -{ - return allocate_dropping_locks_errcode(trans, - __replicas_deltas_realloc(trans, more, _gfp)); -} - -int bch2_update_replicas_list(struct btree_trans *trans, - struct bch_replicas_entry_v1 *r, - s64 sectors) -{ - struct replicas_delta_list *d; - struct replicas_delta *n; - unsigned b; - int ret; - - if (!sectors) - return 0; - - b = replicas_entry_bytes(r) + 8; - ret = bch2_replicas_deltas_realloc(trans, b); - if (ret) - return ret; - - d = trans->fs_usage_deltas; - n = (void *) d->d + d->used; - n->delta = sectors; - unsafe_memcpy((void *) n + offsetof(struct replicas_delta, r), - r, replicas_entry_bytes(r), - "flexible array member embedded in strcuct with padding"); - bch2_replicas_entry_sort(&n->r); - d->used += b; - return 0; -} - -int bch2_update_cached_sectors_list(struct btree_trans *trans, unsigned dev, s64 sectors) -{ - struct bch_replicas_padded r; - - bch2_replicas_entry_cached(&r.e, dev); - - return bch2_update_replicas_list(trans, &r.e, sectors); +err: + printbuf_exit(&buf); + return ret; } -int bch2_mark_metadata_bucket(struct bch_fs *c, struct bch_dev *ca, - size_t b, enum bch_data_type data_type, - unsigned sectors, struct gc_pos pos, - unsigned flags) +static int bucket_ref_update_err(struct btree_trans *trans, struct printbuf *buf, + struct bkey_s_c k, bool insert, enum bch_sb_error_id id) { - struct bucket old, new, *g; - int ret = 0; - - BUG_ON(!(flags & BTREE_TRIGGER_GC)); - BUG_ON(data_type != BCH_DATA_sb && - data_type != BCH_DATA_journal); - - /* - * Backup superblock might be past the end of our normal usable space: - */ - if (b >= ca->mi.nbuckets) - return 0; + struct bch_fs *c = trans->c; - percpu_down_read(&c->mark_lock); - g = gc_bucket(ca, b); + prt_printf(buf, "\nwhile marking "); + bch2_bkey_val_to_text(buf, c, k); + prt_newline(buf); - bucket_lock(g); - old = *g; + bool print = __bch2_count_fsck_err(c, id, buf); - if (bch2_fs_inconsistent_on(g->data_type && - g->data_type != data_type, c, - "different types of data in same bucket: %s, %s", - bch2_data_type_str(g->data_type), - bch2_data_type_str(data_type))) { - ret = -EIO; - goto err; - } + int ret = bch2_run_explicit_recovery_pass(c, buf, + BCH_RECOVERY_PASS_check_allocations, 0); - if (bch2_fs_inconsistent_on((u64) g->dirty_sectors + sectors > ca->mi.bucket_size, c, - "bucket %u:%zu gen %u data type %s sector count overflow: %u + %u > bucket size", - ca->dev_idx, b, g->gen, - bch2_data_type_str(g->data_type ?: data_type), - g->dirty_sectors, sectors)) { - ret = -EIO; - goto err; + if (insert) { + bch2_trans_updates_to_text(buf, trans); + __bch2_inconsistent_error(c, buf); + /* + * If we're in recovery, run_explicit_recovery_pass might give + * us an error code for rewinding recovery + */ + if (!ret) + ret = bch_err_throw(c, bucket_ref_update); + } else { + /* Always ignore overwrite errors, so that deletion works */ + ret = 0; } - g->data_type = data_type; - g->dirty_sectors += sectors; - new = *g; -err: - bucket_unlock(g); - if (!ret) - bch2_dev_usage_update_m(c, ca, &old, &new); - percpu_up_read(&c->mark_lock); + if (print || insert) + bch2_print_str(c, KERN_ERR, buf->buf); return ret; } -int bch2_check_bucket_ref(struct btree_trans *trans, - struct bkey_s_c k, - const struct bch_extent_ptr *ptr, - s64 sectors, enum bch_data_type ptr_data_type, - u8 b_gen, u8 bucket_data_type, - u32 bucket_sectors) +int bch2_bucket_ref_update(struct btree_trans *trans, struct bch_dev *ca, + struct bkey_s_c k, + const struct bch_extent_ptr *ptr, + s64 sectors, enum bch_data_type ptr_data_type, + u8 b_gen, u8 bucket_data_type, + u32 *bucket_sectors) { struct bch_fs *c = trans->c; - struct bch_dev *ca = bch_dev_bkey_exists(c, ptr->dev); size_t bucket_nr = PTR_BUCKET_NR(ca, ptr); struct printbuf buf = PRINTBUF; + bool inserting = sectors > 0; int ret = 0; - if (bucket_data_type == BCH_DATA_cached) - bucket_data_type = BCH_DATA_user; + BUG_ON(!sectors); - if ((bucket_data_type == BCH_DATA_stripe && ptr_data_type == BCH_DATA_user) || - (bucket_data_type == BCH_DATA_user && ptr_data_type == BCH_DATA_stripe)) - bucket_data_type = ptr_data_type = BCH_DATA_stripe; - - if (gen_after(ptr->gen, b_gen)) { - bch2_fsck_err(c, FSCK_CAN_IGNORE|FSCK_NEED_FSCK, - BCH_FSCK_ERR_ptr_gen_newer_than_bucket_gen, - "bucket %u:%zu gen %u data type %s: ptr gen %u newer than bucket gen\n" - "while marking %s", + if (unlikely(gen_after(ptr->gen, b_gen))) { + bch2_log_msg_start(c, &buf); + prt_printf(&buf, + "bucket %u:%zu gen %u data type %s: ptr gen %u newer than bucket gen", ptr->dev, bucket_nr, b_gen, bch2_data_type_str(bucket_data_type ?: ptr_data_type), - ptr->gen, - (bch2_bkey_val_to_text(&buf, c, k), buf.buf)); - ret = -EIO; - goto err; - } + ptr->gen); - if (gen_cmp(b_gen, ptr->gen) > BUCKET_GC_GEN_MAX) { - bch2_fsck_err(c, FSCK_CAN_IGNORE|FSCK_NEED_FSCK, - BCH_FSCK_ERR_ptr_too_stale, - "bucket %u:%zu gen %u data type %s: ptr gen %u too stale\n" - "while marking %s", - ptr->dev, bucket_nr, b_gen, - bch2_data_type_str(bucket_data_type ?: ptr_data_type), - ptr->gen, - (printbuf_reset(&buf), - bch2_bkey_val_to_text(&buf, c, k), buf.buf)); - ret = -EIO; - goto err; + ret = bucket_ref_update_err(trans, &buf, k, inserting, + BCH_FSCK_ERR_ptr_gen_newer_than_bucket_gen); + goto out; } - if (b_gen != ptr->gen && !ptr->cached) { - bch2_fsck_err(c, FSCK_CAN_IGNORE|FSCK_NEED_FSCK, - BCH_FSCK_ERR_stale_dirty_ptr, - "bucket %u:%zu gen %u (mem gen %u) data type %s: stale dirty ptr (gen %u)\n" - "while marking %s", + if (unlikely(gen_cmp(b_gen, ptr->gen) > BUCKET_GC_GEN_MAX)) { + bch2_log_msg_start(c, &buf); + prt_printf(&buf, + "bucket %u:%zu gen %u data type %s: ptr gen %u too stale", ptr->dev, bucket_nr, b_gen, - *bucket_gen(ca, bucket_nr), bch2_data_type_str(bucket_data_type ?: ptr_data_type), - ptr->gen, - (printbuf_reset(&buf), - bch2_bkey_val_to_text(&buf, c, k), buf.buf)); - ret = -EIO; - goto err; - } + ptr->gen); - if (b_gen != ptr->gen) { - ret = 1; + ret = bucket_ref_update_err(trans, &buf, k, inserting, + BCH_FSCK_ERR_ptr_too_stale); goto out; } - if (!data_type_is_empty(bucket_data_type) && - ptr_data_type && - bucket_data_type != ptr_data_type) { - bch2_fsck_err(c, FSCK_CAN_IGNORE|FSCK_NEED_FSCK, - BCH_FSCK_ERR_ptr_bucket_data_type_mismatch, - "bucket %u:%zu gen %u different types of data in same bucket: %s, %s\n" - "while marking %s", - ptr->dev, bucket_nr, b_gen, - bch2_data_type_str(bucket_data_type), - bch2_data_type_str(ptr_data_type), - (printbuf_reset(&buf), - bch2_bkey_val_to_text(&buf, c, k), buf.buf)); - ret = -EIO; - goto err; + if (b_gen != ptr->gen && ptr->cached) { + ret = 1; + goto out; } - if ((u64) bucket_sectors + sectors > U32_MAX) { - bch2_fsck_err(c, FSCK_CAN_IGNORE|FSCK_NEED_FSCK, - BCH_FSCK_ERR_bucket_sector_count_overflow, - "bucket %u:%zu gen %u data type %s sector count overflow: %u + %lli > U32_MAX\n" - "while marking %s", + if (unlikely(b_gen != ptr->gen)) { + bch2_log_msg_start(c, &buf); + prt_printf(&buf, + "bucket %u:%zu gen %u (mem gen %u) data type %s: stale dirty ptr (gen %u)", ptr->dev, bucket_nr, b_gen, + bucket_gen_get(ca, bucket_nr), bch2_data_type_str(bucket_data_type ?: ptr_data_type), - bucket_sectors, sectors, - (printbuf_reset(&buf), - bch2_bkey_val_to_text(&buf, c, k), buf.buf)); - ret = -EIO; - goto err; - } -out: - printbuf_exit(&buf); - return ret; -err: - bch2_dump_trans_updates(trans); - goto out; -} + ptr->gen); -void bch2_trans_fs_usage_revert(struct btree_trans *trans, - struct replicas_delta_list *deltas) -{ - struct bch_fs *c = trans->c; - struct bch_fs_usage *dst; - struct replicas_delta *d, *top = (void *) deltas->d + deltas->used; - s64 added = 0; - unsigned i; - - percpu_down_read(&c->mark_lock); - preempt_disable(); - dst = fs_usage_ptr(c, trans->journal_res.seq, false); - - /* revert changes: */ - for (d = deltas->d; d != top; d = replicas_delta_next(d)) { - switch (d->r.data_type) { - case BCH_DATA_btree: - case BCH_DATA_user: - case BCH_DATA_parity: - added += d->delta; - } - BUG_ON(__update_replicas(c, dst, &d->r, -d->delta)); + ret = bucket_ref_update_err(trans, &buf, k, inserting, + BCH_FSCK_ERR_stale_dirty_ptr); + goto out; } - dst->b.nr_inodes -= deltas->nr_inodes; + if (unlikely(bucket_data_type_mismatch(bucket_data_type, ptr_data_type))) { + bch2_log_msg_start(c, &buf); + prt_printf(&buf, "bucket %u:%zu gen %u different types of data in same bucket: %s, %s", + ptr->dev, bucket_nr, b_gen, + bch2_data_type_str(bucket_data_type), + bch2_data_type_str(ptr_data_type)); - for (i = 0; i < BCH_REPLICAS_MAX; i++) { - added -= deltas->persistent_reserved[i]; - dst->b.reserved -= deltas->persistent_reserved[i]; - dst->persistent_reserved[i] -= deltas->persistent_reserved[i]; + ret = bucket_ref_update_err(trans, &buf, k, inserting, + BCH_FSCK_ERR_ptr_bucket_data_type_mismatch); + goto out; } - if (added > 0) { - trans->disk_res->sectors += added; - this_cpu_add(*c->online_reserved, added); + if (unlikely((u64) *bucket_sectors + sectors > U32_MAX)) { + bch2_log_msg_start(c, &buf); + prt_printf(&buf, + "bucket %u:%zu gen %u data type %s sector count overflow: %u + %lli > U32_MAX", + ptr->dev, bucket_nr, b_gen, + bch2_data_type_str(bucket_data_type ?: ptr_data_type), + *bucket_sectors, sectors); + + ret = bucket_ref_update_err(trans, &buf, k, inserting, + BCH_FSCK_ERR_bucket_sector_count_overflow); + sectors = -*bucket_sectors; + goto out; } - preempt_enable(); - percpu_up_read(&c->mark_lock); + *bucket_sectors += sectors; +out: + printbuf_exit(&buf); + return ret; } void bch2_trans_account_disk_usage_change(struct btree_trans *trans) @@ -702,8 +551,6 @@ void bch2_trans_account_disk_usage_change(struct btree_trans *trans) bool warn = false; percpu_down_read(&c->mark_lock); - preempt_disable(); - struct bch_fs_usage_base *dst = &fs_usage_ptr(c, trans->journal_res.seq, false)->b; struct bch_fs_usage_base *src = &trans->fs_usage_delta; s64 added = src->btree + src->data + src->reserved; @@ -714,13 +561,13 @@ void bch2_trans_account_disk_usage_change(struct btree_trans *trans) */ s64 should_not_have_added = added - (s64) disk_res_sectors; if (unlikely(should_not_have_added > 0)) { - u64 old, new, v = atomic64_read(&c->sectors_available); + u64 old, new; + old = atomic64_read(&c->sectors_available); do { - old = v; new = max_t(s64, 0, old - should_not_have_added); - } while ((v = atomic64_cmpxchg(&c->sectors_available, - old, new)) != old); + } while (!atomic64_try_cmpxchg(&c->sectors_available, + &old, new)); added -= should_not_have_added; warn = true; @@ -731,13 +578,9 @@ void bch2_trans_account_disk_usage_change(struct btree_trans *trans) this_cpu_sub(*c->online_reserved, added); } - dst->hidden += src->hidden; - dst->btree += src->btree; - dst->data += src->data; - dst->cached += src->cached; - dst->reserved += src->reserved; - dst->nr_inodes += src->nr_inodes; - + preempt_disable(); + struct bch_fs_usage_base *dst = this_cpu_ptr(c->usage); + acc_u64s((u64 *) dst, (u64 *) src, sizeof(*src) / sizeof(u64)); preempt_enable(); percpu_up_read(&c->mark_lock); @@ -747,150 +590,111 @@ void bch2_trans_account_disk_usage_change(struct btree_trans *trans) should_not_have_added, disk_res_sectors); } -int bch2_trans_fs_usage_apply(struct btree_trans *trans, - struct replicas_delta_list *deltas) -{ - struct bch_fs *c = trans->c; - struct replicas_delta *d, *d2; - struct replicas_delta *top = (void *) deltas->d + deltas->used; - struct bch_fs_usage *dst; - unsigned i; - - percpu_down_read(&c->mark_lock); - preempt_disable(); - dst = fs_usage_ptr(c, trans->journal_res.seq, false); - - for (d = deltas->d; d != top; d = replicas_delta_next(d)) - if (__update_replicas(c, dst, &d->r, d->delta)) - goto need_mark; - - dst->b.nr_inodes += deltas->nr_inodes; - - for (i = 0; i < BCH_REPLICAS_MAX; i++) { - dst->b.reserved += deltas->persistent_reserved[i]; - dst->persistent_reserved[i] += deltas->persistent_reserved[i]; - } - - preempt_enable(); - percpu_up_read(&c->mark_lock); - return 0; -need_mark: - /* revert changes: */ - for (d2 = deltas->d; d2 != d; d2 = replicas_delta_next(d2)) - BUG_ON(__update_replicas(c, dst, &d2->r, -d2->delta)); - - preempt_enable(); - percpu_up_read(&c->mark_lock); - return -1; -} - /* KEY_TYPE_extent: */ -static int __mark_pointer(struct btree_trans *trans, +static int __mark_pointer(struct btree_trans *trans, struct bch_dev *ca, struct bkey_s_c k, - const struct bch_extent_ptr *ptr, + const struct extent_ptr_decoded *p, s64 sectors, enum bch_data_type ptr_data_type, - u8 bucket_gen, u8 *bucket_data_type, - u32 *dirty_sectors, u32 *cached_sectors) + struct bch_alloc_v4 *a, + bool insert) { - u32 *dst_sectors = !ptr->cached - ? dirty_sectors - : cached_sectors; - int ret = bch2_check_bucket_ref(trans, k, ptr, sectors, ptr_data_type, - bucket_gen, *bucket_data_type, *dst_sectors); + u32 *dst_sectors = p->has_ec ? &a->stripe_sectors : + !p->ptr.cached ? &a->dirty_sectors : + &a->cached_sectors; + int ret = bch2_bucket_ref_update(trans, ca, k, &p->ptr, sectors, ptr_data_type, + a->gen, a->data_type, dst_sectors); if (ret) return ret; - - *dst_sectors += sectors; - - if (!*dirty_sectors && !*cached_sectors) - *bucket_data_type = 0; - else if (*bucket_data_type != BCH_DATA_stripe) - *bucket_data_type = ptr_data_type; - + if (insert) + alloc_data_type_set(a, ptr_data_type); return 0; } static int bch2_trigger_pointer(struct btree_trans *trans, enum btree_id btree_id, unsigned level, struct bkey_s_c k, struct extent_ptr_decoded p, + const union bch_extent_entry *entry, s64 *sectors, - unsigned flags) + enum btree_iter_update_trigger_flags flags) { - bool insert = !(flags & BTREE_TRIGGER_OVERWRITE); - struct bpos bucket; - struct bch_backpointer bp; - - bch2_extent_ptr_to_bp(trans->c, btree_id, level, k, p, &bucket, &bp); - *sectors = insert ? bp.bucket_len : -((s64) bp.bucket_len); + struct bch_fs *c = trans->c; + bool insert = !(flags & BTREE_TRIGGER_overwrite); + struct printbuf buf = PRINTBUF; + int ret = 0; - if (flags & BTREE_TRIGGER_TRANSACTIONAL) { - struct btree_iter iter; - struct bkey_i_alloc_v4 *a = bch2_trans_start_alloc_update(trans, &iter, bucket); - int ret = PTR_ERR_OR_ZERO(a); - if (ret) - return ret; + struct bkey_i_backpointer bp; + bch2_extent_ptr_to_bp(c, btree_id, level, k, p, entry, &bp); - ret = __mark_pointer(trans, k, &p.ptr, *sectors, bp.data_type, - a->v.gen, &a->v.data_type, - &a->v.dirty_sectors, &a->v.cached_sectors) ?: - bch2_trans_update(trans, &iter, &a->k_i, 0); - bch2_trans_iter_exit(trans, &iter); + *sectors = insert ? bp.v.bucket_len : -(s64) bp.v.bucket_len; - if (ret) - return ret; + struct bch_dev *ca = bch2_dev_tryget(c, p.ptr.dev); + if (unlikely(!ca)) { + if (insert && p.ptr.dev != BCH_SB_MEMBER_INVALID) + ret = bch_err_throw(c, trigger_pointer); + goto err; + } - if (!p.ptr.cached) { - ret = bch2_bucket_backpointer_mod(trans, bucket, bp, k, insert); - if (ret) - return ret; + struct bpos bucket = PTR_BUCKET_POS(ca, &p.ptr); + if (!bucket_valid(ca, bucket.offset)) { + if (insert) { + bch2_dev_bucket_missing(ca, bucket.offset); + ret = bch_err_throw(c, trigger_pointer); } + goto err; } - if (flags & BTREE_TRIGGER_GC) { - struct bch_fs *c = trans->c; - struct bch_dev *ca = bch_dev_bkey_exists(c, p.ptr.dev); - enum bch_data_type data_type = bkey_ptr_data_type(btree_id, level, k, p); + if (flags & BTREE_TRIGGER_transactional) { + struct bkey_i_alloc_v4 *a = bch2_trans_start_alloc_update(trans, bucket, 0); + ret = PTR_ERR_OR_ZERO(a) ?: + __mark_pointer(trans, ca, k, &p, *sectors, bp.v.data_type, &a->v, insert); + if (ret) + goto err; - percpu_down_read(&c->mark_lock); - struct bucket *g = PTR_GC_BUCKET(ca, &p.ptr); - bucket_lock(g); - struct bucket old = *g; - - u8 bucket_data_type = g->data_type; - int ret = __mark_pointer(trans, k, &p.ptr, *sectors, - data_type, g->gen, - &bucket_data_type, - &g->dirty_sectors, - &g->cached_sectors); - if (ret) { - bucket_unlock(g); - percpu_up_read(&c->mark_lock); - return ret; + ret = bch2_bucket_backpointer_mod(trans, k, &bp, insert); + if (ret) + goto err; + } + + if (flags & BTREE_TRIGGER_gc) { + struct bucket *g = gc_bucket(ca, bucket.offset); + if (bch2_fs_inconsistent_on(!g, c, "reference to invalid bucket on device %u\n %s", + p.ptr.dev, + (bch2_bkey_val_to_text(&buf, c, k), buf.buf))) { + ret = bch_err_throw(c, trigger_pointer); + goto err; } - g->data_type = bucket_data_type; - struct bucket new = *g; + bucket_lock(g); + struct bch_alloc_v4 old = bucket_m_to_alloc(*g), new = old; + ret = __mark_pointer(trans, ca, k, &p, *sectors, bp.v.data_type, &new, insert); + alloc_to_bucket(g, new); bucket_unlock(g); - bch2_dev_usage_update_m(c, ca, &old, &new); - percpu_up_read(&c->mark_lock); - } - return 0; + if (!ret) + ret = bch2_alloc_key_to_dev_counters(trans, ca, &old, &new, flags); + } +err: + bch2_dev_put(ca); + printbuf_exit(&buf); + return ret; } static int bch2_trigger_stripe_ptr(struct btree_trans *trans, struct bkey_s_c k, struct extent_ptr_decoded p, enum bch_data_type data_type, - s64 sectors, unsigned flags) + s64 sectors, + enum btree_iter_update_trigger_flags flags) { - if (flags & BTREE_TRIGGER_TRANSACTIONAL) { + struct bch_fs *c = trans->c; + + if (flags & BTREE_TRIGGER_transactional) { struct btree_iter iter; struct bkey_i_stripe *s = bch2_bkey_get_mut_typed(trans, &iter, BTREE_ID_stripes, POS(0, p.ec.idx), - BTREE_ITER_WITH_UPDATES, stripe); + BTREE_ITER_with_updates, stripe); int ret = PTR_ERR_OR_ZERO(s); if (unlikely(ret)) { bch2_trans_inconsistent_on(bch2_err_matches(ret, ENOENT), trans, @@ -903,7 +707,7 @@ static int bch2_trigger_stripe_ptr(struct btree_trans *trans, bch2_trans_inconsistent(trans, "stripe pointer doesn't match stripe %llu", (u64) p.ec.idx); - ret = -EIO; + ret = bch_err_throw(c, trigger_stripe_pointer); goto err; } @@ -911,47 +715,52 @@ static int bch2_trigger_stripe_ptr(struct btree_trans *trans, stripe_blockcount_get(&s->v, p.ec.block) + sectors); - struct bch_replicas_padded r; - bch2_bkey_to_replicas(&r.e, bkey_i_to_s_c(&s->k_i)); - r.e.data_type = data_type; - ret = bch2_update_replicas_list(trans, &r.e, sectors); + struct disk_accounting_pos acc; + memset(&acc, 0, sizeof(acc)); + acc.type = BCH_DISK_ACCOUNTING_replicas; + bch2_bkey_to_replicas(&acc.replicas, bkey_i_to_s_c(&s->k_i)); + acc.replicas.data_type = data_type; + ret = bch2_disk_accounting_mod(trans, &acc, §ors, 1, false); err: bch2_trans_iter_exit(trans, &iter); return ret; } - if (flags & BTREE_TRIGGER_GC) { - struct bch_fs *c = trans->c; - - BUG_ON(!(flags & BTREE_TRIGGER_GC)); - + if (flags & BTREE_TRIGGER_gc) { struct gc_stripe *m = genradix_ptr_alloc(&c->gc_stripes, p.ec.idx, GFP_KERNEL); if (!m) { bch_err(c, "error allocating memory for gc_stripes, idx %llu", (u64) p.ec.idx); - return -BCH_ERR_ENOMEM_mark_stripe_ptr; + return bch_err_throw(c, ENOMEM_mark_stripe_ptr); } - mutex_lock(&c->ec_stripes_heap_lock); + gc_stripe_lock(m); if (!m || !m->alive) { - mutex_unlock(&c->ec_stripes_heap_lock); + gc_stripe_unlock(m); struct printbuf buf = PRINTBUF; + bch2_log_msg_start(c, &buf); + prt_printf(&buf, "pointer to nonexistent stripe %llu\n while marking ", + (u64) p.ec.idx); bch2_bkey_val_to_text(&buf, c, k); - bch_err_ratelimited(c, "pointer to nonexistent stripe %llu\n while marking %s", - (u64) p.ec.idx, buf.buf); + __bch2_inconsistent_error(c, &buf); + bch2_print_str(c, KERN_ERR, buf.buf); printbuf_exit(&buf); - bch2_inconsistent_error(c); - return -EIO; + return bch_err_throw(c, trigger_stripe_pointer); } m->block_sectors[p.ec.block] += sectors; - struct bch_replicas_padded r = m->r; - mutex_unlock(&c->ec_stripes_heap_lock); + struct disk_accounting_pos acc; + memset(&acc, 0, sizeof(acc)); + acc.type = BCH_DISK_ACCOUNTING_replicas; + unsafe_memcpy(&acc.replicas, &m->r.e, replicas_entry_bytes(&m->r.e), "VLA"); + gc_stripe_unlock(m); - r.e.data_type = data_type; - bch2_update_replicas(c, k, &r.e, sectors, trans->journal_res.seq, true); + acc.replicas.data_type = data_type; + int ret = bch2_disk_accounting_mod(trans, &acc, §ors, 1, true); + if (ret) + return ret; } return 0; @@ -959,45 +768,48 @@ err: static int __trigger_extent(struct btree_trans *trans, enum btree_id btree_id, unsigned level, - struct bkey_s_c k, unsigned flags) + struct bkey_s_c k, + enum btree_iter_update_trigger_flags flags) { - bool gc = flags & BTREE_TRIGGER_GC; - struct bch_fs *c = trans->c; + bool gc = flags & BTREE_TRIGGER_gc; struct bkey_ptrs_c ptrs = bch2_bkey_ptrs_c(k); const union bch_extent_entry *entry; struct extent_ptr_decoded p; - struct bch_replicas_padded r; enum bch_data_type data_type = bkey_is_btree_ptr(k.k) ? BCH_DATA_btree : BCH_DATA_user; - s64 dirty_sectors = 0; int ret = 0; - r.e.data_type = data_type; - r.e.nr_devs = 0; - r.e.nr_required = 1; + s64 replicas_sectors = 0; + + struct disk_accounting_pos acc_replicas_key; + memset(&acc_replicas_key, 0, sizeof(acc_replicas_key)); + acc_replicas_key.type = BCH_DISK_ACCOUNTING_replicas; + acc_replicas_key.replicas.data_type = data_type; + acc_replicas_key.replicas.nr_devs = 0; + acc_replicas_key.replicas.nr_required = 1; + + unsigned cur_compression_type = 0; + u64 compression_acct[3] = { 1, 0, 0 }; bkey_for_each_ptr_decode(k.k, ptrs, p, entry) { - s64 disk_sectors; - ret = bch2_trigger_pointer(trans, btree_id, level, k, p, &disk_sectors, flags); + s64 disk_sectors = 0; + ret = bch2_trigger_pointer(trans, btree_id, level, k, p, entry, &disk_sectors, flags); if (ret < 0) return ret; bool stale = ret > 0; + if (p.ptr.cached && stale) + continue; + if (p.ptr.cached) { - if (!stale) { - ret = !gc - ? bch2_update_cached_sectors_list(trans, p.ptr.dev, disk_sectors) - : update_cached_sectors(c, k, p.ptr.dev, disk_sectors, 0, true); - bch2_fs_fatal_err_on(ret && gc, c, "%s: no replicas entry while updating cached sectors", - bch2_err_str(ret)); - if (ret) - return ret; - } + ret = bch2_mod_dev_cached_sectors(trans, p.ptr.dev, disk_sectors, gc); + if (ret) + return ret; } else if (!p.has_ec) { - dirty_sectors += disk_sectors; - r.e.devs[r.e.nr_devs++] = p.ptr.dev; + replicas_sectors += disk_sectors; + replicas_entry_add_dev(&acc_replicas_key.replicas, p.ptr.dev); } else { ret = bch2_trigger_stripe_ptr(trans, k, p, data_type, disk_sectors, flags); if (ret) @@ -1008,21 +820,66 @@ static int __trigger_extent(struct btree_trans *trans, * if so they're not required for mounting if we have an * erasure coded pointer in this extent: */ - r.e.nr_required = 0; + acc_replicas_key.replicas.nr_required = 0; } - } - if (r.e.nr_devs) { - ret = !gc - ? bch2_update_replicas_list(trans, &r.e, dirty_sectors) - : bch2_update_replicas(c, k, &r.e, dirty_sectors, 0, true); - if (unlikely(ret && gc)) { - struct printbuf buf = PRINTBUF; + if (cur_compression_type && + cur_compression_type != p.crc.compression_type) { + if (flags & BTREE_TRIGGER_overwrite) + bch2_u64s_neg(compression_acct, ARRAY_SIZE(compression_acct)); - bch2_bkey_val_to_text(&buf, c, k); - bch2_fs_fatal_error(c, ": no replicas entry for %s", buf.buf); - printbuf_exit(&buf); + ret = bch2_disk_accounting_mod2(trans, gc, compression_acct, + compression, cur_compression_type); + if (ret) + return ret; + + compression_acct[0] = 1; + compression_acct[1] = 0; + compression_acct[2] = 0; + } + + cur_compression_type = p.crc.compression_type; + if (p.crc.compression_type) { + compression_acct[1] += p.crc.uncompressed_size; + compression_acct[2] += p.crc.compressed_size; } + } + + if (acc_replicas_key.replicas.nr_devs) { + ret = bch2_disk_accounting_mod(trans, &acc_replicas_key, &replicas_sectors, 1, gc); + if (ret) + return ret; + } + + if (acc_replicas_key.replicas.nr_devs && !level && k.k->p.snapshot) { + ret = bch2_disk_accounting_mod2_nr(trans, gc, &replicas_sectors, 1, snapshot, k.k->p.snapshot); + if (ret) + return ret; + } + + if (cur_compression_type) { + if (flags & BTREE_TRIGGER_overwrite) + bch2_u64s_neg(compression_acct, ARRAY_SIZE(compression_acct)); + + ret = bch2_disk_accounting_mod2(trans, gc, compression_acct, + compression, cur_compression_type); + if (ret) + return ret; + } + + if (level) { + ret = bch2_disk_accounting_mod2_nr(trans, gc, &replicas_sectors, 1, btree, btree_id); + if (ret) + return ret; + } else { + bool insert = !(flags & BTREE_TRIGGER_overwrite); + + s64 v[3] = { + insert ? 1 : -1, + insert ? k.k->size : -((s64) k.k->size), + replicas_sectors, + }; + ret = bch2_disk_accounting_mod2(trans, gc, v, inum, k.k->p.inode); if (ret) return ret; } @@ -1031,15 +888,19 @@ static int __trigger_extent(struct btree_trans *trans, } int bch2_trigger_extent(struct btree_trans *trans, - enum btree_id btree_id, unsigned level, + enum btree_id btree, unsigned level, struct bkey_s_c old, struct bkey_s new, - unsigned flags) + enum btree_iter_update_trigger_flags flags) { + struct bch_fs *c = trans->c; struct bkey_ptrs_c new_ptrs = bch2_bkey_ptrs_c(new.s_c); struct bkey_ptrs_c old_ptrs = bch2_bkey_ptrs_c(old); unsigned new_ptrs_bytes = (void *) new_ptrs.end - (void *) new_ptrs.start; unsigned old_ptrs_bytes = (void *) old_ptrs.end - (void *) old_ptrs.start; + if (unlikely(flags & BTREE_TRIGGER_check_repair)) + return bch2_check_fix_ptrs(trans, btree, level, new.s_c, flags); + /* if pointers aren't changing - nothing to do: */ if (new_ptrs_bytes == old_ptrs_bytes && !memcmp(new_ptrs.start, @@ -1047,21 +908,46 @@ int bch2_trigger_extent(struct btree_trans *trans, new_ptrs_bytes)) return 0; - if (flags & BTREE_TRIGGER_TRANSACTIONAL) { - struct bch_fs *c = trans->c; - int mod = (int) bch2_bkey_needs_rebalance(c, new.s_c) - - (int) bch2_bkey_needs_rebalance(c, old); + if (flags & (BTREE_TRIGGER_transactional|BTREE_TRIGGER_gc)) { + if (old.k->type) { + int ret = __trigger_extent(trans, btree, level, old, + flags & ~BTREE_TRIGGER_insert); + if (ret) + return ret; + } + + if (new.k->type) { + int ret = __trigger_extent(trans, btree, level, new.s_c, + flags & ~BTREE_TRIGGER_overwrite); + if (ret) + return ret; + } + + int need_rebalance_delta = 0; + s64 need_rebalance_sectors_delta[1] = { 0 }; - if (mod) { + s64 s = bch2_bkey_sectors_need_rebalance(c, old); + need_rebalance_delta -= s != 0; + need_rebalance_sectors_delta[0] -= s; + + s = bch2_bkey_sectors_need_rebalance(c, new.s_c); + need_rebalance_delta += s != 0; + need_rebalance_sectors_delta[0] += s; + + if ((flags & BTREE_TRIGGER_transactional) && need_rebalance_delta) { int ret = bch2_btree_bit_mod_buffered(trans, BTREE_ID_rebalance_work, - new.k->p, mod > 0); + new.k->p, need_rebalance_delta > 0); if (ret) return ret; } - } - if (flags & (BTREE_TRIGGER_TRANSACTIONAL|BTREE_TRIGGER_GC)) - return trigger_run_overwrite_then_insert(__trigger_extent, trans, btree_id, level, old, new, flags); + if (need_rebalance_sectors_delta[0]) { + int ret = bch2_disk_accounting_mod2(trans, flags & BTREE_TRIGGER_gc, + need_rebalance_sectors_delta, rebalance_work); + if (ret) + return ret; + } + } return 0; } @@ -1069,39 +955,17 @@ int bch2_trigger_extent(struct btree_trans *trans, /* KEY_TYPE_reservation */ static int __trigger_reservation(struct btree_trans *trans, - enum btree_id btree_id, unsigned level, - struct bkey_s_c k, unsigned flags) + enum btree_id btree_id, unsigned level, struct bkey_s_c k, + enum btree_iter_update_trigger_flags flags) { - struct bch_fs *c = trans->c; - unsigned replicas = bkey_s_c_to_reservation(k).v->nr_replicas; - s64 sectors = (s64) k.k->size * replicas; - - if (flags & BTREE_TRIGGER_OVERWRITE) - sectors = -sectors; - - if (flags & BTREE_TRIGGER_TRANSACTIONAL) { - int ret = bch2_replicas_deltas_realloc(trans, 0); - if (ret) - return ret; + if (flags & (BTREE_TRIGGER_transactional|BTREE_TRIGGER_gc)) { + s64 sectors[1] = { k.k->size }; - struct replicas_delta_list *d = trans->fs_usage_deltas; - replicas = min(replicas, ARRAY_SIZE(d->persistent_reserved)); + if (flags & BTREE_TRIGGER_overwrite) + sectors[0] = -sectors[0]; - d->persistent_reserved[replicas - 1] += sectors; - } - - if (flags & BTREE_TRIGGER_GC) { - percpu_down_read(&c->mark_lock); - preempt_disable(); - - struct bch_fs_usage *fs_usage = this_cpu_ptr(c->usage_gc); - - replicas = min(replicas, ARRAY_SIZE(fs_usage->persistent_reserved)); - fs_usage->b.reserved += sectors; - fs_usage->persistent_reserved[replicas - 1] += sectors; - - preempt_enable(); - percpu_up_read(&c->mark_lock); + return bch2_disk_accounting_mod2(trans, flags & BTREE_TRIGGER_gc, sectors, + persistent_reserved, bkey_s_c_to_reservation(k).v->nr_replicas); } return 0; @@ -1110,7 +974,7 @@ static int __trigger_reservation(struct btree_trans *trans, int bch2_trigger_reservation(struct btree_trans *trans, enum btree_id btree_id, unsigned level, struct bkey_s_c old, struct bkey_s new, - unsigned flags) + enum btree_iter_update_trigger_flags flags) { return trigger_run_overwrite_then_insert(__trigger_reservation, trans, btree_id, level, old, new, flags); } @@ -1118,35 +982,39 @@ int bch2_trigger_reservation(struct btree_trans *trans, /* Mark superblocks: */ static int __bch2_trans_mark_metadata_bucket(struct btree_trans *trans, - struct bch_dev *ca, size_t b, + struct bch_dev *ca, u64 b, enum bch_data_type type, unsigned sectors) { struct bch_fs *c = trans->c; struct btree_iter iter; - struct bkey_i_alloc_v4 *a; int ret = 0; - /* - * Backup superblock might be past the end of our normal usable space: - */ - if (b >= ca->mi.nbuckets) - return 0; - - a = bch2_trans_start_alloc_update(trans, &iter, POS(ca->dev_idx, b)); + struct bkey_i_alloc_v4 *a = + bch2_trans_start_alloc_update_noupdate(trans, &iter, POS(ca->dev_idx, b)); if (IS_ERR(a)) return PTR_ERR(a); if (a->v.data_type && type && a->v.data_type != type) { - bch2_fsck_err(c, FSCK_CAN_IGNORE|FSCK_NEED_FSCK, - BCH_FSCK_ERR_bucket_metadata_type_mismatch, - "bucket %llu:%llu gen %u different types of data in same bucket: %s, %s\n" - "while marking %s", - iter.pos.inode, iter.pos.offset, a->v.gen, - bch2_data_type_str(a->v.data_type), - bch2_data_type_str(type), - bch2_data_type_str(type)); - ret = -EIO; + struct printbuf buf = PRINTBUF; + bch2_log_msg_start(c, &buf); + prt_printf(&buf, "bucket %llu:%llu gen %u different types of data in same bucket: %s, %s\n" + "while marking %s\n", + iter.pos.inode, iter.pos.offset, a->v.gen, + bch2_data_type_str(a->v.data_type), + bch2_data_type_str(type), + bch2_data_type_str(type)); + + bch2_count_fsck_err(c, bucket_metadata_type_mismatch, &buf); + + ret = bch2_run_explicit_recovery_pass(c, &buf, + BCH_RECOVERY_PASS_check_allocations, 0); + + /* Always print, this is always fatal */ + bch2_print_str(c, KERN_ERR, buf.buf); + printbuf_exit(&buf); + if (!ret) + ret = bch_err_throw(c, metadata_bucket_inconsistency); goto err; } @@ -1161,20 +1029,75 @@ err: return ret; } +static int bch2_mark_metadata_bucket(struct btree_trans *trans, struct bch_dev *ca, + u64 b, enum bch_data_type data_type, unsigned sectors, + enum btree_iter_update_trigger_flags flags) +{ + struct bch_fs *c = trans->c; + int ret = 0; + + struct bucket *g = gc_bucket(ca, b); + if (bch2_fs_inconsistent_on(!g, c, "reference to invalid bucket on device %u when marking metadata type %s", + ca->dev_idx, bch2_data_type_str(data_type))) + goto err; + + bucket_lock(g); + struct bch_alloc_v4 old = bucket_m_to_alloc(*g); + + if (bch2_fs_inconsistent_on(g->data_type && + g->data_type != data_type, c, + "different types of data in same bucket: %s, %s", + bch2_data_type_str(g->data_type), + bch2_data_type_str(data_type))) + goto err_unlock; + + if (bch2_fs_inconsistent_on((u64) g->dirty_sectors + sectors > ca->mi.bucket_size, c, + "bucket %u:%llu gen %u data type %s sector count overflow: %u + %u > bucket size", + ca->dev_idx, b, g->gen, + bch2_data_type_str(g->data_type ?: data_type), + g->dirty_sectors, sectors)) + goto err_unlock; + + g->data_type = data_type; + g->dirty_sectors += sectors; + struct bch_alloc_v4 new = bucket_m_to_alloc(*g); + bucket_unlock(g); + ret = bch2_alloc_key_to_dev_counters(trans, ca, &old, &new, flags); + return ret; +err_unlock: + bucket_unlock(g); +err: + return bch_err_throw(c, metadata_bucket_inconsistency); +} + int bch2_trans_mark_metadata_bucket(struct btree_trans *trans, - struct bch_dev *ca, size_t b, - enum bch_data_type type, - unsigned sectors) + struct bch_dev *ca, u64 b, + enum bch_data_type type, unsigned sectors, + enum btree_iter_update_trigger_flags flags) { - return commit_do(trans, NULL, NULL, 0, - __bch2_trans_mark_metadata_bucket(trans, ca, b, type, sectors)); + BUG_ON(type != BCH_DATA_free && + type != BCH_DATA_sb && + type != BCH_DATA_journal); + + /* + * Backup superblock might be past the end of our normal usable space: + */ + if (b >= ca->mi.nbuckets) + return 0; + + if (flags & BTREE_TRIGGER_gc) + return bch2_mark_metadata_bucket(trans, ca, b, type, sectors, flags); + else if (flags & BTREE_TRIGGER_transactional) + return commit_do(trans, NULL, NULL, 0, + __bch2_trans_mark_metadata_bucket(trans, ca, b, type, sectors)); + else + BUG(); } static int bch2_trans_mark_metadata_sectors(struct btree_trans *trans, - struct bch_dev *ca, - u64 start, u64 end, - enum bch_data_type type, - u64 *bucket, unsigned *bucket_sectors) + struct bch_dev *ca, u64 start, u64 end, + enum bch_data_type type, u64 *bucket, unsigned *bucket_sectors, + enum btree_iter_update_trigger_flags flags) { do { u64 b = sector_to_bucket(ca, start); @@ -1183,7 +1106,7 @@ static int bch2_trans_mark_metadata_sectors(struct btree_trans *trans, if (b != *bucket && *bucket_sectors) { int ret = bch2_trans_mark_metadata_bucket(trans, ca, *bucket, - type, *bucket_sectors); + type, *bucket_sectors, flags); if (ret) return ret; @@ -1198,35 +1121,40 @@ static int bch2_trans_mark_metadata_sectors(struct btree_trans *trans, return 0; } -static int __bch2_trans_mark_dev_sb(struct btree_trans *trans, - struct bch_dev *ca) +static int __bch2_trans_mark_dev_sb(struct btree_trans *trans, struct bch_dev *ca, + enum btree_iter_update_trigger_flags flags) { - struct bch_sb_layout *layout = &ca->disk_sb.sb->layout; + struct bch_fs *c = trans->c; + + mutex_lock(&c->sb_lock); + struct bch_sb_layout layout = ca->disk_sb.sb->layout; + mutex_unlock(&c->sb_lock); + u64 bucket = 0; unsigned i, bucket_sectors = 0; int ret; - for (i = 0; i < layout->nr_superblocks; i++) { - u64 offset = le64_to_cpu(layout->sb_offset[i]); + for (i = 0; i < layout.nr_superblocks; i++) { + u64 offset = le64_to_cpu(layout.sb_offset[i]); if (offset == BCH_SB_SECTOR) { ret = bch2_trans_mark_metadata_sectors(trans, ca, 0, BCH_SB_SECTOR, - BCH_DATA_sb, &bucket, &bucket_sectors); + BCH_DATA_sb, &bucket, &bucket_sectors, flags); if (ret) return ret; } ret = bch2_trans_mark_metadata_sectors(trans, ca, offset, - offset + (1 << layout->sb_max_size_bits), - BCH_DATA_sb, &bucket, &bucket_sectors); + offset + (1 << layout.sb_max_size_bits), + BCH_DATA_sb, &bucket, &bucket_sectors, flags); if (ret) return ret; } if (bucket_sectors) { ret = bch2_trans_mark_metadata_bucket(trans, ca, - bucket, BCH_DATA_sb, bucket_sectors); + bucket, BCH_DATA_sb, bucket_sectors, flags); if (ret) return ret; } @@ -1234,7 +1162,7 @@ static int __bch2_trans_mark_dev_sb(struct btree_trans *trans, for (i = 0; i < ca->journal.nr; i++) { ret = bch2_trans_mark_metadata_bucket(trans, ca, ca->journal.buckets[i], - BCH_DATA_journal, ca->mi.bucket_size); + BCH_DATA_journal, ca->mi.bucket_size, flags); if (ret) return ret; } @@ -1242,20 +1170,22 @@ static int __bch2_trans_mark_dev_sb(struct btree_trans *trans, return 0; } -int bch2_trans_mark_dev_sb(struct bch_fs *c, struct bch_dev *ca) +int bch2_trans_mark_dev_sb(struct bch_fs *c, struct bch_dev *ca, + enum btree_iter_update_trigger_flags flags) { - int ret = bch2_trans_run(c, __bch2_trans_mark_dev_sb(trans, ca)); - + int ret = bch2_trans_run(c, + __bch2_trans_mark_dev_sb(trans, ca, flags)); bch_err_fn(c, ret); return ret; } -int bch2_trans_mark_dev_sbs(struct bch_fs *c) +int bch2_trans_mark_dev_sbs_flags(struct bch_fs *c, + enum btree_iter_update_trigger_flags flags) { - for_each_online_member(c, ca) { - int ret = bch2_trans_mark_dev_sb(c, ca); + for_each_online_member(c, ca, BCH_DEV_READ_REF_trans_mark_dev_sbs) { + int ret = bch2_trans_mark_dev_sb(c, ca, flags); if (ret) { - percpu_ref_put(&ca->ref); + enumerated_ref_put(&ca->io_ref[READ], BCH_DEV_READ_REF_trans_mark_dev_sbs); return ret; } } @@ -1263,16 +1193,46 @@ int bch2_trans_mark_dev_sbs(struct bch_fs *c) return 0; } +int bch2_trans_mark_dev_sbs(struct bch_fs *c) +{ + return bch2_trans_mark_dev_sbs_flags(c, BTREE_TRIGGER_transactional); +} + +bool bch2_is_superblock_bucket(struct bch_dev *ca, u64 b) +{ + struct bch_sb_layout *layout = &ca->disk_sb.sb->layout; + u64 b_offset = bucket_to_sector(ca, b); + u64 b_end = bucket_to_sector(ca, b + 1); + unsigned i; + + if (!b) + return true; + + for (i = 0; i < layout->nr_superblocks; i++) { + u64 offset = le64_to_cpu(layout->sb_offset[i]); + u64 end = offset + (1 << layout->sb_max_size_bits); + + if (!(offset >= b_end || end <= b_offset)) + return true; + } + + for (i = 0; i < ca->journal.nr; i++) + if (b == ca->journal.buckets[i]) + return true; + + return false; +} + /* Disk reservations: */ #define SECTORS_CACHE 1024 int __bch2_disk_reservation_add(struct bch_fs *c, struct disk_reservation *res, - u64 sectors, int flags) + u64 sectors, enum bch_reservation_flags flags) { struct bch_fs_pcpu *pcpu; - u64 old, v, get; - s64 sectors_available; + u64 old, get; + u64 sectors_available; int ret; percpu_down_read(&c->mark_lock); @@ -1282,17 +1242,16 @@ int __bch2_disk_reservation_add(struct bch_fs *c, struct disk_reservation *res, if (sectors <= pcpu->sectors_available) goto out; - v = atomic64_read(&c->sectors_available); + old = atomic64_read(&c->sectors_available); do { - old = v; get = min((u64) sectors + SECTORS_CACHE, old); if (get < sectors) { preempt_enable(); goto recalculate; } - } while ((v = atomic64_cmpxchg(&c->sectors_available, - old, old - get)) != old); + } while (!atomic64_try_cmpxchg(&c->sectors_available, + &old, old - get)); pcpu->sectors_available += get; @@ -1311,6 +1270,9 @@ recalculate: percpu_u64_set(&c->pcpu->sectors_available, 0); sectors_available = avail_factor(__bch2_fs_usage_read_short(c).free); + if (sectors_available && (flags & BCH_DISK_RESERVATION_PARTIAL)) + sectors = min(sectors, sectors_available); + if (sectors <= sectors_available || (flags & BCH_DISK_RESERVATION_NOFAIL)) { atomic64_set(&c->sectors_available, @@ -1320,7 +1282,7 @@ recalculate: ret = 0; } else { atomic64_set(&c->sectors_available, sectors_available); - ret = -BCH_ERR_ENOSPC_disk_reservation; + ret = bch_err_throw(c, ENOSPC_disk_reservation); } mutex_unlock(&c->sectors_available_lock); @@ -1331,6 +1293,31 @@ recalculate: /* Startup/shutdown: */ +void bch2_buckets_nouse_free(struct bch_fs *c) +{ + for_each_member_device(c, ca) { + kvfree_rcu_mightsleep(ca->buckets_nouse); + ca->buckets_nouse = NULL; + } +} + +int bch2_buckets_nouse_alloc(struct bch_fs *c) +{ + for_each_member_device(c, ca) { + BUG_ON(ca->buckets_nouse); + + ca->buckets_nouse = bch2_kvmalloc(BITS_TO_LONGS(ca->mi.nbuckets) * + sizeof(unsigned long), + GFP_KERNEL|__GFP_ZERO); + if (!ca->buckets_nouse) { + bch2_dev_put(ca); + return bch_err_throw(c, ENOMEM_buckets_nouse); + } + } + + return 0; +} + static void bucket_gens_free_rcu(struct rcu_head *rcu) { struct bucket_gens *buckets = @@ -1342,63 +1329,49 @@ static void bucket_gens_free_rcu(struct rcu_head *rcu) int bch2_dev_buckets_resize(struct bch_fs *c, struct bch_dev *ca, u64 nbuckets) { struct bucket_gens *bucket_gens = NULL, *old_bucket_gens = NULL; - unsigned long *buckets_nouse = NULL; bool resize = ca->bucket_gens != NULL; int ret; - if (!(bucket_gens = kvmalloc(sizeof(struct bucket_gens) + nbuckets, - GFP_KERNEL|__GFP_ZERO))) { - ret = -BCH_ERR_ENOMEM_bucket_gens; - goto err; - } + if (resize) + lockdep_assert_held(&c->state_lock); - if ((c->opts.buckets_nouse && - !(buckets_nouse = kvmalloc(BITS_TO_LONGS(nbuckets) * - sizeof(unsigned long), - GFP_KERNEL|__GFP_ZERO)))) { - ret = -BCH_ERR_ENOMEM_buckets_nouse; + if (resize && ca->buckets_nouse) + return bch_err_throw(c, no_resize_with_buckets_nouse); + + bucket_gens = bch2_kvmalloc(struct_size(bucket_gens, b, nbuckets), + GFP_KERNEL|__GFP_ZERO); + if (!bucket_gens) { + ret = bch_err_throw(c, ENOMEM_bucket_gens); goto err; } bucket_gens->first_bucket = ca->mi.first_bucket; bucket_gens->nbuckets = nbuckets; - - if (resize) { - down_write(&c->gc_lock); - down_write(&ca->bucket_lock); - percpu_down_write(&c->mark_lock); - } + bucket_gens->nbuckets_minus_first = + bucket_gens->nbuckets - bucket_gens->first_bucket; old_bucket_gens = rcu_dereference_protected(ca->bucket_gens, 1); if (resize) { - size_t n = min(bucket_gens->nbuckets, old_bucket_gens->nbuckets); - + u64 copy = min(bucket_gens->nbuckets, + old_bucket_gens->nbuckets); memcpy(bucket_gens->b, old_bucket_gens->b, - n); - if (buckets_nouse) - memcpy(buckets_nouse, - ca->buckets_nouse, - BITS_TO_LONGS(n) * sizeof(unsigned long)); + sizeof(bucket_gens->b[0]) * copy); } + ret = bch2_bucket_bitmap_resize(ca, &ca->bucket_backpointer_mismatch, + ca->mi.nbuckets, nbuckets) ?: + bch2_bucket_bitmap_resize(ca, &ca->bucket_backpointer_empty, + ca->mi.nbuckets, nbuckets); + rcu_assign_pointer(ca->bucket_gens, bucket_gens); bucket_gens = old_bucket_gens; - swap(ca->buckets_nouse, buckets_nouse); - nbuckets = ca->mi.nbuckets; - if (resize) { - percpu_up_write(&c->mark_lock); - up_write(&ca->bucket_lock); - up_write(&c->gc_lock); - } - ret = 0; err: - kvfree(buckets_nouse); if (bucket_gens) call_rcu(&bucket_gens->rcu, bucket_gens_free_rcu); @@ -1409,23 +1382,14 @@ void bch2_dev_buckets_free(struct bch_dev *ca) { kvfree(ca->buckets_nouse); kvfree(rcu_dereference_protected(ca->bucket_gens, 1)); - - for (unsigned i = 0; i < ARRAY_SIZE(ca->usage); i++) - free_percpu(ca->usage[i]); - kfree(ca->usage_base); + free_percpu(ca->usage); } int bch2_dev_buckets_alloc(struct bch_fs *c, struct bch_dev *ca) { - ca->usage_base = kzalloc(sizeof(struct bch_dev_usage), GFP_KERNEL); - if (!ca->usage_base) - return -BCH_ERR_ENOMEM_usage_init; - - for (unsigned i = 0; i < ARRAY_SIZE(ca->usage); i++) { - ca->usage[i] = alloc_percpu(struct bch_dev_usage); - if (!ca->usage[i]) - return -BCH_ERR_ENOMEM_usage_init; - } + ca->usage = alloc_percpu(struct bch_dev_usage_full); + if (!ca->usage) + return bch_err_throw(c, ENOMEM_usage_init); return bch2_dev_buckets_resize(c, ca, ca->mi.nbuckets); } |