#include "reflink.h"
#include "replicas.h"
#include "subvolume.h"
+#include "trace.h"
#include <linux/preempt.h>
-#include <trace/events/bcachefs.h>
static inline void fs_usage_data_type_to_base(struct bch_fs_usage *fs_usage,
enum bch_data_type data_type,
} while (read_seqcount_retry(&c->usage_lock, seq));
}
-static inline struct bch_fs_usage *fs_usage_ptr(struct bch_fs *c,
- unsigned journal_seq,
- bool gc)
-{
- percpu_rwsem_assert_held(&c->mark_lock);
- BUG_ON(!gc && !journal_seq);
-
- return this_cpu_ptr(gc
- ? c->usage_gc
- : c->usage[journal_seq & JOURNAL_BUF_MASK]);
-}
-
u64 bch2_fs_usage_read_one(struct bch_fs *c, u64 *v)
{
ssize_t offset = v - (u64 *) c->usage_base;
struct bch_fs_usage_online *bch2_fs_usage_read(struct bch_fs *c)
{
struct bch_fs_usage_online *ret;
- unsigned seq, i, v, u64s = fs_usage_u64s(c) + 1;
+ unsigned nr_replicas = READ_ONCE(c->replicas.nr);
+ unsigned seq, i;
retry:
- ret = kmalloc(u64s * sizeof(u64), GFP_NOFS);
+ ret = kmalloc(__fs_usage_online_u64s(nr_replicas) * sizeof(u64), GFP_KERNEL);
if (unlikely(!ret))
return NULL;
percpu_down_read(&c->mark_lock);
- v = fs_usage_u64s(c) + 1;
- if (unlikely(u64s != v)) {
- u64s = v;
+ if (nr_replicas != c->replicas.nr) {
+ nr_replicas = c->replicas.nr;
percpu_up_read(&c->mark_lock);
kfree(ret);
goto retry;
do {
seq = read_seqcount_begin(&c->usage_lock);
- unsafe_memcpy(&ret->u, c->usage_base, u64s * sizeof(u64),
+ 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], u64s);
+ 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;
struct bch_replicas_entry *r, s64 sectors,
unsigned journal_seq, bool gc)
{
- struct bch_fs_usage __percpu *fs_usage;
+ struct bch_fs_usage *fs_usage;
int idx, ret = 0;
struct printbuf buf = PRINTBUF;
return update_replicas(c, k, &r.e, sectors, journal_seq, gc);
}
-static struct replicas_delta_list *
-replicas_deltas_realloc(struct btree_trans *trans, unsigned more)
+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;
WARN_ON_ONCE(alloc_size > REPLICAS_DELTA_LIST_MAX);
if (!d || d->used + more > d->size) {
- d = krealloc(d, alloc_size, GFP_NOIO|__GFP_ZERO);
+ d = krealloc(d, alloc_size, gfp|__GFP_ZERO);
+
+ if (unlikely(!d)) {
+ if (alloc_size > REPLICAS_DELTA_LIST_MAX)
+ return -ENOMEM;
- BUG_ON(!d && alloc_size > REPLICAS_DELTA_LIST_MAX);
+ d = mempool_alloc(&trans->c->replicas_delta_pool, gfp);
+ if (!d)
+ return -ENOMEM;
- if (!d) {
- d = mempool_alloc(&trans->c->replicas_delta_pool, GFP_NOIO);
memset(d, 0, REPLICAS_DELTA_LIST_MAX);
if (trans->fs_usage_deltas)
d->size = new_size;
trans->fs_usage_deltas = d;
}
- return d;
+
+ return 0;
}
-static inline void update_replicas_list(struct btree_trans *trans,
+int bch2_replicas_deltas_realloc(struct btree_trans *trans, unsigned more)
+{
+ return allocate_dropping_locks_errcode(trans,
+ __replicas_deltas_realloc(trans, more, _gfp));
+}
+
+static inline int update_replicas_list(struct btree_trans *trans,
struct bch_replicas_entry *r,
s64 sectors)
{
struct replicas_delta_list *d;
struct replicas_delta *n;
unsigned b;
+ int ret;
if (!sectors)
- return;
+ return 0;
b = replicas_entry_bytes(r) + 8;
- d = replicas_deltas_realloc(trans, b);
+ 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;
memcpy((void *) n + offsetof(struct replicas_delta, r),
r, replicas_entry_bytes(r));
bch2_replicas_entry_sort(&n->r);
d->used += b;
+ return 0;
}
-static inline void update_cached_sectors_list(struct btree_trans *trans,
+static inline int update_cached_sectors_list(struct btree_trans *trans,
unsigned dev, s64 sectors)
{
struct bch_replicas_padded r;
bch2_replicas_entry_cached(&r.e, dev);
- update_replicas_list(trans, &r.e, sectors);
+ return update_replicas_list(trans, &r.e, sectors);
}
int bch2_mark_alloc(struct btree_trans *trans,
+ enum btree_id btree, unsigned level,
struct bkey_s_c old, struct bkey_s_c new,
unsigned flags)
{
if (b_gen != ptr->gen) {
ret = 1;
- goto err;
+ goto out;
}
if (!data_type_is_empty(bucket_data_type) &&
const struct bch_stripe *s = bkey_s_c_to_stripe(k).v;
unsigned nr_data = s->nr_blocks - s->nr_redundant;
bool parity = ptr_idx >= nr_data;
- enum bch_data_type data_type = parity ? BCH_DATA_parity : 0;
+ enum bch_data_type data_type = parity ? BCH_DATA_parity : BCH_DATA_stripe;
s64 sectors = parity ? le16_to_cpu(s->sectors) : 0;
const struct bch_extent_ptr *ptr = s->ptrs + ptr_idx;
struct bch_dev *ca = bch_dev_bkey_exists(c, ptr->dev);
if (ret)
goto err;
- if (data_type)
- g->data_type = data_type;
+ g->data_type = data_type;
g->dirty_sectors += sectors;
g->stripe = k.k->p.offset;
}
static int bch2_mark_pointer(struct btree_trans *trans,
+ enum btree_id btree_id, unsigned level,
struct bkey_s_c k,
struct extent_ptr_decoded p,
- s64 sectors, enum bch_data_type data_type,
+ s64 sectors,
unsigned flags)
{
u64 journal_seq = trans->journal_res.seq;
struct bch_fs *c = trans->c;
struct bch_dev *ca = bch_dev_bkey_exists(c, p.ptr.dev);
struct bucket old, new, *g;
+ enum bch_data_type data_type = bkey_ptr_data_type(btree_id, level, k, p);
u8 bucket_data_type;
int ret = 0;
if (!m) {
bch_err(c, "error allocating memory for gc_stripes, idx %llu",
(u64) p.idx);
- return -ENOMEM;
+ return -BCH_ERR_ENOMEM_mark_stripe_ptr;
}
- spin_lock(&c->ec_stripes_heap_lock);
+ mutex_lock(&c->ec_stripes_heap_lock);
if (!m || !m->alive) {
- spin_unlock(&c->ec_stripes_heap_lock);
+ mutex_unlock(&c->ec_stripes_heap_lock);
bch_err_ratelimited(c, "pointer to nonexistent stripe %llu",
(u64) p.idx);
bch2_inconsistent_error(c);
m->block_sectors[p.block] += sectors;
r = m->r;
- spin_unlock(&c->ec_stripes_heap_lock);
+ mutex_unlock(&c->ec_stripes_heap_lock);
r.e.data_type = data_type;
update_replicas(c, k, &r.e, sectors, trans->journal_res.seq, true);
}
int bch2_mark_extent(struct btree_trans *trans,
+ enum btree_id btree_id, unsigned level,
struct bkey_s_c old, struct bkey_s_c new,
unsigned flags)
{
if (flags & BTREE_TRIGGER_OVERWRITE)
disk_sectors = -disk_sectors;
- ret = bch2_mark_pointer(trans, k, p, disk_sectors,
- data_type, flags);
+ ret = bch2_mark_pointer(trans, btree_id, level, k, p, disk_sectors, flags);
if (ret < 0)
return ret;
}
int bch2_mark_stripe(struct btree_trans *trans,
+ enum btree_id btree_id, unsigned level,
struct bkey_s_c old, struct bkey_s_c new,
unsigned flags)
{
if (!gc) {
struct stripe *m = genradix_ptr(&c->stripes, idx);
- if (!m || (old_s && !m->alive)) {
+ if (!m) {
struct printbuf buf1 = PRINTBUF;
struct printbuf buf2 = PRINTBUF;
}
if (!new_s) {
- spin_lock(&c->ec_stripes_heap_lock);
bch2_stripes_heap_del(c, m, idx);
- spin_unlock(&c->ec_stripes_heap_lock);
memset(m, 0, sizeof(*m));
} else {
- m->alive = true;
m->sectors = le16_to_cpu(new_s->sectors);
m->algorithm = new_s->algorithm;
m->nr_blocks = new_s->nr_blocks;
for (i = 0; i < new_s->nr_blocks; i++)
m->blocks_nonempty += !!stripe_blockcount_get(new_s, i);
- spin_lock(&c->ec_stripes_heap_lock);
- bch2_stripes_heap_update(c, m, idx);
- spin_unlock(&c->ec_stripes_heap_lock);
+ if (!old_s)
+ bch2_stripes_heap_insert(c, m, idx);
+ else
+ bch2_stripes_heap_update(c, m, idx);
}
} else {
struct gc_stripe *m =
if (!m) {
bch_err(c, "error allocating memory for gc_stripes, idx %llu",
idx);
- return -ENOMEM;
+ return -BCH_ERR_ENOMEM_mark_stripe;
}
/*
* This will be wrong when we bring back runtime gc: we should
return 0;
}
-int bch2_mark_inode(struct btree_trans *trans,
- struct bkey_s_c old, struct bkey_s_c new,
- unsigned flags)
-{
- struct bch_fs *c = trans->c;
- struct bch_fs_usage __percpu *fs_usage;
- u64 journal_seq = trans->journal_res.seq;
-
- if (flags & BTREE_TRIGGER_INSERT) {
- struct bch_inode_v3 *v = (struct bch_inode_v3 *) new.v;
-
- BUG_ON(!journal_seq);
- BUG_ON(new.k->type != KEY_TYPE_inode_v3);
-
- v->bi_journal_seq = cpu_to_le64(journal_seq);
- }
-
- if (flags & BTREE_TRIGGER_GC) {
- percpu_down_read(&c->mark_lock);
- preempt_disable();
-
- fs_usage = fs_usage_ptr(c, journal_seq, flags & BTREE_TRIGGER_GC);
- fs_usage->nr_inodes += bkey_is_inode(new.k);
- fs_usage->nr_inodes -= bkey_is_inode(old.k);
-
- preempt_enable();
- percpu_up_read(&c->mark_lock);
- }
- return 0;
-}
-
int bch2_mark_reservation(struct btree_trans *trans,
+ enum btree_id btree_id, unsigned level,
struct bkey_s_c old, struct bkey_s_c new,
unsigned flags)
{
struct bch_fs *c = trans->c;
struct bkey_s_c k = flags & BTREE_TRIGGER_OVERWRITE ? old : new;
- struct bch_fs_usage __percpu *fs_usage;
+ struct bch_fs_usage *fs_usage;
unsigned replicas = bkey_s_c_to_reservation(k).v->nr_replicas;
s64 sectors = (s64) k.k->size;
}
int bch2_mark_reflink_p(struct btree_trans *trans,
+ enum btree_id btree_id, unsigned level,
struct bkey_s_c old, struct bkey_s_c new,
unsigned flags)
{
BUG_ON(!(flags & BTREE_TRIGGER_GC));
- if (c->sb.version >= bcachefs_metadata_version_reflink_p_fix) {
+ if (c->sb.version_upgrade_complete >= bcachefs_metadata_version_reflink_p_fix) {
idx -= le32_to_cpu(p.v->front_pad);
end += le32_to_cpu(p.v->back_pad);
}
bool insert = !(flags & BTREE_TRIGGER_OVERWRITE);
struct btree_iter iter;
struct bkey_i_alloc_v4 *a;
- struct bpos bucket_pos;
+ struct bpos bucket;
struct bch_backpointer bp;
s64 sectors;
int ret;
- bch2_extent_ptr_to_bp(trans->c, btree_id, level, k, p, &bucket_pos, &bp);
+ bch2_extent_ptr_to_bp(trans->c, btree_id, level, k, p, &bucket, &bp);
sectors = bp.bucket_len;
if (!insert)
sectors = -sectors;
- a = bch2_trans_start_alloc_update(trans, &iter, bucket_pos);
+ a = bch2_trans_start_alloc_update(trans, &iter, bucket);
if (IS_ERR(a))
return PTR_ERR(a);
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);
+ &a->v.dirty_sectors, &a->v.cached_sectors) ?:
+ bch2_trans_update(trans, &iter, &a->k_i, 0);
+ bch2_trans_iter_exit(trans, &iter);
+
if (ret)
- goto err;
+ return ret;
if (!p.ptr.cached) {
- ret = bch2_bucket_backpointer_mod(trans, a, bp, k, insert);
+ ret = bch2_bucket_backpointer_mod(trans, bucket, bp, k, insert);
if (ret)
- goto err;
+ return ret;
}
- ret = bch2_trans_update(trans, &iter, &a->k_i, 0);
-err:
- bch2_trans_iter_exit(trans, &iter);
- return ret;
+ return 0;
}
static int bch2_trans_mark_stripe_ptr(struct btree_trans *trans,
struct bch_replicas_padded r;
int ret = 0;
- bch2_trans_iter_init(trans, &iter, BTREE_ID_stripes, POS(0, p.ec.idx),
- BTREE_ITER_INTENT|
- BTREE_ITER_WITH_UPDATES);
- s = bch2_bkey_get_mut_typed(trans, &iter, stripe);
+ s = bch2_bkey_get_mut_typed(trans, &iter,
+ BTREE_ID_stripes, POS(0, p.ec.idx),
+ BTREE_ITER_WITH_UPDATES, stripe);
ret = PTR_ERR_OR_ZERO(s);
if (unlikely(ret)) {
- bch2_trans_inconsistent_on(ret == -ENOENT, trans,
+ bch2_trans_inconsistent_on(bch2_err_matches(ret, ENOENT), trans,
"pointer to nonexistent stripe %llu",
(u64) p.ec.idx);
goto err;
stripe_blockcount_get(&s->v, p.ec.block) +
sectors);
- ret = bch2_trans_update(trans, &iter, &s->k_i, 0);
- if (ret)
- goto err;
-
bch2_bkey_to_replicas(&r.e, bkey_i_to_s_c(&s->k_i));
r.e.data_type = data_type;
- update_replicas_list(trans, &r.e, sectors);
+ ret = update_replicas_list(trans, &r.e, sectors);
err:
bch2_trans_iter_exit(trans, &iter);
return ret;
: k.k->size;
s64 dirty_sectors = 0;
bool stale;
- int ret;
+ int ret = 0;
r.e.data_type = data_type;
r.e.nr_devs = 0;
stale = ret > 0;
if (p.ptr.cached) {
- if (!stale)
- update_cached_sectors_list(trans, p.ptr.dev,
- disk_sectors);
+ if (!stale) {
+ ret = update_cached_sectors_list(trans, p.ptr.dev,
+ disk_sectors);
+ if (ret)
+ return ret;
+ }
} else if (!p.has_ec) {
dirty_sectors += disk_sectors;
r.e.devs[r.e.nr_devs++] = p.ptr.dev;
}
if (r.e.nr_devs)
- update_replicas_list(trans, &r.e, dirty_sectors);
+ ret = update_replicas_list(trans, &r.e, dirty_sectors);
- return 0;
+ return ret;
}
static int bch2_trans_mark_stripe_bucket(struct btree_trans *trans,
a->v.stripe = s.k->p.offset;
a->v.stripe_redundancy = s.v->nr_redundant;
+ a->v.data_type = BCH_DATA_stripe;
} else {
if (bch2_trans_inconsistent_on(a->v.stripe != s.k->p.offset ||
a->v.stripe_redundancy != s.v->nr_redundant, trans,
a->v.stripe = 0;
a->v.stripe_redundancy = 0;
+ a->v.data_type = alloc_data_type(a->v, BCH_DATA_user);
}
a->v.dirty_sectors += sectors;
s64 sectors = le16_to_cpu(new_s->sectors);
bch2_bkey_to_replicas(&r.e, bkey_i_to_s_c(new));
- update_replicas_list(trans, &r.e, sectors * new_s->nr_redundant);
+ ret = update_replicas_list(trans, &r.e, sectors * new_s->nr_redundant);
+ if (ret)
+ return ret;
}
if (old_s) {
s64 sectors = -((s64) le16_to_cpu(old_s->sectors));
bch2_bkey_to_replicas(&r.e, old);
- update_replicas_list(trans, &r.e, sectors * old_s->nr_redundant);
+ ret = update_replicas_list(trans, &r.e, sectors * old_s->nr_redundant);
+ if (ret)
+ return ret;
}
for (i = 0; i < nr_blocks; i++) {
return ret;
}
-int bch2_trans_mark_inode(struct btree_trans *trans,
- enum btree_id btree_id, unsigned level,
- struct bkey_s_c old,
- struct bkey_i *new,
- unsigned flags)
-{
- int nr = bkey_is_inode(&new->k) - bkey_is_inode(old.k);
-
- if (nr) {
- struct replicas_delta_list *d =
- replicas_deltas_realloc(trans, 0);
- d->nr_inodes += nr;
- }
-
- return 0;
-}
-
int bch2_trans_mark_reservation(struct btree_trans *trans,
enum btree_id btree_id, unsigned level,
struct bkey_s_c old,
unsigned replicas = bkey_s_c_to_reservation(k).v->nr_replicas;
s64 sectors = (s64) k.k->size;
struct replicas_delta_list *d;
+ int ret;
if (flags & BTREE_TRIGGER_OVERWRITE)
sectors = -sectors;
sectors *= replicas;
- d = replicas_deltas_realloc(trans, 0);
+ ret = bch2_replicas_deltas_realloc(trans, 0);
+ if (ret)
+ return ret;
+ d = trans->fs_usage_deltas;
replicas = clamp_t(unsigned, replicas, 1,
ARRAY_SIZE(d->persistent_reserved));
struct printbuf buf = PRINTBUF;
int ret;
- bch2_trans_iter_init(trans, &iter, BTREE_ID_reflink, POS(0, *idx),
- BTREE_ITER_INTENT|
- BTREE_ITER_WITH_UPDATES);
- k = bch2_bkey_get_mut(trans, &iter);
+ k = bch2_bkey_get_mut_noupdate(trans, &iter,
+ BTREE_ID_reflink, POS(0, *idx),
+ BTREE_ITER_WITH_UPDATES);
ret = PTR_ERR_OR_ZERO(k);
if (ret)
goto err;
if (IS_ERR(a))
return PTR_ERR(a);
- if (a->v.data_type && a->v.data_type != type) {
+ if (a->v.data_type && type && a->v.data_type != type) {
bch2_fsck_err(c, FSCK_CAN_IGNORE|FSCK_NEED_FSCK,
"bucket %llu:%llu gen %u different types of data in same bucket: %s, %s\n"
"while marking %s",
int bch2_trans_mark_dev_sb(struct bch_fs *c, struct bch_dev *ca)
{
- return bch2_trans_run(c, __bch2_trans_mark_dev_sb(&trans, ca));
+ int ret = bch2_trans_run(c, __bch2_trans_mark_dev_sb(&trans, ca));
+ if (ret)
+ bch_err_fn(c, ret);
+ return ret;
}
/* Disk reservations: */
struct bucket_gens *bucket_gens = NULL, *old_bucket_gens = NULL;
unsigned long *buckets_nouse = NULL;
bool resize = ca->bucket_gens != NULL;
- int ret = -ENOMEM;
+ int ret;
if (!(bucket_gens = kvpmalloc(sizeof(struct bucket_gens) + nbuckets,
- GFP_KERNEL|__GFP_ZERO)) ||
- (c->opts.buckets_nouse &&
+ GFP_KERNEL|__GFP_ZERO))) {
+ ret = -BCH_ERR_ENOMEM_bucket_gens;
+ goto err;
+ }
+
+ if ((c->opts.buckets_nouse &&
!(buckets_nouse = kvpmalloc(BITS_TO_LONGS(nbuckets) *
sizeof(unsigned long),
- GFP_KERNEL|__GFP_ZERO))))
+ GFP_KERNEL|__GFP_ZERO)))) {
+ ret = -BCH_ERR_ENOMEM_buckets_nouse;
goto err;
+ }
bucket_gens->first_bucket = ca->mi.first_bucket;
bucket_gens->nbuckets = nbuckets;
ca->usage_base = kzalloc(sizeof(struct bch_dev_usage), GFP_KERNEL);
if (!ca->usage_base)
- return -ENOMEM;
+ return -BCH_ERR_ENOMEM_usage_init;
for (i = 0; i < ARRAY_SIZE(ca->usage); i++) {
ca->usage[i] = alloc_percpu(struct bch_dev_usage);
if (!ca->usage[i])
- return -ENOMEM;
+ return -BCH_ERR_ENOMEM_usage_init;
}
return bch2_dev_buckets_resize(c, ca, ca->mi.nbuckets);