Soon, bset.c won't need to depend on struct btree.
Signed-off-by: Kent Overstreet <kmo@daterainc.com>
unsigned error_decay;
unsigned short journal_delay_ms;
+ bool expensive_debug_checks;
unsigned verify:1;
unsigned key_merging_disabled:1;
- unsigned expensive_debug_checks:1;
unsigned gc_always_rewrite:1;
unsigned shrinker_disabled:1;
unsigned copy_gc_enabled:1;
#define BSET_CACHELINE 128
/* Space required for the btree node keys */
-static inline size_t btree_keys_bytes(struct btree *b)
+static inline size_t btree_keys_bytes(struct btree_keys *b)
{
return PAGE_SIZE << b->page_order;
}
-static inline size_t btree_keys_cachelines(struct btree *b)
+static inline size_t btree_keys_cachelines(struct btree_keys *b)
{
return btree_keys_bytes(b) / BSET_CACHELINE;
}
/* Space required for the auxiliary search trees */
-static inline size_t bset_tree_bytes(struct btree *b)
+static inline size_t bset_tree_bytes(struct btree_keys *b)
{
return btree_keys_cachelines(b) * sizeof(struct bkey_float);
}
/* Space required for the prev pointers */
-static inline size_t bset_prev_bytes(struct btree *b)
+static inline size_t bset_prev_bytes(struct btree_keys *b)
{
return btree_keys_cachelines(b) * sizeof(uint8_t);
}
/* Memory allocation */
-void bch_btree_keys_free(struct btree *b)
+void bch_btree_keys_free(struct btree_keys *b)
{
- struct bset_tree *t = b->sets;
+ struct bset_tree *t = b->set;
if (bset_prev_bytes(b) < PAGE_SIZE)
kfree(t->prev);
t->tree = NULL;
t->data = NULL;
}
+EXPORT_SYMBOL(bch_btree_keys_free);
-int bch_btree_keys_alloc(struct btree *b, unsigned page_order, gfp_t gfp)
+int bch_btree_keys_alloc(struct btree_keys *b, unsigned page_order, gfp_t gfp)
{
- struct bset_tree *t = b->sets;
+ struct bset_tree *t = b->set;
BUG_ON(t->data);
bch_btree_keys_free(b);
return -ENOMEM;
}
+EXPORT_SYMBOL(bch_btree_keys_alloc);
+
+void bch_btree_keys_init(struct btree_keys *b, const struct btree_keys_ops *ops,
+ bool *expensive_debug_checks)
+{
+ unsigned i;
+
+ b->ops = ops;
+ b->expensive_debug_checks = expensive_debug_checks;
+ b->nsets = 0;
+ b->last_set_unwritten = 0;
+
+ /* XXX: shouldn't be needed */
+ for (i = 0; i < MAX_BSETS; i++)
+ b->set[i].size = 0;
+ /*
+ * Second loop starts at 1 because b->keys[0]->data is the memory we
+ * allocated
+ */
+ for (i = 1; i < MAX_BSETS; i++)
+ b->set[i].data = NULL;
+}
+EXPORT_SYMBOL(bch_btree_keys_init);
/* Binary tree stuff for auxiliary search trees */
f->exponent = 127;
}
-static void bset_alloc_tree(struct btree *b, struct bset_tree *t)
+static void bset_alloc_tree(struct btree_keys *b, struct bset_tree *t)
{
- if (t != b->sets) {
+ if (t != b->set) {
unsigned j = roundup(t[-1].size,
64 / sizeof(struct bkey_float));
t->prev = t[-1].prev + j;
}
- while (t < b->sets + MAX_BSETS)
+ while (t < b->set + MAX_BSETS)
t++->size = 0;
}
-static void bch_bset_build_unwritten_tree(struct btree *b)
+static void bch_bset_build_unwritten_tree(struct btree_keys *b)
{
struct bset_tree *t = bset_tree_last(b);
+ BUG_ON(b->last_set_unwritten);
+ b->last_set_unwritten = 1;
+
bset_alloc_tree(b, t);
- if (t->tree != b->sets->tree + btree_keys_cachelines(b)) {
+ if (t->tree != b->set->tree + btree_keys_cachelines(b)) {
t->prev[0] = bkey_to_cacheline_offset(t->data->start);
t->size = 1;
}
}
-void bch_bset_init_next(struct btree *b, struct bset *i, uint64_t magic)
+void bch_bset_init_next(struct btree_keys *b, struct bset *i, uint64_t magic)
{
- if (i != b->sets->data) {
- b->sets[++b->nsets].data = i;
- i->seq = b->sets->data->seq;
+ if (i != b->set->data) {
+ b->set[++b->nsets].data = i;
+ i->seq = b->set->data->seq;
} else
get_random_bytes(&i->seq, sizeof(uint64_t));
bch_bset_build_unwritten_tree(b);
}
+EXPORT_SYMBOL(bch_bset_init_next);
-static void bset_build_written_tree(struct btree *b)
+void bch_bset_build_written_tree(struct btree_keys *b)
{
struct bset_tree *t = bset_tree_last(b);
struct bkey *k = t->data->start;
unsigned j, cacheline = 1;
+ b->last_set_unwritten = 0;
+
bset_alloc_tree(b, t);
t->size = min_t(unsigned,
bkey_to_cacheline(t, bset_bkey_last(t->data)),
- b->sets->tree + btree_keys_cachelines(b) - t->tree);
+ b->set->tree + btree_keys_cachelines(b) - t->tree);
if (t->size < 2) {
t->size = 0;
j = inorder_next(j, t->size))
make_bfloat(t, j);
}
+EXPORT_SYMBOL(bch_bset_build_written_tree);
-void bch_bset_fix_invalidated_key(struct btree *b, struct bkey *k)
+void bch_bset_fix_invalidated_key(struct btree_keys *b, struct bkey *k)
{
struct bset_tree *t;
unsigned inorder, j = 1;
- for (t = b->sets; t <= bset_tree_last(b); t++)
+ for (t = b->set; t <= bset_tree_last(b); t++)
if (k < bset_bkey_last(t->data))
goto found_set;
j = j * 2 + 1;
} while (j < t->size);
}
+EXPORT_SYMBOL(bch_bset_fix_invalidated_key);
-static void bch_bset_fix_lookup_table(struct btree *b,
+static void bch_bset_fix_lookup_table(struct btree_keys *b,
struct bset_tree *t,
struct bkey *k)
{
}
}
- if (t->size == b->sets->tree + btree_keys_cachelines(b) - t->tree)
+ if (t->size == b->set->tree + btree_keys_cachelines(b) - t->tree)
return;
/* Possibly add a new entry to the end of the lookup table */
}
}
-void bch_bset_insert(struct btree *b, struct bkey *where,
+void bch_bset_insert(struct btree_keys *b, struct bkey *where,
struct bkey *insert)
{
struct bset_tree *t = bset_tree_last(b);
- BUG_ON(t->data != write_block(b));
+ BUG_ON(!b->last_set_unwritten);
BUG_ON(bset_byte_offset(b, t->data) +
__set_bytes(t->data, t->data->keys + bkey_u64s(insert)) >
PAGE_SIZE << b->page_order);
bkey_copy(where, insert);
bch_bset_fix_lookup_table(b, t, where);
}
+EXPORT_SYMBOL(bch_bset_insert);
struct bset_search_iter {
struct bkey *l, *r;
};
-static struct bset_search_iter bset_search_write_set(struct btree *b,
- struct bset_tree *t,
+static struct bset_search_iter bset_search_write_set(struct bset_tree *t,
const struct bkey *search)
{
unsigned li = 0, ri = t->size;
- BUG_ON(!b->nsets &&
- t->size < bkey_to_cacheline(t, bset_bkey_last(t->data)));
-
while (li + 1 != ri) {
unsigned m = (li + ri) >> 1;
};
}
-static struct bset_search_iter bset_search_tree(struct btree *b,
- struct bset_tree *t,
+static struct bset_search_iter bset_search_tree(struct bset_tree *t,
const struct bkey *search)
{
struct bkey *l, *r;
if (unlikely(!t->size)) {
i.l = t->data->start;
i.r = bset_bkey_last(t->data);
- } else if (bset_written(b, t)) {
+ } else if (bset_written(&b->keys, t)) {
/*
* Each node in the auxiliary search tree covers a certain range
* of bits, and keys above and below the set it covers might
if (unlikely(bkey_cmp(search, t->data->start) < 0))
return t->data->start;
- i = bset_search_tree(b, t, search);
- } else
- i = bset_search_write_set(b, t, search);
+ i = bset_search_tree(t, search);
+ } else {
+ BUG_ON(!b->keys.nsets &&
+ t->size < bkey_to_cacheline(t, bset_bkey_last(t->data)));
+
+ i = bset_search_write_set(t, search);
+ }
if (expensive_debug_checks(b->c)) {
- BUG_ON(bset_written(b, t) &&
+ BUG_ON(bset_written(&b->keys, t) &&
i.l != t->data->start &&
bkey_cmp(tree_to_prev_bkey(t,
inorder_to_tree(bkey_to_cacheline(t, i.l), t)),
return i.l;
}
+EXPORT_SYMBOL(__bch_bset_search);
/* Btree iterator */
iter->b = b;
#endif
- for (; start <= &b->sets[b->nsets]; start++) {
+ for (; start <= bset_tree_last(&b->keys); start++) {
ret = bch_bset_search(b, start, search);
bch_btree_iter_push(iter, ret, bset_bkey_last(start->data));
}
struct btree_iter *iter,
struct bkey *search)
{
- return __bch_btree_iter_init(b, iter, search, b->sets);
+ return __bch_btree_iter_init(b, iter, search, b->keys.set);
}
+EXPORT_SYMBOL(bch_btree_iter_init);
static inline struct bkey *__bch_btree_iter_next(struct btree_iter *iter,
btree_iter_cmp_fn *cmp)
return __bch_btree_iter_next(iter, btree_iter_cmp);
}
+EXPORT_SYMBOL(bch_btree_iter_next);
struct bkey *bch_btree_iter_next_filter(struct btree_iter *iter,
- struct btree *b, ptr_filter_fn fn)
+ struct btree_keys *b, ptr_filter_fn fn)
{
struct bkey *ret;
return 0;
}
+EXPORT_SYMBOL(bch_bset_sort_state_init);
-static void btree_mergesort(struct btree *b, struct bset *out,
+static void btree_mergesort(struct btree_keys *b, struct bset *out,
struct btree_iter *iter,
bool fixup, bool remove_stale)
{
int i;
struct bkey *k, *last = NULL;
BKEY_PADDED(k) tmp;
- bool (*bad)(struct btree *, const struct bkey *) = remove_stale
+ bool (*bad)(struct btree_keys *, const struct bkey *) = remove_stale
? bch_ptr_bad
: bch_ptr_invalid;
pr_debug("sorted %i keys", out->keys);
}
-static void __btree_sort(struct btree *b, struct btree_iter *iter,
+static void __btree_sort(struct btree_keys *b, struct btree_iter *iter,
unsigned start, unsigned order, bool fixup,
struct bset_sort_state *state)
{
out = page_address(mempool_alloc(state->pool, GFP_NOIO));
used_mempool = true;
- order = ilog2(bucket_pages(b->c));
+ order = state->page_order;
}
start_time = local_clock();
* memcpy()
*/
- out->magic = bset_magic(&b->c->sb);
- out->seq = b->sets[0].data->seq;
- out->version = b->sets[0].data->version;
- swap(out, b->sets[0].data);
+ out->magic = b->set->data->magic;
+ out->seq = b->set->data->seq;
+ out->version = b->set->data->version;
+ swap(out, b->set->data);
} else {
- b->sets[start].data->keys = out->keys;
- memcpy(b->sets[start].data->start, out->start,
+ b->set[start].data->keys = out->keys;
+ memcpy(b->set[start].data->start, out->start,
(void *) bset_bkey_last(out) - (void *) out->start);
}
else
free_pages((unsigned long) out, order);
- bset_build_written_tree(b);
+ bch_bset_build_written_tree(b);
if (!start)
bch_time_stats_update(&state->time, start_time);
void bch_btree_sort_partial(struct btree *b, unsigned start,
struct bset_sort_state *state)
{
- size_t order = b->page_order, keys = 0;
+ size_t order = b->keys.page_order, keys = 0;
struct btree_iter iter;
int oldsize = bch_count_data(b);
- __bch_btree_iter_init(b, &iter, NULL, &b->sets[start]);
-
- BUG_ON(!bset_written(b, bset_tree_last(b)) &&
- (bset_tree_last(b)->size || b->nsets));
+ __bch_btree_iter_init(b, &iter, NULL, &b->keys.set[start]);
if (start) {
unsigned i;
- for (i = start; i <= b->nsets; i++)
- keys += b->sets[i].data->keys;
+ for (i = start; i <= b->keys.nsets; i++)
+ keys += b->keys.set[i].data->keys;
- order = roundup_pow_of_two(__set_bytes(b->sets->data,
+ order = roundup_pow_of_two(__set_bytes(b->keys.set->data,
keys)) / PAGE_SIZE;
if (order)
order = ilog2(order);
}
- __btree_sort(b, &iter, start, order, false, state);
+ __btree_sort(&b->keys, &iter, start, order, false, state);
EBUG_ON(b->written && oldsize >= 0 && bch_count_data(b) != oldsize);
}
EXPORT_SYMBOL(bch_btree_sort_partial);
-void bch_btree_sort_and_fix_extents(struct btree *b, struct btree_iter *iter,
+void bch_btree_sort_and_fix_extents(struct btree_keys *b,
+ struct btree_iter *iter,
struct bset_sort_state *state)
{
__btree_sort(b, iter, 0, b->page_order, true, state);
struct btree_iter iter;
bch_btree_iter_init(b, &iter, NULL);
- btree_mergesort(b, new->sets->data, &iter, false, true);
+ btree_mergesort(&b->keys, new->keys.set->data, &iter, false, true);
bch_time_stats_update(&state->time, start_time);
- new->sets->size = 0;
+ new->keys.set->size = 0; // XXX: why?
}
#define SORT_CRIT (4096 / sizeof(uint64_t))
unsigned crit = SORT_CRIT;
int i;
+ b->keys.last_set_unwritten = 0;
+
/* Don't sort if nothing to do */
- if (!b->nsets)
+ if (!b->keys.nsets)
goto out;
- for (i = b->nsets - 1; i >= 0; --i) {
+ for (i = b->keys.nsets - 1; i >= 0; --i) {
crit *= state->crit_factor;
- if (b->sets[i].data->keys < crit) {
+ if (b->keys.set[i].data->keys < crit) {
bch_btree_sort_partial(b, i, state);
return;
}
}
/* Sort if we'd overflow */
- if (b->nsets + 1 == MAX_BSETS) {
+ if (b->keys.nsets + 1 == MAX_BSETS) {
bch_btree_sort(b, state);
return;
}
out:
- bset_build_written_tree(b);
+ bch_bset_build_written_tree(&b->keys);
}
+EXPORT_SYMBOL(bch_btree_sort_lazy);
/* Sysfs stuff */
stats->nodes++;
- for (i = 0; i <= b->nsets; i++) {
- struct bset_tree *t = &b->sets[i];
+ for (i = 0; i <= b->keys.nsets; i++) {
+ struct bset_tree *t = &b->keys.set[i];
size_t bytes = t->data->keys * sizeof(uint64_t);
size_t j;
- if (bset_written(b, t)) {
+ if (bset_written(&b->keys, t)) {
stats->sets_written++;
stats->bytes_written += bytes;
*/
struct btree;
+struct btree_keys;
+struct btree_iter;
+struct btree_iter_set;
struct bkey_float;
#define MAX_BSETS 4U
struct bset *data;
};
+struct btree_keys_ops {
+ bool (*sort_cmp)(struct btree_iter_set,
+ struct btree_iter_set);
+ struct bkey *(*sort_fixup)(struct btree_iter *, struct bkey *);
+ bool (*key_invalid)(struct btree_keys *,
+ const struct bkey *);
+ bool (*key_bad)(struct btree_keys *, const struct bkey *);
+ bool (*key_merge)(struct btree_keys *,
+ struct bkey *, struct bkey *);
+
+ /*
+ * Only used for deciding whether to use START_KEY(k) or just the key
+ * itself in a couple places
+ */
+ bool is_extents;
+};
+
+struct btree_keys {
+ const struct btree_keys_ops *ops;
+ uint8_t page_order;
+ uint8_t nsets;
+ unsigned last_set_unwritten:1;
+ bool *expensive_debug_checks;
+
+ /*
+ * Sets of sorted keys - the real btree node - plus a binary search tree
+ *
+ * set[0] is special; set[0]->tree, set[0]->prev and set[0]->data point
+ * to the memory we have allocated for this btree node. Additionally,
+ * set[0]->data points to the entire btree node as it exists on disk.
+ */
+ struct bset_tree set[MAX_BSETS];
+};
+
+static inline struct bset_tree *bset_tree_last(struct btree_keys *b)
+{
+ return b->set + b->nsets;
+}
+
+static inline bool bset_written(struct btree_keys *b, struct bset_tree *t)
+{
+ return t <= b->set + b->nsets - b->last_set_unwritten;
+}
+
+static inline bool bkey_written(struct btree_keys *b, struct bkey *k)
+{
+ return !b->last_set_unwritten || k < b->set[b->nsets].data->start;
+}
+
+static inline unsigned bset_byte_offset(struct btree_keys *b, struct bset *i)
+{
+ return ((size_t) i) - ((size_t) b->set->data);
+}
+
+static inline unsigned bset_sector_offset(struct btree_keys *b, struct bset *i)
+{
+ return bset_byte_offset(b, i) >> 9;
+}
+
+static inline bool btree_keys_expensive_checks(struct btree_keys *b)
+{
+#ifdef CONFIG_BCACHE_DEBUG
+ return *b->expensive_debug_checks;
+#else
+ return false;
+#endif
+}
+
#define __set_bytes(i, k) (sizeof(*(i)) + (k) * sizeof(uint64_t))
#define set_bytes(i) __set_bytes(i, i->keys)
#define set_blocks(i, block_bytes) \
__set_blocks(i, (i)->keys, block_bytes)
-void bch_btree_keys_free(struct btree *);
-int bch_btree_keys_alloc(struct btree *, unsigned, gfp_t);
+static inline struct bset *bset_next_set(struct btree_keys *b,
+ unsigned block_bytes)
+{
+ struct bset *i = bset_tree_last(b)->data;
+
+ return ((void *) i) + roundup(set_bytes(i), block_bytes);
+}
+
+void bch_btree_keys_free(struct btree_keys *);
+int bch_btree_keys_alloc(struct btree_keys *, unsigned, gfp_t);
+void bch_btree_keys_init(struct btree_keys *, const struct btree_keys_ops *,
+ bool *);
-void bch_bset_fix_invalidated_key(struct btree *, struct bkey *);
-void bch_bset_init_next(struct btree *, struct bset *, uint64_t);
-void bch_bset_insert(struct btree *, struct bkey *, struct bkey *);
+void bch_bset_init_next(struct btree_keys *, struct bset *, uint64_t);
+void bch_bset_build_written_tree(struct btree_keys *);
+void bch_bset_fix_invalidated_key(struct btree_keys *, struct bkey *);
+void bch_bset_insert(struct btree_keys *, struct bkey *, struct bkey *);
+
+/*
+ * Tries to merge l and r: l should be lower than r
+ * Returns true if we were able to merge. If we did merge, l will be the merged
+ * key, r will be untouched.
+ */
+static inline bool bch_bkey_try_merge(struct btree_keys *b,
+ struct bkey *l, struct bkey *r)
+{
+ return b->ops->key_merge ? b->ops->key_merge(b, l, r) : false;
+}
/* Btree key iteration */
} data[MAX_BSETS];
};
-typedef bool (*ptr_filter_fn)(struct btree *, const struct bkey *);
+typedef bool (*ptr_filter_fn)(struct btree_keys *, const struct bkey *);
struct bkey *bch_btree_iter_next(struct btree_iter *);
struct bkey *bch_btree_iter_next_filter(struct btree_iter *,
- struct btree *, ptr_filter_fn);
+ struct btree_keys *, ptr_filter_fn);
void bch_btree_iter_push(struct btree_iter *, struct bkey *, struct bkey *);
struct bkey *bch_btree_iter_init(struct btree *, struct btree_iter *,
void bch_btree_sort_lazy(struct btree *, struct bset_sort_state *);
void bch_btree_sort_into(struct btree *, struct btree *,
struct bset_sort_state *);
-void bch_btree_sort_and_fix_extents(struct btree *, struct btree_iter *,
+void bch_btree_sort_and_fix_extents(struct btree_keys *, struct btree_iter *,
struct bset_sort_state *);
void bch_btree_sort_partial(struct btree *, unsigned,
struct bset_sort_state *);
_ret; \
})
+static inline bool bch_ptr_invalid(struct btree_keys *b, const struct bkey *k)
+{
+ return b->ops->key_invalid(b, k);
+}
+
+static inline bool bch_ptr_bad(struct btree_keys *b, const struct bkey *k)
+{
+ return b->ops->key_bad(b, k);
+}
+
/* Keylists */
struct keylist {
static struct workqueue_struct *btree_io_wq;
-static inline bool should_split(struct btree *b)
-{
- struct bset *i = write_block(b);
- return b->written >= btree_blocks(b) ||
- (b->written + __set_blocks(i, i->keys + 15, block_bytes(b->c))
- > btree_blocks(b));
-}
-
#define insert_lock(s, b) ((b)->level <= (s)->lock)
/*
_r; \
})
+static inline struct bset *write_block(struct btree *b)
+{
+ return ((void *) btree_bset_first(b)) + b->written * block_bytes(b->c);
+}
+
+static inline bool should_split(struct btree *b)
+{
+ struct bset *i = write_block(b);
+ return b->written >= btree_blocks(b) ||
+ (b->written + __set_blocks(i, i->keys + 15, block_bytes(b->c))
+ > btree_blocks(b));
+}
+
/* Btree key manipulation */
void bkey_put(struct cache_set *c, struct bkey *k)
goto err;
for (;
- b->written < btree_blocks(b) && i->seq == b->sets[0].data->seq;
+ b->written < btree_blocks(b) && i->seq == b->keys.set[0].data->seq;
i = write_block(b)) {
err = "unsupported bset version";
if (i->version > BCACHE_BSET_VERSION)
}
err = "empty set";
- if (i != b->sets[0].data && !i->keys)
+ if (i != b->keys.set[0].data && !i->keys)
goto err;
bch_btree_iter_push(iter, i->start, bset_bkey_last(i));
err = "corrupted btree";
for (i = write_block(b);
- bset_sector_offset(b, i) < KEY_SIZE(&b->key);
+ bset_sector_offset(&b->keys, i) < KEY_SIZE(&b->key);
i = ((void *) i) + block_bytes(b->c))
- if (i->seq == b->sets[0].data->seq)
+ if (i->seq == b->keys.set[0].data->seq)
goto err;
- bch_btree_sort_and_fix_extents(b, iter, &b->c->sort);
+ bch_btree_sort_and_fix_extents(&b->keys, iter, &b->c->sort);
- i = b->sets[0].data;
+ i = b->keys.set[0].data;
err = "short btree key";
- if (b->sets[0].size &&
- bkey_cmp(&b->key, &b->sets[0].end) < 0)
+ if (b->keys.set[0].size &&
+ bkey_cmp(&b->key, &b->keys.set[0].end) < 0)
goto err;
if (b->written < btree_blocks(b))
- bch_bset_init_next(b, write_block(b), bset_magic(&b->c->sb));
+ bch_bset_init_next(&b->keys, write_block(b),
+ bset_magic(&b->c->sb));
out:
mempool_free(iter, b->c->fill_iter);
return;
bio->bi_end_io = btree_node_read_endio;
bio->bi_private = &cl;
- bch_bio_map(bio, b->sets[0].data);
+ bch_bio_map(bio, b->keys.set[0].data);
bch_submit_bbio(bio, b->c, &b->key, 0);
closure_sync(&cl);
bkey_copy(&k.key, &b->key);
SET_PTR_OFFSET(&k.key, 0, PTR_OFFSET(&k.key, 0) +
- bset_sector_offset(b, i));
+ bset_sector_offset(&b->keys, i));
if (!bio_alloc_pages(b->bio, GFP_NOIO)) {
int j;
do_btree_node_write(b);
- b->written += set_blocks(i, block_bytes(b->c));
atomic_long_add(set_blocks(i, block_bytes(b->c)) * b->c->sb.block_size,
&PTR_CACHE(b->c, &b->key, 0)->btree_sectors_written);
+ b->written += set_blocks(i, block_bytes(b->c));
+
/* If not a leaf node, always sort */
- if (b->level && b->nsets)
+ if (b->level && b->keys.nsets)
bch_btree_sort(b, &b->c->sort);
else
bch_btree_sort_lazy(b, &b->c->sort);
* do verify if there was more than one set initially (i.e. we did a
* sort) and we sorted down to a single set:
*/
- if (i != b->sets->data && !b->nsets)
+ if (i != b->keys.set->data && !b->keys.nsets)
bch_btree_verify(b);
if (b->written < btree_blocks(b))
- bch_bset_init_next(b, write_block(b), bset_magic(&b->c->sb));
+ bch_bset_init_next(&b->keys, write_block(b),
+ bset_magic(&b->c->sb));
}
static void bch_btree_node_write_sync(struct btree *b)
* mca -> memory cache
*/
-static void mca_reinit(struct btree *b)
-{
- unsigned i;
-
- b->flags = 0;
- b->written = 0;
- b->nsets = 0;
-
- for (i = 0; i < MAX_BSETS; i++)
- b->sets[i].size = 0;
- /*
- * Second loop starts at 1 because b->sets[0]->data is the memory we
- * allocated
- */
- for (i = 1; i < MAX_BSETS; i++)
- b->sets[i].data = NULL;
-}
-
#define mca_reserve(c) (((c->root && c->root->level) \
? c->root->level : 1) * 8 + 16)
#define mca_can_free(c) \
{
BUG_ON(b->io_mutex.count != 1);
- bch_btree_keys_free(b);
+ bch_btree_keys_free(&b->keys);
b->c->bucket_cache_used--;
list_move(&b->list, &b->c->btree_cache_freed);
static void mca_data_alloc(struct btree *b, struct bkey *k, gfp_t gfp)
{
- if (!bch_btree_keys_alloc(b,
+ if (!bch_btree_keys_alloc(&b->keys,
max_t(unsigned,
ilog2(b->c->btree_pages),
btree_order(k)),
if (!down_write_trylock(&b->lock))
return -ENOMEM;
- BUG_ON(btree_node_dirty(b) && !b->sets[0].data);
+ BUG_ON(btree_node_dirty(b) && !b->keys.set[0].data);
- if (b->page_order < min_order)
+ if (b->keys.page_order < min_order)
goto out_unlock;
if (!flush) {
c->verify_data = mca_bucket_alloc(c, &ZERO_KEY, GFP_KERNEL);
if (c->verify_data &&
- c->verify_data->sets[0].data)
+ c->verify_data->keys.set->data)
list_del_init(&c->verify_data->list);
else
c->verify_data = NULL;
list_for_each_entry(b, &c->btree_cache_freed, list)
if (!mca_reap(b, 0, false)) {
mca_data_alloc(b, k, __GFP_NOWARN|GFP_NOIO);
- if (!b->sets[0].data)
+ if (!b->keys.set[0].data)
goto err;
else
goto out;
goto err;
BUG_ON(!down_write_trylock(&b->lock));
- if (!b->sets->data)
+ if (!b->keys.set->data)
goto err;
out:
BUG_ON(b->io_mutex.count != 1);
hlist_add_head_rcu(&b->hash, mca_hash(c, k));
lock_set_subclass(&b->lock.dep_map, level + 1, _THIS_IP_);
- b->level = level;
b->parent = (void *) ~0UL;
+ b->flags = 0;
+ b->written = 0;
+ b->level = level;
if (!b->level)
- b->ops = &bch_extent_keys_ops;
+ bch_btree_keys_init(&b->keys, &bch_extent_keys_ops,
+ &b->c->expensive_debug_checks);
else
- b->ops = &bch_btree_keys_ops;
-
- mca_reinit(b);
+ bch_btree_keys_init(&b->keys, &bch_btree_keys_ops,
+ &b->c->expensive_debug_checks);
return b;
err:
b->accessed = 1;
- for (; i <= b->nsets && b->sets[i].size; i++) {
- prefetch(b->sets[i].tree);
- prefetch(b->sets[i].data);
+ for (; i <= b->keys.nsets && b->keys.set[i].size; i++) {
+ prefetch(b->keys.set[i].tree);
+ prefetch(b->keys.set[i].data);
}
- for (; i <= b->nsets; i++)
- prefetch(b->sets[i].data);
+ for (; i <= b->keys.nsets; i++)
+ prefetch(b->keys.set[i].data);
if (btree_node_io_error(b)) {
rw_unlock(write, b);
}
b->accessed = 1;
- bch_bset_init_next(b, b->sets->data, bset_magic(&b->c->sb));
+ bch_bset_init_next(&b->keys, b->keys.set->data, bset_magic(&b->c->sb));
mutex_unlock(&c->bucket_lock);
stale = max(stale, btree_mark_key(b, k));
keys++;
- if (bch_ptr_bad(b, k))
+ if (bch_ptr_bad(&b->keys, k))
continue;
gc->key_bytes += bkey_u64s(k);
gc->data += KEY_SIZE(k);
}
- for (t = b->sets; t <= &b->sets[b->nsets]; t++)
+ for (t = b->keys.set; t <= &b->keys.set[b->keys.nsets]; t++)
btree_bug_on(t->size &&
- bset_written(b, t) &&
+ bset_written(&b->keys, t) &&
bkey_cmp(&b->key, &t->end) < 0,
b, "found short btree key in gc");
blocks = btree_default_blocks(b->c) * 2 / 3;
if (nodes < 2 ||
- __set_blocks(b->sets[0].data, keys,
+ __set_blocks(b->keys.set[0].data, keys,
block_bytes(b->c)) > blocks * (nodes - 1))
return 0;
r[i].b = ERR_PTR(-EINTR);
while (1) {
- k = bch_btree_iter_next_filter(&iter, b, bch_ptr_bad);
+ k = bch_btree_iter_next_filter(&iter, &b->keys, bch_ptr_bad);
if (k) {
r->b = bch_btree_node_get(b->c, k, b->level - 1, true);
if (IS_ERR(r->b)) {
bch_btree_iter_init(b, &iter, NULL);
do {
- k = bch_btree_iter_next_filter(&iter, b, bch_ptr_bad);
+ k = bch_btree_iter_next_filter(&iter, &b->keys,
+ bch_ptr_bad);
if (k)
btree_node_prefetch(b->c, k, b->level - 1);
subtract_dirty(k, KEY_START(insert), KEY_SIZE(insert));
- if (bkey_written(b, k)) {
+ if (bkey_written(&b->keys, k)) {
/*
* We insert a new key to cover the top of the
* old key, and the old key is modified in place
* depends on us inserting a new key for the top
* here.
*/
- top = bch_bset_search(b, bset_tree_last(b),
+ top = bch_bset_search(b,
+ bset_tree_last(&b->keys),
insert);
- bch_bset_insert(b, top, k);
+ bch_bset_insert(&b->keys, top, k);
} else {
BKEY_PADDED(key) temp;
bkey_copy(&temp.key, k);
- bch_bset_insert(b, k, &temp.key);
+ bch_bset_insert(&b->keys, k, &temp.key);
top = bkey_next(k);
}
bch_cut_front(insert, top);
bch_cut_back(&START_KEY(insert), k);
- bch_bset_fix_invalidated_key(b, k);
+ bch_bset_fix_invalidated_key(&b->keys, k);
return false;
}
if (bkey_cmp(&START_KEY(insert), &START_KEY(k)) > 0)
old_offset = KEY_START(insert);
- if (bkey_written(b, k) &&
+ if (bkey_written(&b->keys, k) &&
bkey_cmp(&START_KEY(insert), &START_KEY(k)) <= 0) {
/*
* Completely overwrote, so we don't have to
bch_cut_front(k, k);
} else {
__bch_cut_back(&START_KEY(insert), k);
- bch_bset_fix_invalidated_key(b, k);
+ bch_bset_fix_invalidated_key(&b->keys, k);
}
}
* the previous key.
*/
prev = NULL;
- m = bch_btree_iter_init(b, &iter, PRECEDING_KEY(&START_KEY(k)));
+ m = bch_btree_iter_init(b, &iter,
+ PRECEDING_KEY(&START_KEY(k)));
if (fix_overlapping_extents(b, k, &iter, replace_key)) {
op->insert_collision = true;
/* prev is in the tree, if we merge we're done */
status = BTREE_INSERT_STATUS_BACK_MERGE;
if (prev &&
- bch_bkey_try_merge(b, prev, k))
+ bch_bkey_try_merge(&b->keys, prev, k))
goto merged;
status = BTREE_INSERT_STATUS_OVERWROTE;
status = BTREE_INSERT_STATUS_FRONT_MERGE;
if (m != bset_bkey_last(i) &&
- bch_bkey_try_merge(b, k, m))
+ bch_bkey_try_merge(&b->keys, k, m))
goto copy;
} else {
BUG_ON(replace_key);
- m = bch_bset_search(b, bset_tree_last(b), k);
+ m = bch_bset_search(b, bset_tree_last(&b->keys), k);
}
-insert: bch_bset_insert(b, m, k);
+insert: bch_bset_insert(&b->keys, m, k);
copy: bkey_copy(m, k);
merged:
bch_check_keys(b, "%u for %s", status,
bch_btree_iter_init(b, &iter, from);
- while ((k = bch_btree_iter_next_filter(&iter, b,
+ while ((k = bch_btree_iter_next_filter(&iter, &b->keys,
bch_ptr_bad))) {
ret = btree(map_nodes_recurse, k, b,
op, from, fn, flags);
bch_btree_iter_init(b, &iter, from);
- while ((k = bch_btree_iter_next_filter(&iter, b, bch_ptr_bad))) {
+ while ((k = bch_btree_iter_next_filter(&iter, &b->keys, bch_ptr_bad))) {
ret = !b->level
? fn(op, b, k)
: btree(map_keys_recurse, k, b, op, from, fn, flags);
int prio_blocked;
};
-struct btree_keys_ops {
- bool (*sort_cmp)(struct btree_iter_set,
- struct btree_iter_set);
- struct bkey *(*sort_fixup)(struct btree_iter *,
- struct bkey *);
- bool (*key_invalid)(struct btree *,
- const struct bkey *);
- bool (*key_bad)(struct btree *,
- const struct bkey *);
- bool (*key_merge)(struct btree *,
- struct bkey *, struct bkey *);
-
-
- /*
- * Only used for deciding whether to use START_KEY(k) or just the key
- * itself in a couple places
- */
- bool is_extents;
-};
-
struct btree {
- const struct btree_keys_ops *ops;
/* Hottest entries first */
struct hlist_node hash;
unsigned long flags;
uint16_t written; /* would be nice to kill */
uint8_t level;
- uint8_t nsets;
- uint8_t page_order;
-
- /*
- * Set of sorted keys - the real btree node - plus a binary search tree
- *
- * sets[0] is special; set[0]->tree, set[0]->prev and set[0]->data point
- * to the memory we have allocated for this btree node. Additionally,
- * set[0]->data points to the entire btree node as it exists on disk.
- */
- struct bset_tree sets[MAX_BSETS];
+
+ struct btree_keys keys;
/* For outstanding btree writes, used as a lock - protects write_idx */
struct closure io;
return b->writes + (btree_node_write_idx(b) ^ 1);
}
-static inline struct bset_tree *bset_tree_last(struct btree *b)
-{
- return b->sets + b->nsets;
-}
-
static inline struct bset *btree_bset_first(struct btree *b)
{
- return b->sets->data;
+ return b->keys.set->data;
}
static inline struct bset *btree_bset_last(struct btree *b)
{
- return bset_tree_last(b)->data;
-}
-
-static inline unsigned bset_byte_offset(struct btree *b, struct bset *i)
-{
- return ((size_t) i) - ((size_t) b->sets->data);
-}
-
-static inline unsigned bset_sector_offset(struct btree *b, struct bset *i)
-{
- return (((void *) i) - ((void *) btree_bset_first(b))) >> 9;
+ return bset_tree_last(&b->keys)->data;
}
static inline unsigned bset_block_offset(struct btree *b, struct bset *i)
{
- return bset_sector_offset(b, i) >> b->c->block_bits;
-}
-
-static inline struct bset *write_block(struct btree *b)
-{
- return ((void *) b->sets[0].data) + b->written * block_bytes(b->c);
-}
-
-static inline bool bset_written(struct btree *b, struct bset_tree *t)
-{
- return t->data < write_block(b);
-}
-
-static inline bool bkey_written(struct btree *b, struct bkey *k)
-{
- return k < write_block(b)->start;
+ return bset_sector_offset(&b->keys, i) >> b->c->block_bits;
}
static inline void set_gc_sectors(struct cache_set *c)
atomic_set(&c->sectors_to_gc, c->sb.bucket_size * c->nbuckets / 16);
}
-static inline bool bch_ptr_invalid(struct btree *b, const struct bkey *k)
-{
- return b->ops->key_invalid(b, k);
-}
-
-static inline bool bch_ptr_bad(struct btree *b, const struct bkey *k)
-{
- return b->ops->key_bad(b, k);
-}
-
-/*
- * Tries to merge l and r: l should be lower than r
- * Returns true if we were able to merge. If we did merge, l will be the merged
- * key, r will be untouched.
- */
-static inline bool bch_bkey_try_merge(struct btree *b,
- struct bkey *l, struct bkey *r)
-{
- return b->ops->key_merge ? b->ops->key_merge(b, l, r) : false;
-}
-
void bkey_put(struct cache_set *c, struct bkey *k);
/* Looping macros */
#define for_each_key_filter(b, k, iter, filter) \
for (bch_btree_iter_init((b), (iter), NULL); \
- ((k) = bch_btree_iter_next_filter((iter), b, filter));)
+ ((k) = bch_btree_iter_next_filter((iter), &(b)->keys, filter));)
#define for_each_key(b, k, iter) \
for (bch_btree_iter_init((b), (iter), NULL); \
unsigned i;
console_lock();
- for (i = 0; i <= b->nsets; i++)
- dump_bset(b, b->sets[i].data,
- bset_block_offset(b, b->sets[i].data));
+ for (i = 0; i <= b->keys.nsets; i++)
+ dump_bset(b, b->keys.set[i].data,
+ bset_block_offset(b, b->keys.set[i].data));
console_unlock();
}
mutex_lock(&b->c->verify_lock);
ondisk = b->c->verify_ondisk;
- sorted = b->c->verify_data->sets->data;
- inmemory = b->sets->data;
+ sorted = b->c->verify_data->keys.set->data;
+ inmemory = b->keys.set->data;
bkey_copy(&v->key, &b->key);
v->written = 0;
v->level = b->level;
- v->ops = b->ops;
+ v->keys.ops = b->keys.ops;
bio = bch_bbio_alloc(b->c);
bio->bi_bdev = PTR_CACHE(b->c, &b->key, 0)->bdev;
memcpy(ondisk, sorted, KEY_SIZE(&v->key) << 9);
bch_btree_node_read_done(v);
- sorted = v->sets->data;
+ sorted = v->keys.set->data;
if (inmemory->keys != sorted->keys ||
memcmp(inmemory->start,
if (p && bkey_cmp(&START_KEY(p), &START_KEY(k)) > 0)
goto bug;
- if (bch_ptr_invalid(b, k))
+ if (bch_ptr_invalid(&b->keys, k))
continue;
err = "Overlapping keys";
if (p && bkey_cmp(p, &START_KEY(k)) > 0)
goto bug;
} else {
- if (bch_ptr_bad(b, k))
+ if (bch_ptr_bad(&b->keys, k))
continue;
err = "Duplicate keys";
return true;
}
-static bool bch_btree_ptr_invalid(struct btree *b, const struct bkey *k)
+static bool bch_btree_ptr_invalid(struct btree_keys *bk, const struct bkey *k)
{
+ struct btree *b = container_of(bk, struct btree, keys);
return __bch_btree_ptr_invalid(b->c, k);
}
return true;
}
-static bool bch_btree_ptr_bad(struct btree *b, const struct bkey *k)
+static bool bch_btree_ptr_bad(struct btree_keys *bk, const struct bkey *k)
{
+ struct btree *b = container_of(bk, struct btree, keys);
unsigned i;
if (!bkey_cmp(k, &ZERO_KEY) ||
!KEY_PTRS(k) ||
- bch_ptr_invalid(b, k))
+ bch_ptr_invalid(bk, k))
return true;
for (i = 0; i < KEY_PTRS(k); i++)
return NULL;
}
-static bool bch_extent_invalid(struct btree *b, const struct bkey *k)
+static bool bch_extent_invalid(struct btree_keys *bk, const struct bkey *k)
{
+ struct btree *b = container_of(bk, struct btree, keys);
char buf[80];
if (!KEY_SIZE(k))
return true;
}
-static bool bch_extent_bad(struct btree *b, const struct bkey *k)
+static bool bch_extent_bad(struct btree_keys *bk, const struct bkey *k)
{
+ struct btree *b = container_of(bk, struct btree, keys);
struct bucket *g;
unsigned i, stale;
if (!KEY_PTRS(k) ||
- bch_extent_invalid(b, k))
+ bch_extent_invalid(bk, k))
return true;
for (i = 0; i < KEY_PTRS(k); i++)
~((uint64_t)1 << 63);
}
-static bool bch_extent_merge(struct btree *b, struct bkey *l, struct bkey *r)
+static bool bch_extent_merge(struct btree_keys *bk, struct bkey *l, struct bkey *r)
{
+ struct btree *b = container_of(bk, struct btree, keys);
unsigned i;
if (key_merging_disabled(b->c))
mutex_lock(&c->bucket_lock);
list_for_each_entry(b, &c->btree_cache, list)
- ret += 1 << (b->page_order + PAGE_SHIFT);
+ ret += 1 << (b->keys.page_order + PAGE_SHIFT);
mutex_unlock(&c->bucket_lock);
return ret;
TP_fast_assign(
__entry->bucket = PTR_BUCKET_NR(b->c, &b->key, 0);
__entry->block = b->written;
- __entry->keys = b->sets[b->nsets].data->keys;
+ __entry->keys = b->keys.set[b->keys.nsets].data->keys;
),
TP_printk("bucket %zu", __entry->bucket)
projects / openwrt / staging / blogic.git / commitdiff