summaryrefslogtreecommitdiff
path: root/libbcachefs/buckets.h
blob: 3b82d7f32120e2552ee1ab88f46df728a0067be8 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
/*
 * Code for manipulating bucket marks for garbage collection.
 *
 * Copyright 2014 Datera, Inc.
 */

#ifndef _BUCKETS_H
#define _BUCKETS_H

#include "buckets_types.h"
#include "super.h"

#define for_each_bucket(b, ca)					\
	for (b = (ca)->buckets + (ca)->mi.first_bucket;		\
	     b < (ca)->buckets + (ca)->mi.nbuckets; b++)

#define bucket_cmpxchg(g, new, expr)				\
({								\
	u64 _v = READ_ONCE((g)->_mark.counter);			\
	struct bucket_mark _old;				\
								\
	do {							\
		(new).counter = _old.counter = _v;		\
		expr;						\
	} while ((_v = cmpxchg(&(g)->_mark.counter,		\
			       _old.counter,			\
			       (new).counter)) != _old.counter);\
	_old;							\
})

/*
 * bucket_gc_gen() returns the difference between the bucket's current gen and
 * the oldest gen of any pointer into that bucket in the btree.
 */

static inline u8 bucket_gc_gen(struct bch_dev *ca, struct bucket *g)
{
	unsigned long r = g - ca->buckets;
	return g->mark.gen - ca->oldest_gens[r];
}

static inline size_t PTR_BUCKET_NR(const struct bch_dev *ca,
				   const struct bch_extent_ptr *ptr)
{
	return sector_to_bucket(ca, ptr->offset);
}

/*
 * Returns 0 if no pointers or device offline - only for tracepoints!
 */
static inline size_t PTR_BUCKET_NR_TRACE(const struct bch_fs *c,
					 const struct bkey_i *k,
					 unsigned ptr)
{
	size_t bucket = 0;
#if 0
	if (bkey_extent_is_data(&k->k)) {
		const struct bch_extent_ptr *ptr;

		extent_for_each_ptr(bkey_i_to_s_c_extent(k), ptr) {
			const struct bch_dev *ca = c->devs[ptr->dev];
			bucket = PTR_BUCKET_NR(ca, ptr);
			break;
		}
	}
#endif
	return bucket;
}

static inline struct bucket *PTR_BUCKET(const struct bch_dev *ca,
					const struct bch_extent_ptr *ptr)
{
	return ca->buckets + PTR_BUCKET_NR(ca, ptr);
}

static inline int gen_cmp(u8 a, u8 b)
{
	return (s8) (a - b);
}

static inline int gen_after(u8 a, u8 b)
{
	int r = gen_cmp(a, b);

	return r > 0 ? r : 0;
}

/**
 * ptr_stale() - check if a pointer points into a bucket that has been
 * invalidated.
 */
static inline u8 ptr_stale(const struct bch_dev *ca,
			   const struct bch_extent_ptr *ptr)
{
	return gen_after(PTR_BUCKET(ca, ptr)->mark.gen, ptr->gen);
}

/* bucket heaps */

static inline bool bucket_min_cmp(struct bucket_heap_entry l,
				  struct bucket_heap_entry r)
{
	return l.val < r.val;
}

static inline bool bucket_max_cmp(struct bucket_heap_entry l,
				  struct bucket_heap_entry r)
{
	return l.val > r.val;
}

static inline void bucket_heap_push(struct bch_dev *ca, struct bucket *g,
				    unsigned long val)
{
	struct bucket_heap_entry new = { g, val };

	if (!heap_full(&ca->heap))
		heap_add(&ca->heap, new, bucket_min_cmp);
	else if (bucket_min_cmp(new, heap_peek(&ca->heap))) {
		ca->heap.data[0] = new;
		heap_sift(&ca->heap, 0, bucket_min_cmp);
	}
}

/* bucket gc marks */

/* The dirty and cached sector counts saturate. If this occurs,
 * reference counting alone will not free the bucket, and a btree
 * GC must be performed. */
#define GC_MAX_SECTORS_USED ((1U << 15) - 1)

static inline bool bucket_unused(struct bucket *g)
{
	return !g->mark.counter;
}

static inline unsigned bucket_sectors_used(struct bucket *g)
{
	return g->mark.dirty_sectors + g->mark.cached_sectors;
}

/* Per device stats: */

struct bch_dev_usage __bch2_dev_usage_read(struct bch_dev *);
struct bch_dev_usage bch2_dev_usage_read(struct bch_dev *);

static inline u64 __dev_buckets_available(struct bch_dev *ca,
					  struct bch_dev_usage stats)
{
	return max_t(s64, 0,
		     ca->mi.nbuckets - ca->mi.first_bucket -
		     stats.buckets_dirty -
		     stats.buckets_alloc -
		     stats.buckets_meta);
}

/*
 * Number of reclaimable buckets - only for use by the allocator thread:
 */
static inline u64 dev_buckets_available(struct bch_dev *ca)
{
	return __dev_buckets_available(ca, bch2_dev_usage_read(ca));
}

static inline u64 __dev_buckets_free(struct bch_dev *ca,
				       struct bch_dev_usage stats)
{
	return __dev_buckets_available(ca, stats) +
		fifo_used(&ca->free[RESERVE_NONE]) +
		fifo_used(&ca->free_inc);
}

static inline u64 dev_buckets_free(struct bch_dev *ca)
{
	return __dev_buckets_free(ca, bch2_dev_usage_read(ca));
}

/* Cache set stats: */

struct bch_fs_usage __bch2_fs_usage_read(struct bch_fs *);
struct bch_fs_usage bch2_fs_usage_read(struct bch_fs *);
void bch2_fs_usage_apply(struct bch_fs *, struct bch_fs_usage *,
			struct disk_reservation *, struct gc_pos);

static inline u64 __bch2_fs_sectors_used(struct bch_fs *c)
{
	struct bch_fs_usage stats = __bch2_fs_usage_read(c);
	u64 reserved = stats.persistent_reserved +
		stats.online_reserved;

	return stats.s[S_COMPRESSED][S_META] +
		stats.s[S_COMPRESSED][S_DIRTY] +
		reserved +
		(reserved >> 7);
}

static inline u64 bch2_fs_sectors_used(struct bch_fs *c)
{
	return min(c->capacity, __bch2_fs_sectors_used(c));
}

static inline bool is_available_bucket(struct bucket_mark mark)
{
	return (!mark.owned_by_allocator &&
		mark.data_type == BUCKET_DATA &&
		!mark.dirty_sectors &&
		!mark.nouse);
}

static inline bool bucket_needs_journal_commit(struct bucket_mark m,
					       u16 last_seq_ondisk)
{
	return m.journal_seq_valid &&
		((s16) m.journal_seq - (s16) last_seq_ondisk > 0);
}

void bch2_bucket_seq_cleanup(struct bch_fs *);

void bch2_invalidate_bucket(struct bch_dev *, struct bucket *);
void bch2_mark_free_bucket(struct bch_dev *, struct bucket *);
void bch2_mark_alloc_bucket(struct bch_dev *, struct bucket *, bool);
void bch2_mark_metadata_bucket(struct bch_dev *, struct bucket *,
			       enum bucket_data_type, bool);

void __bch2_gc_mark_key(struct bch_fs *, struct bkey_s_c, s64, bool,
		       struct bch_fs_usage *);
void bch2_gc_mark_key(struct bch_fs *, struct bkey_s_c, s64, bool);
void bch2_mark_key(struct bch_fs *, struct bkey_s_c, s64, bool,
		  struct gc_pos, struct bch_fs_usage *, u64);

void bch2_recalc_sectors_available(struct bch_fs *);

void bch2_disk_reservation_put(struct bch_fs *,
			      struct disk_reservation *);

#define BCH_DISK_RESERVATION_NOFAIL		(1 << 0)
#define BCH_DISK_RESERVATION_METADATA		(1 << 1)
#define BCH_DISK_RESERVATION_GC_LOCK_HELD	(1 << 2)
#define BCH_DISK_RESERVATION_BTREE_LOCKS_HELD	(1 << 3)

int bch2_disk_reservation_add(struct bch_fs *,
			     struct disk_reservation *,
			     unsigned, int);
int bch2_disk_reservation_get(struct bch_fs *,
			     struct disk_reservation *,
			     unsigned, int);

#endif /* _BUCKETS_H */