[linux-2.6-block.git] / fs / bcachefs / buckets.h

/* SPDX-License-Identifier: GPL-2.0 */
/*
 * Code for manipulating bucket marks for garbage collection.
 *
 * Copyright 2014 Datera, Inc.
 */

#ifndef _BUCKETS_H
#define _BUCKETS_H

#include "buckets_types.h"
#include "extents.h"
#include "sb-members.h"

static inline u64 sector_to_bucket(const struct bch_dev *ca, sector_t s)
{
	return div_u64(s, ca->mi.bucket_size);
}

static inline sector_t bucket_to_sector(const struct bch_dev *ca, size_t b)
{
	return ((sector_t) b) * ca->mi.bucket_size;
}

static inline sector_t bucket_remainder(const struct bch_dev *ca, sector_t s)
{
	u32 remainder;

	div_u64_rem(s, ca->mi.bucket_size, &remainder);
	return remainder;
}

static inline u64 sector_to_bucket_and_offset(const struct bch_dev *ca, sector_t s, u32 *offset)
{
	return div_u64_rem(s, ca->mi.bucket_size, offset);
}

#define for_each_bucket(_b, _buckets)				\
	for (_b = (_buckets)->b + (_buckets)->first_bucket;	\
	     _b < (_buckets)->b + (_buckets)->nbuckets; _b++)

/*
 * Ugly hack alert:
 *
 * We need to cram a spinlock in a single byte, because that's what we have left
 * in struct bucket, and we care about the size of these - during fsck, we need
 * in memory state for every single bucket on every device.
 *
 * We used to do
 *   while (xchg(&b->lock, 1) cpu_relax();
 * but, it turns out not all architectures support xchg on a single byte.
 *
 * So now we use bit_spin_lock(), with fun games since we can't burn a whole
 * ulong for this - we just need to make sure the lock bit always ends up in the
 * first byte.
 */

#if __BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__
#define BUCKET_LOCK_BITNR	0
#else
#define BUCKET_LOCK_BITNR	(BITS_PER_LONG - 1)
#endif

union ulong_byte_assert {
	ulong	ulong;
	u8	byte;
};

static inline void bucket_unlock(struct bucket *b)
{
	BUILD_BUG_ON(!((union ulong_byte_assert) { .ulong = 1UL << BUCKET_LOCK_BITNR }).byte);

	clear_bit_unlock(BUCKET_LOCK_BITNR, (void *) &b->lock);
	wake_up_bit((void *) &b->lock, BUCKET_LOCK_BITNR);
}

static inline void bucket_lock(struct bucket *b)
{
	wait_on_bit_lock((void *) &b->lock, BUCKET_LOCK_BITNR,
			 TASK_UNINTERRUPTIBLE);
}

static inline struct bucket_array *gc_bucket_array(struct bch_dev *ca)
{
	return rcu_dereference_check(ca->buckets_gc,
				     !ca->fs ||
				     percpu_rwsem_is_held(&ca->fs->mark_lock) ||
				     lockdep_is_held(&ca->fs->gc_lock) ||
				     lockdep_is_held(&ca->bucket_lock));
}

static inline struct bucket *gc_bucket(struct bch_dev *ca, size_t b)
{
	struct bucket_array *buckets = gc_bucket_array(ca);

	BUG_ON(!bucket_valid(ca, b));
	return buckets->b + b;
}

static inline struct bucket_gens *bucket_gens(struct bch_dev *ca)
{
	return rcu_dereference_check(ca->bucket_gens,
				     !ca->fs ||
				     percpu_rwsem_is_held(&ca->fs->mark_lock) ||
				     lockdep_is_held(&ca->fs->gc_lock) ||
				     lockdep_is_held(&ca->bucket_lock));
}

static inline u8 *bucket_gen(struct bch_dev *ca, size_t b)
{
	struct bucket_gens *gens = bucket_gens(ca);

	BUG_ON(!bucket_valid(ca, b));
	return gens->b + b;
}

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);
}

static inline struct bpos PTR_BUCKET_POS(const struct bch_dev *ca,
					 const struct bch_extent_ptr *ptr)
{
	return POS(ptr->dev, PTR_BUCKET_NR(ca, ptr));
}

static inline struct bpos PTR_BUCKET_POS_OFFSET(const struct bch_dev *ca,
						const struct bch_extent_ptr *ptr,
						u32 *bucket_offset)
{
	return POS(ptr->dev, sector_to_bucket_and_offset(ca, ptr->offset, bucket_offset));
}

static inline struct bucket *PTR_GC_BUCKET(struct bch_dev *ca,
					   const struct bch_extent_ptr *ptr)
{
	return gc_bucket(ca, PTR_BUCKET_NR(ca, ptr));
}

static inline enum bch_data_type ptr_data_type(const struct bkey *k,
					       const struct bch_extent_ptr *ptr)
{
	if (bkey_is_btree_ptr(k))
		return BCH_DATA_btree;

	return ptr->cached ? BCH_DATA_cached : BCH_DATA_user;
}

static inline s64 ptr_disk_sectors(s64 sectors, struct extent_ptr_decoded p)
{
	EBUG_ON(sectors < 0);

	return crc_is_compressed(p.crc)
		? DIV_ROUND_UP_ULL(sectors * p.crc.compressed_size,
				   p.crc.uncompressed_size)
		: sectors;
}

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;
}

static inline u8 dev_ptr_stale_rcu(struct bch_dev *ca, const struct bch_extent_ptr *ptr)
{
	return gen_after(*bucket_gen(ca, PTR_BUCKET_NR(ca, ptr)), ptr->gen);
}

/**
 * dev_ptr_stale() - check if a pointer points into a bucket that has been
 * invalidated.
 */
static inline u8 dev_ptr_stale(struct bch_dev *ca, const struct bch_extent_ptr *ptr)
{
	rcu_read_lock();
	u8 ret = dev_ptr_stale_rcu(ca, ptr);
	rcu_read_unlock();

	return ret;
}

/* Device usage: */

void bch2_dev_usage_read_fast(struct bch_dev *, struct bch_dev_usage *);
static inline struct bch_dev_usage bch2_dev_usage_read(struct bch_dev *ca)
{
	struct bch_dev_usage ret;

	bch2_dev_usage_read_fast(ca, &ret);
	return ret;
}

void bch2_dev_usage_init(struct bch_dev *);
void bch2_dev_usage_to_text(struct printbuf *, struct bch_dev_usage *);

static inline u64 bch2_dev_buckets_reserved(struct bch_dev *ca, enum bch_watermark watermark)
{
	s64 reserved = 0;

	switch (watermark) {
	case BCH_WATERMARK_NR:
		BUG();
	case BCH_WATERMARK_stripe:
		reserved += ca->mi.nbuckets >> 6;
		fallthrough;
	case BCH_WATERMARK_normal:
		reserved += ca->mi.nbuckets >> 6;
		fallthrough;
	case BCH_WATERMARK_copygc:
		reserved += ca->nr_btree_reserve;
		fallthrough;
	case BCH_WATERMARK_btree:
		reserved += ca->nr_btree_reserve;
		fallthrough;
	case BCH_WATERMARK_btree_copygc:
	case BCH_WATERMARK_reclaim:
	case BCH_WATERMARK_interior_updates:
		break;
	}

	return reserved;
}

static inline u64 dev_buckets_free(struct bch_dev *ca,
				   struct bch_dev_usage usage,
				   enum bch_watermark watermark)
{
	return max_t(s64, 0,
		     usage.d[BCH_DATA_free].buckets -
		     ca->nr_open_buckets -
		     bch2_dev_buckets_reserved(ca, watermark));
}

static inline u64 __dev_buckets_available(struct bch_dev *ca,
					  struct bch_dev_usage usage,
					  enum bch_watermark watermark)
{
	return max_t(s64, 0,
		       usage.d[BCH_DATA_free].buckets
		     + usage.d[BCH_DATA_cached].buckets
		     + usage.d[BCH_DATA_need_gc_gens].buckets
		     + usage.d[BCH_DATA_need_discard].buckets
		     - ca->nr_open_buckets
		     - bch2_dev_buckets_reserved(ca, watermark));
}

static inline u64 dev_buckets_available(struct bch_dev *ca,
					enum bch_watermark watermark)
{
	return __dev_buckets_available(ca, bch2_dev_usage_read(ca), watermark);
}

/* Filesystem usage: */

static inline unsigned __fs_usage_u64s(unsigned nr_replicas)
{
	return sizeof(struct bch_fs_usage) / sizeof(u64) + nr_replicas;
}

static inline unsigned fs_usage_u64s(struct bch_fs *c)
{
	return __fs_usage_u64s(READ_ONCE(c->replicas.nr));
}

static inline unsigned __fs_usage_online_u64s(unsigned nr_replicas)
{
	return sizeof(struct bch_fs_usage_online) / sizeof(u64) + nr_replicas;
}

static inline unsigned fs_usage_online_u64s(struct bch_fs *c)
{
	return __fs_usage_online_u64s(READ_ONCE(c->replicas.nr));
}

static inline unsigned dev_usage_u64s(void)
{
	return sizeof(struct bch_dev_usage) / sizeof(u64);
}

u64 bch2_fs_usage_read_one(struct bch_fs *, u64 *);

struct bch_fs_usage_online *bch2_fs_usage_read(struct bch_fs *);

void bch2_fs_usage_acc_to_base(struct bch_fs *, unsigned);

void bch2_fs_usage_to_text(struct printbuf *,
			   struct bch_fs *, struct bch_fs_usage_online *);

u64 bch2_fs_sectors_used(struct bch_fs *, struct bch_fs_usage_online *);

struct bch_fs_usage_short
bch2_fs_usage_read_short(struct bch_fs *);

void bch2_dev_usage_update(struct bch_fs *, struct bch_dev *,
			   const struct bch_alloc_v4 *,
			   const struct bch_alloc_v4 *, u64, bool);

/* key/bucket marking: */

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]);
}

int bch2_update_replicas(struct bch_fs *, struct bkey_s_c,
			 struct bch_replicas_entry_v1 *, s64,
			 unsigned, bool);
int bch2_update_replicas_list(struct btree_trans *,
			 struct bch_replicas_entry_v1 *, s64);
int bch2_update_cached_sectors_list(struct btree_trans *, unsigned, s64);
int bch2_replicas_deltas_realloc(struct btree_trans *, unsigned);

void bch2_fs_usage_initialize(struct bch_fs *);

int bch2_bucket_ref_update(struct btree_trans *, struct bch_dev *,
			   struct bkey_s_c, const struct bch_extent_ptr *,
			   s64, enum bch_data_type, u8, u8, u32 *);

int bch2_check_fix_ptrs(struct btree_trans *,
			enum btree_id, unsigned, struct bkey_s_c,
			enum btree_iter_update_trigger_flags);

int bch2_trigger_extent(struct btree_trans *, enum btree_id, unsigned,
			struct bkey_s_c, struct bkey_s,
			enum btree_iter_update_trigger_flags);
int bch2_trigger_reservation(struct btree_trans *, enum btree_id, unsigned,
			  struct bkey_s_c, struct bkey_s,
			  enum btree_iter_update_trigger_flags);

#define trigger_run_overwrite_then_insert(_fn, _trans, _btree_id, _level, _old, _new, _flags)\
({												\
	int ret = 0;										\
												\
	if (_old.k->type)									\
		ret = _fn(_trans, _btree_id, _level, _old, _flags & ~BTREE_TRIGGER_insert);	\
	if (!ret && _new.k->type)								\
		ret = _fn(_trans, _btree_id, _level, _new.s_c, _flags & ~BTREE_TRIGGER_overwrite);\
	ret;											\
})

void bch2_trans_account_disk_usage_change(struct btree_trans *);

void bch2_trans_fs_usage_revert(struct btree_trans *, struct replicas_delta_list *);
int bch2_trans_fs_usage_apply(struct btree_trans *, struct replicas_delta_list *);

int bch2_trans_mark_metadata_bucket(struct btree_trans *, struct bch_dev *, u64,
				    enum bch_data_type, unsigned,
				    enum btree_iter_update_trigger_flags);
int bch2_trans_mark_dev_sb(struct bch_fs *, struct bch_dev *,
				    enum btree_iter_update_trigger_flags);
int bch2_trans_mark_dev_sbs_flags(struct bch_fs *,
				    enum btree_iter_update_trigger_flags);
int bch2_trans_mark_dev_sbs(struct bch_fs *);

static inline bool 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;
	}

	return false;
}

static inline const char *bch2_data_type_str(enum bch_data_type type)
{
	return type < BCH_DATA_NR
		? __bch2_data_types[type]
		: "(invalid data type)";
}

/* disk reservations: */

static inline void bch2_disk_reservation_put(struct bch_fs *c,
					     struct disk_reservation *res)
{
	if (res->sectors) {
		this_cpu_sub(*c->online_reserved, res->sectors);
		res->sectors = 0;
	}
}

#define BCH_DISK_RESERVATION_NOFAIL		(1 << 0)

int __bch2_disk_reservation_add(struct bch_fs *,
				struct disk_reservation *,
				u64, int);

static inline int bch2_disk_reservation_add(struct bch_fs *c, struct disk_reservation *res,
					    u64 sectors, int flags)
{
#ifdef __KERNEL__
	u64 old, new;

	do {
		old = this_cpu_read(c->pcpu->sectors_available);
		if (sectors > old)
			return __bch2_disk_reservation_add(c, res, sectors, flags);

		new = old - sectors;
	} while (this_cpu_cmpxchg(c->pcpu->sectors_available, old, new) != old);

	this_cpu_add(*c->online_reserved, sectors);
	res->sectors			+= sectors;
	return 0;
#else
	return __bch2_disk_reservation_add(c, res, sectors, flags);
#endif
}

static inline struct disk_reservation
bch2_disk_reservation_init(struct bch_fs *c, unsigned nr_replicas)
{
	return (struct disk_reservation) {
		.sectors	= 0,
#if 0
		/* not used yet: */
		.gen		= c->capacity_gen,
#endif
		.nr_replicas	= nr_replicas,
	};
}

static inline int bch2_disk_reservation_get(struct bch_fs *c,
					    struct disk_reservation *res,
					    u64 sectors, unsigned nr_replicas,
					    int flags)
{
	*res = bch2_disk_reservation_init(c, nr_replicas);

	return bch2_disk_reservation_add(c, res, sectors * nr_replicas, flags);
}

#define RESERVE_FACTOR	6

static inline u64 avail_factor(u64 r)
{
	return div_u64(r << RESERVE_FACTOR, (1 << RESERVE_FACTOR) + 1);
}

void bch2_buckets_nouse_free(struct bch_fs *);
int bch2_buckets_nouse_alloc(struct bch_fs *);

int bch2_dev_buckets_resize(struct bch_fs *, struct bch_dev *, u64);
void bch2_dev_buckets_free(struct bch_dev *);
int bch2_dev_buckets_alloc(struct bch_fs *, struct bch_dev *);

#endif /* _BUCKETS_H */
Commit	Line	Data
1c6fdbd8 KO	1	/* SPDX-License-Identifier: GPL-2.0 */
	2	/*
	3	* Code for manipulating bucket marks for garbage collection.
	4	*
	5	* Copyright 2014 Datera, Inc.
	6	*/
	7
	8	#ifndef _BUCKETS_H
	9	#define _BUCKETS_H
	10
	11	#include "buckets_types.h"
f0ac7df2	12	#include "extents.h"
fb8e5b4c KO	13	#include "sb-members.h"
fb8e5b4c KO	14
d155272b	15	static inline u64 sector_to_bucket(const struct bch_dev *ca, sector_t s)
fb8e5b4c KO	16	{
	17	return div_u64(s, ca->mi.bucket_size);
	18	}
	19
	20	static inline sector_t bucket_to_sector(const struct bch_dev *ca, size_t b)
	21	{
	22	return ((sector_t) b) * ca->mi.bucket_size;
	23	}
	24
	25	static inline sector_t bucket_remainder(const struct bch_dev *ca, sector_t s)
	26	{
	27	u32 remainder;
	28
	29	div_u64_rem(s, ca->mi.bucket_size, &remainder);
	30	return remainder;
	31	}
	32
d155272b	33	static inline u64 sector_to_bucket_and_offset(const struct bch_dev ca, sector_t s, u32 offset)
fb8e5b4c KO	34	{
	35	return div_u64_rem(s, ca->mi.bucket_size, offset);
	36	}
1c6fdbd8 KO	37
	38	#define for_each_bucket(_b, _buckets) \
	39	for (_b = (_buckets)->b + (_buckets)->first_bucket; \
	40	_b < (_buckets)->b + (_buckets)->nbuckets; _b++)
	41
8c2d82a6 KO	42	/*
	43	* Ugly hack alert:
	44	*
	45	* We need to cram a spinlock in a single byte, because that's what we have left
	46	* in struct bucket, and we care about the size of these - during fsck, we need
	47	* in memory state for every single bucket on every device.
	48	*
	49	* We used to do
	50	* while (xchg(&b->lock, 1) cpu_relax();
	51	* but, it turns out not all architectures support xchg on a single byte.
	52	*
	53	* So now we use bit_spin_lock(), with fun games since we can't burn a whole
	54	* ulong for this - we just need to make sure the lock bit always ends up in the
	55	* first byte.
	56	*/
	57
	58	#if __BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__
	59	#define BUCKET_LOCK_BITNR 0
	60	#else
	61	#define BUCKET_LOCK_BITNR (BITS_PER_LONG - 1)
	62	#endif
	63
	64	union ulong_byte_assert {
	65	ulong ulong;
	66	u8 byte;
	67	};
	68
66d90823 KO	69	static inline void bucket_unlock(struct bucket *b)
66d90823 KO	70	{
8c2d82a6	71	BUILD_BUG_ON(!((union ulong_byte_assert) { .ulong = 1UL << BUCKET_LOCK_BITNR }).byte);
73bbeaa2 KO	72
	73	clear_bit_unlock(BUCKET_LOCK_BITNR, (void *) &b->lock);
	74	wake_up_bit((void *) &b->lock, BUCKET_LOCK_BITNR);
66d90823 KO	75	}
	76
	77	static inline void bucket_lock(struct bucket *b)
	78	{
73bbeaa2 KO	79	wait_on_bit_lock((void *) &b->lock, BUCKET_LOCK_BITNR,
73bbeaa2 KO	80	TASK_UNINTERRUPTIBLE);
66d90823	81	}
1c6fdbd8	82
5735608c	83	static inline struct bucket_array gc_bucket_array(struct bch_dev ca)
1c6fdbd8	84	{
5735608c	85	return rcu_dereference_check(ca->buckets_gc,
1c6fdbd8	86	!ca->fs \|\|
9166b41d	87	percpu_rwsem_is_held(&ca->fs->mark_lock) \|\|
1c6fdbd8 KO	88	lockdep_is_held(&ca->fs->gc_lock) \|\|
	89	lockdep_is_held(&ca->bucket_lock));
	90	}
	91
5735608c	92	static inline struct bucket gc_bucket(struct bch_dev ca, size_t b)
1c6fdbd8	93	{
5735608c	94	struct bucket_array *buckets = gc_bucket_array(ca);
1c6fdbd8	95
d155272b	96	BUG_ON(!bucket_valid(ca, b));
1c6fdbd8 KO	97	return buckets->b + b;
	98	}
	99
a7860877 KO	100	static inline struct bucket_gens bucket_gens(struct bch_dev ca)
	101	{
	102	return rcu_dereference_check(ca->bucket_gens,
	103	!ca->fs \|\|
	104	percpu_rwsem_is_held(&ca->fs->mark_lock) \|\|
	105	lockdep_is_held(&ca->fs->gc_lock) \|\|
	106	lockdep_is_held(&ca->bucket_lock));
a7860877 KO	107	}
	108
	109	static inline u8 bucket_gen(struct bch_dev ca, size_t b)
	110	{
	111	struct bucket_gens *gens = bucket_gens(ca);
	112
d155272b	113	BUG_ON(!bucket_valid(ca, b));
a7860877 KO	114	return gens->b + b;
	115	}
	116
1c6fdbd8 KO	117	static inline size_t PTR_BUCKET_NR(const struct bch_dev *ca,
	118	const struct bch_extent_ptr *ptr)
	119	{
	120	return sector_to_bucket(ca, ptr->offset);
	121	}
	122
1f2f92ec KO	123	static inline struct bpos PTR_BUCKET_POS(const struct bch_dev *ca,
1f2f92ec KO	124	const struct bch_extent_ptr *ptr)
3d48a7f8	125	{
3d48a7f8 KO	126	return POS(ptr->dev, PTR_BUCKET_NR(ca, ptr));
	127	}
	128
1f2f92ec	129	static inline struct bpos PTR_BUCKET_POS_OFFSET(const struct bch_dev *ca,
a8c752bb KO	130	const struct bch_extent_ptr *ptr,
	131	u32 *bucket_offset)
	132	{
a8c752bb KO	133	return POS(ptr->dev, sector_to_bucket_and_offset(ca, ptr->offset, bucket_offset));
	134	}
	135
47ac34ec KO	136	static inline struct bucket PTR_GC_BUCKET(struct bch_dev ca,
47ac34ec KO	137	const struct bch_extent_ptr *ptr)
1c6fdbd8	138	{
47ac34ec	139	return gc_bucket(ca, PTR_BUCKET_NR(ca, ptr));
1c6fdbd8 KO	140	}
1c6fdbd8 KO	141
06ab329c KO	142	static inline enum bch_data_type ptr_data_type(const struct bkey *k,
	143	const struct bch_extent_ptr *ptr)
	144	{
f0ac7df2	145	if (bkey_is_btree_ptr(k))
89fd25be	146	return BCH_DATA_btree;
06ab329c	147
89fd25be	148	return ptr->cached ? BCH_DATA_cached : BCH_DATA_user;
06ab329c KO	149	}
06ab329c KO	150
a8c752bb KO	151	static inline s64 ptr_disk_sectors(s64 sectors, struct extent_ptr_decoded p)
	152	{
	153	EBUG_ON(sectors < 0);
	154
	155	return crc_is_compressed(p.crc)
	156	? DIV_ROUND_UP_ULL(sectors * p.crc.compressed_size,
	157	p.crc.uncompressed_size)
	158	: sectors;
	159	}
	160
1c6fdbd8 KO	161	static inline int gen_cmp(u8 a, u8 b)
	162	{
	163	return (s8) (a - b);
	164	}
	165
	166	static inline int gen_after(u8 a, u8 b)
	167	{
	168	int r = gen_cmp(a, b);
	169
	170	return r > 0 ? r : 0;
	171	}
	172
3858aa42 KO	173	static inline u8 dev_ptr_stale_rcu(struct bch_dev ca, const struct bch_extent_ptr ptr)
	174	{
	175	return gen_after(*bucket_gen(ca, PTR_BUCKET_NR(ca, ptr)), ptr->gen);
	176	}
	177
1c6fdbd8	178	/**
3858aa42	179	* dev_ptr_stale() - check if a pointer points into a bucket that has been
1c6fdbd8 KO	180	* invalidated.
1c6fdbd8 KO	181	*/
3858aa42	182	static inline u8 dev_ptr_stale(struct bch_dev ca, const struct bch_extent_ptr ptr)
1c6fdbd8	183	{
4b674b09	184	rcu_read_lock();
3858aa42	185	u8 ret = dev_ptr_stale_rcu(ca, ptr);
4b674b09 KO	186	rcu_read_unlock();
	187
	188	return ret;
1c6fdbd8 KO	189	}
1c6fdbd8 KO	190
1c6fdbd8 KO	191	/* Device usage: */
1c6fdbd8 KO	192
ed80c569 KO	193	void bch2_dev_usage_read_fast(struct bch_dev , struct bch_dev_usage );
	194	static inline struct bch_dev_usage bch2_dev_usage_read(struct bch_dev *ca)
	195	{
	196	struct bch_dev_usage ret;
	197
	198	bch2_dev_usage_read_fast(ca, &ret);
	199	return ret;
	200	}
	201
822835ff	202	void bch2_dev_usage_init(struct bch_dev *);
ed0cd515	203	void bch2_dev_usage_to_text(struct printbuf , struct bch_dev_usage );
1c6fdbd8	204
e53a961c	205	static inline u64 bch2_dev_buckets_reserved(struct bch_dev *ca, enum bch_watermark watermark)
f25d8215	206	{
f25d8215 KO	207	s64 reserved = 0;
f25d8215 KO	208
e53a961c KO	209	switch (watermark) {
e53a961c KO	210	case BCH_WATERMARK_NR:
3764647b	211	BUG();
e53a961c	212	case BCH_WATERMARK_stripe:
e84face6 KO	213	reserved += ca->mi.nbuckets >> 6;
e84face6 KO	214	fallthrough;
e53a961c	215	case BCH_WATERMARK_normal:
f25d8215 KO	216	reserved += ca->mi.nbuckets >> 6;
f25d8215 KO	217	fallthrough;
e53a961c	218	case BCH_WATERMARK_copygc:
f25d8215 KO	219	reserved += ca->nr_btree_reserve;
f25d8215 KO	220	fallthrough;
e53a961c	221	case BCH_WATERMARK_btree:
f25d8215 KO	222	reserved += ca->nr_btree_reserve;
f25d8215 KO	223	fallthrough;
e53a961c	224	case BCH_WATERMARK_btree_copygc:
494036d8	225	case BCH_WATERMARK_reclaim:
e2a316b3	226	case BCH_WATERMARK_interior_updates:
f25d8215	227	break;
f25d8215	228	}
1c6fdbd8	229
822835ff KO	230	return reserved;
822835ff KO	231	}
1c6fdbd8	232
30f0349d KO	233	static inline u64 dev_buckets_free(struct bch_dev *ca,
30f0349d KO	234	struct bch_dev_usage usage,
e53a961c	235	enum bch_watermark watermark)
30f0349d KO	236	{
	237	return max_t(s64, 0,
	238	usage.d[BCH_DATA_free].buckets -
	239	ca->nr_open_buckets -
e53a961c	240	bch2_dev_buckets_reserved(ca, watermark));
30f0349d KO	241	}
30f0349d KO	242
822835ff KO	243	static inline u64 __dev_buckets_available(struct bch_dev *ca,
822835ff KO	244	struct bch_dev_usage usage,
e53a961c	245	enum bch_watermark watermark)
822835ff	246	{
f25d8215	247	return max_t(s64, 0,
58aaa083 DH	248	usage.d[BCH_DATA_free].buckets
	249	+ usage.d[BCH_DATA_cached].buckets
	250	+ usage.d[BCH_DATA_need_gc_gens].buckets
	251	+ usage.d[BCH_DATA_need_discard].buckets
	252	- ca->nr_open_buckets
e53a961c	253	- bch2_dev_buckets_reserved(ca, watermark));
1c6fdbd8 KO	254	}
1c6fdbd8 KO	255
f25d8215	256	static inline u64 dev_buckets_available(struct bch_dev *ca,
e53a961c	257	enum bch_watermark watermark)
1c6fdbd8	258	{
e53a961c	259	return __dev_buckets_available(ca, bch2_dev_usage_read(ca), watermark);
1c6fdbd8 KO	260	}
	261
	262	/* Filesystem usage: */
	263
962210b2 KO	264	static inline unsigned __fs_usage_u64s(unsigned nr_replicas)
	265	{
	266	return sizeof(struct bch_fs_usage) / sizeof(u64) + nr_replicas;
	267	}
	268
ecf37a4a	269	static inline unsigned fs_usage_u64s(struct bch_fs *c)
7ef2a73a	270	{
962210b2 KO	271	return __fs_usage_u64s(READ_ONCE(c->replicas.nr));
	272	}
	273
	274	static inline unsigned __fs_usage_online_u64s(unsigned nr_replicas)
	275	{
	276	return sizeof(struct bch_fs_usage_online) / sizeof(u64) + nr_replicas;
	277	}
	278
	279	static inline unsigned fs_usage_online_u64s(struct bch_fs *c)
	280	{
	281	return __fs_usage_online_u64s(READ_ONCE(c->replicas.nr));
ecf37a4a	282	}
7ef2a73a	283
180fb49d KO	284	static inline unsigned dev_usage_u64s(void)
	285	{
	286	return sizeof(struct bch_dev_usage) / sizeof(u64);
	287	}
	288
5e82a9a1	289	u64 bch2_fs_usage_read_one(struct bch_fs , u64 );
1c6fdbd8	290
5e82a9a1 KO	291	struct bch_fs_usage_online bch2_fs_usage_read(struct bch_fs );
	292
	293	void bch2_fs_usage_acc_to_base(struct bch_fs *, unsigned);
	294
	295	void bch2_fs_usage_to_text(struct printbuf *,
	296	struct bch_fs , struct bch_fs_usage_online );
	297
	298	u64 bch2_fs_sectors_used(struct bch_fs , struct bch_fs_usage_online );
1c6fdbd8	299
5663a415 KO	300	struct bch_fs_usage_short
5663a415 KO	301	bch2_fs_usage_read_short(struct bch_fs *);
b092dadd	302
6820ac2c KO	303	void bch2_dev_usage_update(struct bch_fs , struct bch_dev ,
	304	const struct bch_alloc_v4 *,
	305	const struct bch_alloc_v4 *, u64, bool);
	306
5663a415 KO	307	/* key/bucket marking: */
5663a415 KO	308
4dc5bb9a KO	309	static inline struct bch_fs_usage fs_usage_ptr(struct bch_fs c,
	310	unsigned journal_seq,
	311	bool gc)
	312	{
	313	percpu_rwsem_assert_held(&c->mark_lock);
	314	BUG_ON(!gc && !journal_seq);
	315
	316	return this_cpu_ptr(gc
	317	? c->usage_gc
	318	: c->usage[journal_seq & JOURNAL_BUF_MASK]);
	319	}
	320
f4f78779 KO	321	int bch2_update_replicas(struct bch_fs *, struct bkey_s_c,
	322	struct bch_replicas_entry_v1 *, s64,
	323	unsigned, bool);
25f64e99 KO	324	int bch2_update_replicas_list(struct btree_trans *,
	325	struct bch_replicas_entry_v1 *, s64);
	326	int bch2_update_cached_sectors_list(struct btree_trans *, unsigned, s64);
4dc5bb9a KO	327	int bch2_replicas_deltas_realloc(struct btree_trans *, unsigned);
4dc5bb9a KO	328
3e0745e2	329	void bch2_fs_usage_initialize(struct bch_fs *);
1c6fdbd8	330
07d7c4da KO	331	int bch2_bucket_ref_update(struct btree_trans , struct bch_dev ,
07d7c4da KO	332	struct bkey_s_c, const struct bch_extent_ptr *,
70e3e039	333	s64, enum bch_data_type, u8, u8, u32 *);
f4f78779	334
f40d13f9 KO	335	int bch2_check_fix_ptrs(struct btree_trans *,
	336	enum btree_id, unsigned, struct bkey_s_c,
	337	enum btree_iter_update_trigger_flags);
	338
4f9ec59f	339	int bch2_trigger_extent(struct btree_trans *, enum btree_id, unsigned,
5dd8c60e KO	340	struct bkey_s_c, struct bkey_s,
5dd8c60e KO	341	enum btree_iter_update_trigger_flags);
6cacd0c4	342	int bch2_trigger_reservation(struct btree_trans *, enum btree_id, unsigned,
5dd8c60e KO	343	struct bkey_s_c, struct bkey_s,
5dd8c60e KO	344	enum btree_iter_update_trigger_flags);
880e2275	345
282e7c37	346	#define trigger_run_overwrite_then_insert(_fn, _trans, _btree_id, _level, _old, _new, _flags)\
523f33ef KO	347	({ \
	348	int ret = 0; \
	349	\
	350	if (_old.k->type) \
5dd8c60e	351	ret = _fn(_trans, _btree_id, _level, _old, _flags & ~BTREE_TRIGGER_insert); \
523f33ef	352	if (!ret && _new.k->type) \
5dd8c60e	353	ret = _fn(_trans, _btree_id, _level, _new.s_c, _flags & ~BTREE_TRIGGER_overwrite);\
523f33ef KO	354	ret; \
	355	})
	356
5b14ce35 KO	357	void bch2_trans_account_disk_usage_change(struct btree_trans *);
5b14ce35 KO	358
920e69bc	359	void bch2_trans_fs_usage_revert(struct btree_trans , struct replicas_delta_list );
502cfb35	360	int bch2_trans_fs_usage_apply(struct btree_trans , struct replicas_delta_list );
4d8100da	361
c281db0f	362	int bch2_trans_mark_metadata_bucket(struct btree_trans , struct bch_dev , u64,
5dd8c60e KO	363	enum bch_data_type, unsigned,
	364	enum btree_iter_update_trigger_flags);
	365	int bch2_trans_mark_dev_sb(struct bch_fs , struct bch_dev ,
	366	enum btree_iter_update_trigger_flags);
	367	int bch2_trans_mark_dev_sbs_flags(struct bch_fs *,
	368	enum btree_iter_update_trigger_flags);
bbe682c7	369	int bch2_trans_mark_dev_sbs(struct bch_fs *);
bfcf840d	370
fb8e5b4c KO	371	static inline bool is_superblock_bucket(struct bch_dev *ca, u64 b)
	372	{
	373	struct bch_sb_layout *layout = &ca->disk_sb.sb->layout;
	374	u64 b_offset = bucket_to_sector(ca, b);
	375	u64 b_end = bucket_to_sector(ca, b + 1);
	376	unsigned i;
	377
	378	if (!b)
	379	return true;
	380
	381	for (i = 0; i < layout->nr_superblocks; i++) {
	382	u64 offset = le64_to_cpu(layout->sb_offset[i]);
	383	u64 end = offset + (1 << layout->sb_max_size_bits);
	384
	385	if (!(offset >= b_end \|\| end <= b_offset))
	386	return true;
	387	}
	388
	389	return false;
	390	}
	391
e58f963c KO	392	static inline const char *bch2_data_type_str(enum bch_data_type type)
	393	{
	394	return type < BCH_DATA_NR
	395	? __bch2_data_types[type]
	396	: "(invalid data type)";
	397	}
	398
5663a415	399	/* disk reservations: */
1c6fdbd8	400
1c6fdbd8 KO	401	static inline void bch2_disk_reservation_put(struct bch_fs *c,
	402	struct disk_reservation *res)
	403	{
7e94eeff KO	404	if (res->sectors) {
	405	this_cpu_sub(*c->online_reserved, res->sectors);
	406	res->sectors = 0;
	407	}
1c6fdbd8 KO	408	}
	409
	410	#define BCH_DISK_RESERVATION_NOFAIL (1 << 0)
1c6fdbd8	411
7e94eeff KO	412	int __bch2_disk_reservation_add(struct bch_fs *,
	413	struct disk_reservation *,
	414	u64, int);
	415
	416	static inline int bch2_disk_reservation_add(struct bch_fs c, struct disk_reservation res,
	417	u64 sectors, int flags)
	418	{
b2d1d56b	419	#ifdef __KERNEL__
7e94eeff KO	420	u64 old, new;
	421
	422	do {
	423	old = this_cpu_read(c->pcpu->sectors_available);
	424	if (sectors > old)
	425	return __bch2_disk_reservation_add(c, res, sectors, flags);
	426
	427	new = old - sectors;
	428	} while (this_cpu_cmpxchg(c->pcpu->sectors_available, old, new) != old);
	429
	430	this_cpu_add(*c->online_reserved, sectors);
	431	res->sectors += sectors;
	432	return 0;
b2d1d56b KO	433	#else
	434	return __bch2_disk_reservation_add(c, res, sectors, flags);
	435	#endif
7e94eeff	436	}
1c6fdbd8 KO	437
	438	static inline struct disk_reservation
	439	bch2_disk_reservation_init(struct bch_fs *c, unsigned nr_replicas)
	440	{
	441	return (struct disk_reservation) {
	442	.sectors = 0,
	443	#if 0
	444	/* not used yet: */
	445	.gen = c->capacity_gen,
	446	#endif
	447	.nr_replicas = nr_replicas,
	448	};
	449	}
	450
	451	static inline int bch2_disk_reservation_get(struct bch_fs *c,
	452	struct disk_reservation *res,
cd9f3dfe	453	u64 sectors, unsigned nr_replicas,
1c6fdbd8 KO	454	int flags)
	455	{
	456	*res = bch2_disk_reservation_init(c, nr_replicas);
	457
	458	return bch2_disk_reservation_add(c, res, sectors * nr_replicas, flags);
	459	}
	460
ed343411 DR	461	#define RESERVE_FACTOR 6
	462
	463	static inline u64 avail_factor(u64 r)
	464	{
	465	return div_u64(r << RESERVE_FACTOR, (1 << RESERVE_FACTOR) + 1);
	466	}
	467
ffcbec60 KO	468	void bch2_buckets_nouse_free(struct bch_fs *);
	469	int bch2_buckets_nouse_alloc(struct bch_fs *);
	470
1c6fdbd8 KO	471	int bch2_dev_buckets_resize(struct bch_fs , struct bch_dev , u64);
	472	void bch2_dev_buckets_free(struct bch_dev *);
	473	int bch2_dev_buckets_alloc(struct bch_fs , struct bch_dev );
	474
	475	#endif /* _BUCKETS_H */