linux/fs/bcachefs/trace.h

1205 lines
30 KiB
C
Raw Normal View History

/* SPDX-License-Identifier: GPL-2.0 */
#undef TRACE_SYSTEM
#define TRACE_SYSTEM bcachefs
#if !defined(_TRACE_BCACHEFS_H) || defined(TRACE_HEADER_MULTI_READ)
#define _TRACE_BCACHEFS_H
#include <linux/tracepoint.h>
#define TRACE_BPOS_entries(name) \
__field(u64, name##_inode ) \
__field(u64, name##_offset ) \
__field(u32, name##_snapshot )
#define TRACE_BPOS_assign(dst, src) \
__entry->dst##_inode = (src).inode; \
__entry->dst##_offset = (src).offset; \
__entry->dst##_snapshot = (src).snapshot
DECLARE_EVENT_CLASS(bpos,
TP_PROTO(const struct bpos *p),
TP_ARGS(p),
TP_STRUCT__entry(
TRACE_BPOS_entries(p)
),
TP_fast_assign(
TRACE_BPOS_assign(p, *p);
),
TP_printk("%llu:%llu:%u", __entry->p_inode, __entry->p_offset, __entry->p_snapshot)
);
DECLARE_EVENT_CLASS(bkey,
TP_PROTO(const struct bkey *k),
TP_ARGS(k),
TP_STRUCT__entry(
__field(u64, inode )
__field(u64, offset )
__field(u32, size )
),
TP_fast_assign(
__entry->inode = k->p.inode;
__entry->offset = k->p.offset;
__entry->size = k->size;
),
TP_printk("%llu:%llu len %u", __entry->inode,
__entry->offset, __entry->size)
);
DECLARE_EVENT_CLASS(btree_node,
TP_PROTO(struct bch_fs *c, struct btree *b),
TP_ARGS(c, b),
TP_STRUCT__entry(
__field(dev_t, dev )
__field(u8, level )
__field(u8, btree_id )
TRACE_BPOS_entries(pos)
),
TP_fast_assign(
__entry->dev = c->dev;
__entry->level = b->c.level;
__entry->btree_id = b->c.btree_id;
TRACE_BPOS_assign(pos, b->key.k.p);
),
TP_printk("%d,%d %u %s %llu:%llu:%u",
MAJOR(__entry->dev), MINOR(__entry->dev),
__entry->level,
bch2_btree_ids[__entry->btree_id],
__entry->pos_inode, __entry->pos_offset, __entry->pos_snapshot)
);
DECLARE_EVENT_CLASS(bch_fs,
TP_PROTO(struct bch_fs *c),
TP_ARGS(c),
TP_STRUCT__entry(
__field(dev_t, dev )
),
TP_fast_assign(
__entry->dev = c->dev;
),
TP_printk("%d,%d", MAJOR(__entry->dev), MINOR(__entry->dev))
);
DECLARE_EVENT_CLASS(bio,
TP_PROTO(struct bio *bio),
TP_ARGS(bio),
TP_STRUCT__entry(
__field(dev_t, dev )
__field(sector_t, sector )
__field(unsigned int, nr_sector )
__array(char, rwbs, 6 )
),
TP_fast_assign(
__entry->dev = bio->bi_bdev ? bio_dev(bio) : 0;
__entry->sector = bio->bi_iter.bi_sector;
__entry->nr_sector = bio->bi_iter.bi_size >> 9;
blk_fill_rwbs(__entry->rwbs, bio->bi_opf);
),
TP_printk("%d,%d %s %llu + %u",
MAJOR(__entry->dev), MINOR(__entry->dev), __entry->rwbs,
(unsigned long long)__entry->sector, __entry->nr_sector)
);
/* super-io.c: */
TRACE_EVENT(write_super,
TP_PROTO(struct bch_fs *c, unsigned long ip),
TP_ARGS(c, ip),
TP_STRUCT__entry(
__field(dev_t, dev )
__field(unsigned long, ip )
),
TP_fast_assign(
__entry->dev = c->dev;
__entry->ip = ip;
),
TP_printk("%d,%d for %pS",
MAJOR(__entry->dev), MINOR(__entry->dev),
(void *) __entry->ip)
);
/* io.c: */
DEFINE_EVENT(bio, read_promote,
TP_PROTO(struct bio *bio),
TP_ARGS(bio)
);
DEFINE_EVENT(bio, read_bounce,
TP_PROTO(struct bio *bio),
TP_ARGS(bio)
);
DEFINE_EVENT(bio, read_split,
TP_PROTO(struct bio *bio),
TP_ARGS(bio)
);
DEFINE_EVENT(bio, read_retry,
TP_PROTO(struct bio *bio),
TP_ARGS(bio)
);
DEFINE_EVENT(bio, read_reuse_race,
TP_PROTO(struct bio *bio),
TP_ARGS(bio)
);
/* Journal */
DEFINE_EVENT(bch_fs, journal_full,
TP_PROTO(struct bch_fs *c),
TP_ARGS(c)
);
DEFINE_EVENT(bch_fs, journal_entry_full,
TP_PROTO(struct bch_fs *c),
TP_ARGS(c)
);
DEFINE_EVENT(bio, journal_write,
TP_PROTO(struct bio *bio),
TP_ARGS(bio)
);
TRACE_EVENT(journal_reclaim_start,
TP_PROTO(struct bch_fs *c, bool direct, bool kicked,
u64 min_nr, u64 min_key_cache,
u64 prereserved, u64 prereserved_total,
u64 btree_cache_dirty, u64 btree_cache_total,
u64 btree_key_cache_dirty, u64 btree_key_cache_total),
TP_ARGS(c, direct, kicked, min_nr, min_key_cache, prereserved, prereserved_total,
btree_cache_dirty, btree_cache_total,
btree_key_cache_dirty, btree_key_cache_total),
TP_STRUCT__entry(
__field(dev_t, dev )
__field(bool, direct )
__field(bool, kicked )
__field(u64, min_nr )
__field(u64, min_key_cache )
__field(u64, prereserved )
__field(u64, prereserved_total )
__field(u64, btree_cache_dirty )
__field(u64, btree_cache_total )
__field(u64, btree_key_cache_dirty )
__field(u64, btree_key_cache_total )
),
TP_fast_assign(
__entry->dev = c->dev;
__entry->direct = direct;
__entry->kicked = kicked;
__entry->min_nr = min_nr;
__entry->min_key_cache = min_key_cache;
__entry->prereserved = prereserved;
__entry->prereserved_total = prereserved_total;
__entry->btree_cache_dirty = btree_cache_dirty;
__entry->btree_cache_total = btree_cache_total;
__entry->btree_key_cache_dirty = btree_key_cache_dirty;
__entry->btree_key_cache_total = btree_key_cache_total;
),
TP_printk("%d,%d direct %u kicked %u min %llu key cache %llu prereserved %llu/%llu btree cache %llu/%llu key cache %llu/%llu",
MAJOR(__entry->dev), MINOR(__entry->dev),
__entry->direct,
__entry->kicked,
__entry->min_nr,
__entry->min_key_cache,
__entry->prereserved,
__entry->prereserved_total,
__entry->btree_cache_dirty,
__entry->btree_cache_total,
__entry->btree_key_cache_dirty,
__entry->btree_key_cache_total)
);
TRACE_EVENT(journal_reclaim_finish,
TP_PROTO(struct bch_fs *c, u64 nr_flushed),
TP_ARGS(c, nr_flushed),
TP_STRUCT__entry(
__field(dev_t, dev )
__field(u64, nr_flushed )
),
TP_fast_assign(
__entry->dev = c->dev;
__entry->nr_flushed = nr_flushed;
),
TP_printk("%d,%d flushed %llu",
MAJOR(__entry->dev), MINOR(__entry->dev),
__entry->nr_flushed)
);
/* bset.c: */
DEFINE_EVENT(bpos, bkey_pack_pos_fail,
TP_PROTO(const struct bpos *p),
TP_ARGS(p)
);
/* Btree cache: */
TRACE_EVENT(btree_cache_scan,
TP_PROTO(long nr_to_scan, long can_free, long ret),
TP_ARGS(nr_to_scan, can_free, ret),
TP_STRUCT__entry(
__field(long, nr_to_scan )
__field(long, can_free )
__field(long, ret )
),
TP_fast_assign(
__entry->nr_to_scan = nr_to_scan;
__entry->can_free = can_free;
__entry->ret = ret;
),
TP_printk("scanned for %li nodes, can free %li, ret %li",
__entry->nr_to_scan, __entry->can_free, __entry->ret)
);
DEFINE_EVENT(btree_node, btree_cache_reap,
TP_PROTO(struct bch_fs *c, struct btree *b),
TP_ARGS(c, b)
);
DEFINE_EVENT(bch_fs, btree_cache_cannibalize_lock_fail,
TP_PROTO(struct bch_fs *c),
TP_ARGS(c)
);
DEFINE_EVENT(bch_fs, btree_cache_cannibalize_lock,
TP_PROTO(struct bch_fs *c),
TP_ARGS(c)
);
DEFINE_EVENT(bch_fs, btree_cache_cannibalize,
TP_PROTO(struct bch_fs *c),
TP_ARGS(c)
);
DEFINE_EVENT(bch_fs, btree_cache_cannibalize_unlock,
TP_PROTO(struct bch_fs *c),
TP_ARGS(c)
);
/* Btree */
DEFINE_EVENT(btree_node, btree_node_read,
TP_PROTO(struct bch_fs *c, struct btree *b),
TP_ARGS(c, b)
);
TRACE_EVENT(btree_node_write,
TP_PROTO(struct btree *b, unsigned bytes, unsigned sectors),
TP_ARGS(b, bytes, sectors),
TP_STRUCT__entry(
__field(enum btree_node_type, type)
__field(unsigned, bytes )
__field(unsigned, sectors )
),
TP_fast_assign(
__entry->type = btree_node_type(b);
__entry->bytes = bytes;
__entry->sectors = sectors;
),
TP_printk("bkey type %u bytes %u sectors %u",
__entry->type , __entry->bytes, __entry->sectors)
);
DEFINE_EVENT(btree_node, btree_node_alloc,
TP_PROTO(struct bch_fs *c, struct btree *b),
TP_ARGS(c, b)
);
DEFINE_EVENT(btree_node, btree_node_free,
TP_PROTO(struct bch_fs *c, struct btree *b),
TP_ARGS(c, b)
);
TRACE_EVENT(btree_reserve_get_fail,
TP_PROTO(const char *trans_fn,
unsigned long caller_ip,
size_t required,
int ret),
TP_ARGS(trans_fn, caller_ip, required, ret),
TP_STRUCT__entry(
__array(char, trans_fn, 32 )
__field(unsigned long, caller_ip )
__field(size_t, required )
__array(char, ret, 32 )
),
TP_fast_assign(
strscpy(__entry->trans_fn, trans_fn, sizeof(__entry->trans_fn));
__entry->caller_ip = caller_ip;
__entry->required = required;
strscpy(__entry->ret, bch2_err_str(ret), sizeof(__entry->ret));
),
TP_printk("%s %pS required %zu ret %s",
__entry->trans_fn,
(void *) __entry->caller_ip,
__entry->required,
__entry->ret)
);
DEFINE_EVENT(btree_node, btree_node_compact,
TP_PROTO(struct bch_fs *c, struct btree *b),
TP_ARGS(c, b)
);
DEFINE_EVENT(btree_node, btree_node_merge,
TP_PROTO(struct bch_fs *c, struct btree *b),
TP_ARGS(c, b)
);
DEFINE_EVENT(btree_node, btree_node_split,
TP_PROTO(struct bch_fs *c, struct btree *b),
TP_ARGS(c, b)
);
DEFINE_EVENT(btree_node, btree_node_rewrite,
TP_PROTO(struct bch_fs *c, struct btree *b),
TP_ARGS(c, b)
);
DEFINE_EVENT(btree_node, btree_node_set_root,
TP_PROTO(struct bch_fs *c, struct btree *b),
TP_ARGS(c, b)
);
TRACE_EVENT(btree_path_relock_fail,
TP_PROTO(struct btree_trans *trans,
unsigned long caller_ip,
struct btree_path *path,
unsigned level),
TP_ARGS(trans, caller_ip, path, level),
TP_STRUCT__entry(
__array(char, trans_fn, 32 )
__field(unsigned long, caller_ip )
__field(u8, btree_id )
__field(u8, level )
TRACE_BPOS_entries(pos)
__array(char, node, 24 )
__field(u32, iter_lock_seq )
__field(u32, node_lock_seq )
),
TP_fast_assign(
struct btree *b = btree_path_node(path, level);
strscpy(__entry->trans_fn, trans->fn, sizeof(__entry->trans_fn));
__entry->caller_ip = caller_ip;
__entry->btree_id = path->btree_id;
__entry->level = path->level;
TRACE_BPOS_assign(pos, path->pos);
if (IS_ERR(b))
strscpy(__entry->node, bch2_err_str(PTR_ERR(b)), sizeof(__entry->node));
else
scnprintf(__entry->node, sizeof(__entry->node), "%px", b);
__entry->iter_lock_seq = path->l[level].lock_seq;
__entry->node_lock_seq = is_btree_node(path, level) ? path->l[level].b->c.lock.state.seq : 0;
),
TP_printk("%s %pS btree %s pos %llu:%llu:%u level %u node %s iter seq %u lock seq %u",
__entry->trans_fn,
(void *) __entry->caller_ip,
bch2_btree_ids[__entry->btree_id],
__entry->pos_inode,
__entry->pos_offset,
__entry->pos_snapshot,
__entry->level,
__entry->node,
__entry->iter_lock_seq,
__entry->node_lock_seq)
);
TRACE_EVENT(btree_path_upgrade_fail,
TP_PROTO(struct btree_trans *trans,
unsigned long caller_ip,
struct btree_path *path,
unsigned level),
TP_ARGS(trans, caller_ip, path, level),
TP_STRUCT__entry(
__array(char, trans_fn, 32 )
__field(unsigned long, caller_ip )
__field(u8, btree_id )
__field(u8, level )
TRACE_BPOS_entries(pos)
__field(u8, locked )
__field(u8, self_read_count )
__field(u8, self_intent_count)
__field(u8, read_count )
__field(u8, intent_count )
__field(u32, iter_lock_seq )
__field(u32, node_lock_seq )
),
TP_fast_assign(
struct six_lock_count c;
strscpy(__entry->trans_fn, trans->fn, sizeof(__entry->trans_fn));
__entry->caller_ip = caller_ip;
__entry->btree_id = path->btree_id;
__entry->level = level;
TRACE_BPOS_assign(pos, path->pos);
__entry->locked = btree_node_locked(path, level);
c = bch2_btree_node_lock_counts(trans, NULL, &path->l[level].b->c, level),
__entry->self_read_count = c.n[SIX_LOCK_read];
__entry->self_intent_count = c.n[SIX_LOCK_intent];
c = six_lock_counts(&path->l[level].b->c.lock);
__entry->read_count = c.n[SIX_LOCK_read];
__entry->intent_count = c.n[SIX_LOCK_read];
__entry->iter_lock_seq = path->l[level].lock_seq;
__entry->node_lock_seq = is_btree_node(path, level) ? path->l[level].b->c.lock.state.seq : 0;
),
TP_printk("%s %pS btree %s pos %llu:%llu:%u level %u locked %u held %u:%u lock count %u:%u iter seq %u lock seq %u",
__entry->trans_fn,
(void *) __entry->caller_ip,
bch2_btree_ids[__entry->btree_id],
__entry->pos_inode,
__entry->pos_offset,
__entry->pos_snapshot,
__entry->level,
__entry->locked,
__entry->self_read_count,
__entry->self_intent_count,
__entry->read_count,
__entry->intent_count,
__entry->iter_lock_seq,
__entry->node_lock_seq)
);
/* Garbage collection */
DEFINE_EVENT(bch_fs, gc_gens_start,
TP_PROTO(struct bch_fs *c),
TP_ARGS(c)
);
DEFINE_EVENT(bch_fs, gc_gens_end,
TP_PROTO(struct bch_fs *c),
TP_ARGS(c)
);
/* Allocator */
DECLARE_EVENT_CLASS(bucket_alloc,
TP_PROTO(struct bch_dev *ca, const char *alloc_reserve,
u64 bucket,
u64 free,
u64 avail,
u64 copygc_wait_amount,
s64 copygc_waiting_for,
struct bucket_alloc_state *s,
bool nonblocking,
const char *err),
TP_ARGS(ca, alloc_reserve, bucket, free, avail,
copygc_wait_amount, copygc_waiting_for,
s, nonblocking, err),
TP_STRUCT__entry(
__field(u8, dev )
__array(char, reserve, 16 )
__field(u64, bucket )
__field(u64, free )
__field(u64, avail )
__field(u64, copygc_wait_amount )
__field(s64, copygc_waiting_for )
__field(u64, seen )
__field(u64, open )
__field(u64, need_journal_commit )
__field(u64, nouse )
__field(bool, nonblocking )
bcachefs: Nocow support This adds support for nocow mode, where we do writes in-place when possible. Patch components: - New boolean filesystem and inode option, nocow: note that when nocow is enabled, data checksumming and compression are implicitly disabled - To prevent in-place writes from racing with data moves (data_update.c) or bucket reuse (i.e. a bucket being reused and re-allocated while a nocow write is in flight, we have a new locking mechanism. Buckets can be locked for either data update or data move, using a fixed size hash table of two_state_shared locks. We don't have any chaining, meaning updates and moves to different buckets that hash to the same lock will wait unnecessarily - we'll want to watch for this becoming an issue. - The allocator path also needs to check for in-place writes in flight to a given bucket before giving it out: thus we add another counter to bucket_alloc_state so we can track this. - Fsync now may need to issue cache flushes to block devices instead of flushing the journal. We add a device bitmask to bch_inode_info, ei_devs_need_flush, which tracks devices that need to have flushes issued - note that this will lead to unnecessary flushes when other codepaths have already issued flushes, we may want to replace this with a sequence number. - New nocow write path: look up extents, and if they're writable write to them - otherwise fall back to the normal COW write path. XXX: switch to sequence numbers instead of bitmask for devs needing journal flush XXX: ei_quota_lock being a mutex means bch2_nocow_write_done() needs to run in process context - see if we can improve this Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
2022-11-02 21:12:00 +00:00
__field(u64, nocow )
__array(char, err, 32 )
),
TP_fast_assign(
__entry->dev = ca->dev_idx;
strscpy(__entry->reserve, alloc_reserve, sizeof(__entry->reserve));
__entry->bucket = bucket;
__entry->free = free;
__entry->avail = avail;
__entry->copygc_wait_amount = copygc_wait_amount;
__entry->copygc_waiting_for = copygc_waiting_for;
__entry->seen = s->buckets_seen;
__entry->open = s->skipped_open;
__entry->need_journal_commit = s->skipped_need_journal_commit;
__entry->nouse = s->skipped_nouse;
__entry->nonblocking = nonblocking;
bcachefs: Nocow support This adds support for nocow mode, where we do writes in-place when possible. Patch components: - New boolean filesystem and inode option, nocow: note that when nocow is enabled, data checksumming and compression are implicitly disabled - To prevent in-place writes from racing with data moves (data_update.c) or bucket reuse (i.e. a bucket being reused and re-allocated while a nocow write is in flight, we have a new locking mechanism. Buckets can be locked for either data update or data move, using a fixed size hash table of two_state_shared locks. We don't have any chaining, meaning updates and moves to different buckets that hash to the same lock will wait unnecessarily - we'll want to watch for this becoming an issue. - The allocator path also needs to check for in-place writes in flight to a given bucket before giving it out: thus we add another counter to bucket_alloc_state so we can track this. - Fsync now may need to issue cache flushes to block devices instead of flushing the journal. We add a device bitmask to bch_inode_info, ei_devs_need_flush, which tracks devices that need to have flushes issued - note that this will lead to unnecessary flushes when other codepaths have already issued flushes, we may want to replace this with a sequence number. - New nocow write path: look up extents, and if they're writable write to them - otherwise fall back to the normal COW write path. XXX: switch to sequence numbers instead of bitmask for devs needing journal flush XXX: ei_quota_lock being a mutex means bch2_nocow_write_done() needs to run in process context - see if we can improve this Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
2022-11-02 21:12:00 +00:00
__entry->nocow = s->skipped_nocow;
strscpy(__entry->err, err, sizeof(__entry->err));
),
TP_printk("reserve %s bucket %u:%llu free %llu avail %llu copygc_wait %llu/%lli seen %llu open %llu need_journal_commit %llu nouse %llu nocow %llu nonblocking %u err %s",
__entry->reserve,
__entry->dev,
__entry->bucket,
__entry->free,
__entry->avail,
__entry->copygc_wait_amount,
__entry->copygc_waiting_for,
__entry->seen,
__entry->open,
__entry->need_journal_commit,
__entry->nouse,
bcachefs: Nocow support This adds support for nocow mode, where we do writes in-place when possible. Patch components: - New boolean filesystem and inode option, nocow: note that when nocow is enabled, data checksumming and compression are implicitly disabled - To prevent in-place writes from racing with data moves (data_update.c) or bucket reuse (i.e. a bucket being reused and re-allocated while a nocow write is in flight, we have a new locking mechanism. Buckets can be locked for either data update or data move, using a fixed size hash table of two_state_shared locks. We don't have any chaining, meaning updates and moves to different buckets that hash to the same lock will wait unnecessarily - we'll want to watch for this becoming an issue. - The allocator path also needs to check for in-place writes in flight to a given bucket before giving it out: thus we add another counter to bucket_alloc_state so we can track this. - Fsync now may need to issue cache flushes to block devices instead of flushing the journal. We add a device bitmask to bch_inode_info, ei_devs_need_flush, which tracks devices that need to have flushes issued - note that this will lead to unnecessary flushes when other codepaths have already issued flushes, we may want to replace this with a sequence number. - New nocow write path: look up extents, and if they're writable write to them - otherwise fall back to the normal COW write path. XXX: switch to sequence numbers instead of bitmask for devs needing journal flush XXX: ei_quota_lock being a mutex means bch2_nocow_write_done() needs to run in process context - see if we can improve this Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
2022-11-02 21:12:00 +00:00
__entry->nocow,
__entry->nonblocking,
__entry->err)
);
DEFINE_EVENT(bucket_alloc, bucket_alloc,
TP_PROTO(struct bch_dev *ca, const char *alloc_reserve,
u64 bucket,
u64 free,
u64 avail,
u64 copygc_wait_amount,
s64 copygc_waiting_for,
struct bucket_alloc_state *s,
bool nonblocking,
const char *err),
TP_ARGS(ca, alloc_reserve, bucket, free, avail,
copygc_wait_amount, copygc_waiting_for,
s, nonblocking, err)
);
DEFINE_EVENT(bucket_alloc, bucket_alloc_fail,
TP_PROTO(struct bch_dev *ca, const char *alloc_reserve,
u64 bucket,
u64 free,
u64 avail,
u64 copygc_wait_amount,
s64 copygc_waiting_for,
struct bucket_alloc_state *s,
bool nonblocking,
const char *err),
TP_ARGS(ca, alloc_reserve, bucket, free, avail,
copygc_wait_amount, copygc_waiting_for,
s, nonblocking, err)
);
TRACE_EVENT(discard_buckets,
TP_PROTO(struct bch_fs *c, u64 seen, u64 open,
u64 need_journal_commit, u64 discarded, const char *err),
TP_ARGS(c, seen, open, need_journal_commit, discarded, err),
TP_STRUCT__entry(
__field(dev_t, dev )
__field(u64, seen )
__field(u64, open )
__field(u64, need_journal_commit )
__field(u64, discarded )
__array(char, err, 16 )
),
TP_fast_assign(
__entry->dev = c->dev;
__entry->seen = seen;
__entry->open = open;
__entry->need_journal_commit = need_journal_commit;
__entry->discarded = discarded;
strscpy(__entry->err, err, sizeof(__entry->err));
),
TP_printk("%d%d seen %llu open %llu need_journal_commit %llu discarded %llu err %s",
MAJOR(__entry->dev), MINOR(__entry->dev),
__entry->seen,
__entry->open,
__entry->need_journal_commit,
__entry->discarded,
__entry->err)
);
TRACE_EVENT(bucket_invalidate,
TP_PROTO(struct bch_fs *c, unsigned dev, u64 bucket, u32 sectors),
TP_ARGS(c, dev, bucket, sectors),
TP_STRUCT__entry(
__field(dev_t, dev )
__field(u32, dev_idx )
__field(u32, sectors )
__field(u64, bucket )
),
TP_fast_assign(
__entry->dev = c->dev;
__entry->dev_idx = dev;
__entry->sectors = sectors;
__entry->bucket = bucket;
),
TP_printk("%d:%d invalidated %u:%llu cached sectors %u",
MAJOR(__entry->dev), MINOR(__entry->dev),
__entry->dev_idx, __entry->bucket,
__entry->sectors)
);
/* Moving IO */
DEFINE_EVENT(bkey, move_extent_read,
TP_PROTO(const struct bkey *k),
TP_ARGS(k)
);
DEFINE_EVENT(bkey, move_extent_write,
TP_PROTO(const struct bkey *k),
TP_ARGS(k)
);
DEFINE_EVENT(bkey, move_extent_finish,
TP_PROTO(const struct bkey *k),
TP_ARGS(k)
);
DEFINE_EVENT(bkey, move_extent_fail,
TP_PROTO(const struct bkey *k),
TP_ARGS(k)
);
DEFINE_EVENT(bkey, move_extent_alloc_mem_fail,
TP_PROTO(const struct bkey *k),
TP_ARGS(k)
);
TRACE_EVENT(move_data,
TP_PROTO(struct bch_fs *c, u64 sectors_moved,
u64 keys_moved),
TP_ARGS(c, sectors_moved, keys_moved),
TP_STRUCT__entry(
__field(dev_t, dev )
__field(u64, sectors_moved )
__field(u64, keys_moved )
),
TP_fast_assign(
__entry->dev = c->dev;
__entry->sectors_moved = sectors_moved;
__entry->keys_moved = keys_moved;
),
TP_printk("%d,%d sectors_moved %llu keys_moved %llu",
MAJOR(__entry->dev), MINOR(__entry->dev),
__entry->sectors_moved, __entry->keys_moved)
);
TRACE_EVENT(evacuate_bucket,
TP_PROTO(struct bch_fs *c, struct bpos *bucket,
unsigned sectors, unsigned bucket_size,
u64 fragmentation, int ret),
TP_ARGS(c, bucket, sectors, bucket_size, fragmentation, ret),
TP_STRUCT__entry(
__field(dev_t, dev )
__field(u64, member )
__field(u64, bucket )
__field(u32, sectors )
__field(u32, bucket_size )
__field(u64, fragmentation )
__field(int, ret )
),
TP_fast_assign(
__entry->dev = c->dev;
__entry->member = bucket->inode;
__entry->bucket = bucket->offset;
__entry->sectors = sectors;
__entry->bucket_size = bucket_size;
__entry->fragmentation = fragmentation;
__entry->ret = ret;
),
TP_printk("%d,%d %llu:%llu sectors %u/%u fragmentation %llu ret %i",
MAJOR(__entry->dev), MINOR(__entry->dev),
__entry->member, __entry->bucket,
__entry->sectors, __entry->bucket_size,
__entry->fragmentation, __entry->ret)
);
TRACE_EVENT(copygc,
TP_PROTO(struct bch_fs *c,
u64 sectors_moved, u64 sectors_not_moved,
u64 buckets_moved, u64 buckets_not_moved),
TP_ARGS(c,
sectors_moved, sectors_not_moved,
buckets_moved, buckets_not_moved),
TP_STRUCT__entry(
__field(dev_t, dev )
__field(u64, sectors_moved )
__field(u64, sectors_not_moved )
__field(u64, buckets_moved )
__field(u64, buckets_not_moved )
),
TP_fast_assign(
__entry->dev = c->dev;
__entry->sectors_moved = sectors_moved;
__entry->sectors_not_moved = sectors_not_moved;
__entry->buckets_moved = buckets_moved;
__entry->buckets_not_moved = buckets_moved;
),
TP_printk("%d,%d sectors moved %llu remain %llu buckets moved %llu remain %llu",
MAJOR(__entry->dev), MINOR(__entry->dev),
__entry->sectors_moved, __entry->sectors_not_moved,
__entry->buckets_moved, __entry->buckets_not_moved)
);
TRACE_EVENT(copygc_wait,
TP_PROTO(struct bch_fs *c,
u64 wait_amount, u64 until),
TP_ARGS(c, wait_amount, until),
TP_STRUCT__entry(
__field(dev_t, dev )
__field(u64, wait_amount )
__field(u64, until )
),
TP_fast_assign(
__entry->dev = c->dev;
__entry->wait_amount = wait_amount;
__entry->until = until;
),
TP_printk("%d,%u waiting for %llu sectors until %llu",
MAJOR(__entry->dev), MINOR(__entry->dev),
__entry->wait_amount, __entry->until)
);
/* btree transactions: */
DECLARE_EVENT_CLASS(transaction_event,
TP_PROTO(struct btree_trans *trans,
unsigned long caller_ip),
TP_ARGS(trans, caller_ip),
TP_STRUCT__entry(
__array(char, trans_fn, 32 )
__field(unsigned long, caller_ip )
),
TP_fast_assign(
strscpy(__entry->trans_fn, trans->fn, sizeof(__entry->trans_fn));
__entry->caller_ip = caller_ip;
),
TP_printk("%s %pS", __entry->trans_fn, (void *) __entry->caller_ip)
);
DEFINE_EVENT(transaction_event, transaction_commit,
TP_PROTO(struct btree_trans *trans,
unsigned long caller_ip),
TP_ARGS(trans, caller_ip)
);
DEFINE_EVENT(transaction_event, trans_restart_injected,
TP_PROTO(struct btree_trans *trans,
unsigned long caller_ip),
TP_ARGS(trans, caller_ip)
);
DEFINE_EVENT(transaction_event, trans_restart_split_race,
TP_PROTO(struct btree_trans *trans,
unsigned long caller_ip),
TP_ARGS(trans, caller_ip)
);
DEFINE_EVENT(transaction_event, trans_blocked_journal_reclaim,
TP_PROTO(struct btree_trans *trans,
unsigned long caller_ip),
TP_ARGS(trans, caller_ip)
);
DEFINE_EVENT(transaction_event, trans_restart_journal_res_get,
TP_PROTO(struct btree_trans *trans,
unsigned long caller_ip),
TP_ARGS(trans, caller_ip)
);
TRACE_EVENT(trans_restart_journal_preres_get,
TP_PROTO(struct btree_trans *trans,
unsigned long caller_ip,
unsigned flags),
TP_ARGS(trans, caller_ip, flags),
TP_STRUCT__entry(
__array(char, trans_fn, 32 )
__field(unsigned long, caller_ip )
__field(unsigned, flags )
),
TP_fast_assign(
strscpy(__entry->trans_fn, trans->fn, sizeof(__entry->trans_fn));
__entry->caller_ip = caller_ip;
__entry->flags = flags;
),
TP_printk("%s %pS %x", __entry->trans_fn,
(void *) __entry->caller_ip,
__entry->flags)
);
DEFINE_EVENT(transaction_event, trans_restart_journal_reclaim,
TP_PROTO(struct btree_trans *trans,
unsigned long caller_ip),
TP_ARGS(trans, caller_ip)
);
DEFINE_EVENT(transaction_event, trans_restart_fault_inject,
TP_PROTO(struct btree_trans *trans,
unsigned long caller_ip),
TP_ARGS(trans, caller_ip)
);
DEFINE_EVENT(transaction_event, trans_traverse_all,
TP_PROTO(struct btree_trans *trans,
unsigned long caller_ip),
TP_ARGS(trans, caller_ip)
);
DEFINE_EVENT(transaction_event, trans_restart_mark_replicas,
TP_PROTO(struct btree_trans *trans,
unsigned long caller_ip),
TP_ARGS(trans, caller_ip)
);
DEFINE_EVENT(transaction_event, trans_restart_key_cache_raced,
TP_PROTO(struct btree_trans *trans,
unsigned long caller_ip),
TP_ARGS(trans, caller_ip)
);
DEFINE_EVENT(transaction_event, trans_restart_too_many_iters,
TP_PROTO(struct btree_trans *trans,
unsigned long caller_ip),
TP_ARGS(trans, caller_ip)
);
DECLARE_EVENT_CLASS(transaction_restart_iter,
TP_PROTO(struct btree_trans *trans,
unsigned long caller_ip,
struct btree_path *path),
TP_ARGS(trans, caller_ip, path),
TP_STRUCT__entry(
__array(char, trans_fn, 32 )
__field(unsigned long, caller_ip )
__field(u8, btree_id )
TRACE_BPOS_entries(pos)
),
TP_fast_assign(
strscpy(__entry->trans_fn, trans->fn, sizeof(__entry->trans_fn));
__entry->caller_ip = caller_ip;
__entry->btree_id = path->btree_id;
TRACE_BPOS_assign(pos, path->pos)
),
TP_printk("%s %pS btree %s pos %llu:%llu:%u",
__entry->trans_fn,
(void *) __entry->caller_ip,
bch2_btree_ids[__entry->btree_id],
__entry->pos_inode,
__entry->pos_offset,
__entry->pos_snapshot)
);
DEFINE_EVENT(transaction_restart_iter, trans_restart_btree_node_reused,
TP_PROTO(struct btree_trans *trans,
unsigned long caller_ip,
struct btree_path *path),
TP_ARGS(trans, caller_ip, path)
);
DEFINE_EVENT(transaction_restart_iter, trans_restart_btree_node_split,
TP_PROTO(struct btree_trans *trans,
unsigned long caller_ip,
struct btree_path *path),
TP_ARGS(trans, caller_ip, path)
);
TRACE_EVENT(trans_restart_upgrade,
TP_PROTO(struct btree_trans *trans,
unsigned long caller_ip,
struct btree_path *path,
unsigned old_locks_want,
unsigned new_locks_want),
TP_ARGS(trans, caller_ip, path, old_locks_want, new_locks_want),
TP_STRUCT__entry(
__array(char, trans_fn, 32 )
__field(unsigned long, caller_ip )
__field(u8, btree_id )
__field(u8, old_locks_want )
__field(u8, new_locks_want )
TRACE_BPOS_entries(pos)
),
TP_fast_assign(
strscpy(__entry->trans_fn, trans->fn, sizeof(__entry->trans_fn));
__entry->caller_ip = caller_ip;
__entry->btree_id = path->btree_id;
__entry->old_locks_want = old_locks_want;
__entry->new_locks_want = new_locks_want;
TRACE_BPOS_assign(pos, path->pos)
),
TP_printk("%s %pS btree %s pos %llu:%llu:%u locks_want %u -> %u",
__entry->trans_fn,
(void *) __entry->caller_ip,
bch2_btree_ids[__entry->btree_id],
__entry->pos_inode,
__entry->pos_offset,
__entry->pos_snapshot,
__entry->old_locks_want,
__entry->new_locks_want)
);
DEFINE_EVENT(transaction_restart_iter, trans_restart_relock,
TP_PROTO(struct btree_trans *trans,
unsigned long caller_ip,
struct btree_path *path),
TP_ARGS(trans, caller_ip, path)
);
DEFINE_EVENT(transaction_restart_iter, trans_restart_relock_next_node,
TP_PROTO(struct btree_trans *trans,
unsigned long caller_ip,
struct btree_path *path),
TP_ARGS(trans, caller_ip, path)
);
DEFINE_EVENT(transaction_restart_iter, trans_restart_relock_parent_for_fill,
TP_PROTO(struct btree_trans *trans,
unsigned long caller_ip,
struct btree_path *path),
TP_ARGS(trans, caller_ip, path)
);
DEFINE_EVENT(transaction_restart_iter, trans_restart_relock_after_fill,
TP_PROTO(struct btree_trans *trans,
unsigned long caller_ip,
struct btree_path *path),
TP_ARGS(trans, caller_ip, path)
);
DEFINE_EVENT(transaction_event, trans_restart_key_cache_upgrade,
TP_PROTO(struct btree_trans *trans,
unsigned long caller_ip),
TP_ARGS(trans, caller_ip)
);
DEFINE_EVENT(transaction_restart_iter, trans_restart_relock_key_cache_fill,
TP_PROTO(struct btree_trans *trans,
unsigned long caller_ip,
struct btree_path *path),
TP_ARGS(trans, caller_ip, path)
);
DEFINE_EVENT(transaction_restart_iter, trans_restart_relock_path,
TP_PROTO(struct btree_trans *trans,
unsigned long caller_ip,
struct btree_path *path),
TP_ARGS(trans, caller_ip, path)
);
DEFINE_EVENT(transaction_restart_iter, trans_restart_relock_path_intent,
TP_PROTO(struct btree_trans *trans,
unsigned long caller_ip,
struct btree_path *path),
TP_ARGS(trans, caller_ip, path)
);
DEFINE_EVENT(transaction_restart_iter, trans_restart_traverse,
TP_PROTO(struct btree_trans *trans,
unsigned long caller_ip,
struct btree_path *path),
TP_ARGS(trans, caller_ip, path)
);
DEFINE_EVENT(transaction_restart_iter, trans_restart_memory_allocation_failure,
TP_PROTO(struct btree_trans *trans,
unsigned long caller_ip,
struct btree_path *path),
TP_ARGS(trans, caller_ip, path)
);
DEFINE_EVENT(transaction_event, trans_restart_would_deadlock,
TP_PROTO(struct btree_trans *trans,
unsigned long caller_ip),
TP_ARGS(trans, caller_ip)
);
bcachefs: Deadlock cycle detector We've outgrown our own deadlock avoidance strategy. The btree iterator API provides an interface where the user doesn't need to concern themselves with lock ordering - different btree iterators can be traversed in any order. Without special care, this will lead to deadlocks. Our previous strategy was to define a lock ordering internally, and whenever we attempt to take a lock and trylock() fails, we'd check if the current btree transaction is holding any locks that cause a lock ordering violation. If so, we'd issue a transaction restart, and then bch2_trans_begin() would re-traverse all previously used iterators, but in the correct order. That approach had some issues, though. - Sometimes we'd issue transaction restarts unnecessarily, when no deadlock would have actually occured. Lock ordering restarts have become our primary cause of transaction restarts, on some workloads totally 20% of actual transaction commits. - To avoid deadlock or livelock, we'd often have to take intent locks when we only wanted a read lock: with the lock ordering approach, it is actually illegal to hold _any_ read lock while blocking on an intent lock, and this has been causing us unnecessary lock contention. - It was getting fragile - the various lock ordering rules are not trivial, and we'd been seeing occasional livelock issues related to this machinery. So, since bcachefs is already a relational database masquerading as a filesystem, we're stealing the next traditional database technique and switching to a cycle detector for avoiding deadlocks. When we block taking a btree lock, after adding ourself to the waitlist but before sleeping, we do a DFS of btree transactions waiting on other btree transactions, starting with the current transaction and walking our held locks, and transactions blocking on our held locks. If we find a cycle, we emit a transaction restart. Occasionally (e.g. the btree split path) we can not allow the lock() operation to fail, so if necessary we'll tell another transaction that it has to fail. Result: trans_restart_would_deadlock events are reduced by a factor of 10 to 100, and we'll be able to delete a whole bunch of grotty, fragile code. Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com>
2022-08-22 17:23:47 +00:00
DEFINE_EVENT(transaction_event, trans_restart_would_deadlock_recursion_limit,
TP_PROTO(struct btree_trans *trans,
unsigned long caller_ip),
TP_ARGS(trans, caller_ip)
);
TRACE_EVENT(trans_restart_would_deadlock_write,
TP_PROTO(struct btree_trans *trans),
TP_ARGS(trans),
TP_STRUCT__entry(
__array(char, trans_fn, 32 )
),
TP_fast_assign(
strscpy(__entry->trans_fn, trans->fn, sizeof(__entry->trans_fn));
),
TP_printk("%s", __entry->trans_fn)
);
TRACE_EVENT(trans_restart_mem_realloced,
TP_PROTO(struct btree_trans *trans,
unsigned long caller_ip,
unsigned long bytes),
TP_ARGS(trans, caller_ip, bytes),
TP_STRUCT__entry(
__array(char, trans_fn, 32 )
__field(unsigned long, caller_ip )
__field(unsigned long, bytes )
),
TP_fast_assign(
strscpy(__entry->trans_fn, trans->fn, sizeof(__entry->trans_fn));
__entry->caller_ip = caller_ip;
__entry->bytes = bytes;
),
TP_printk("%s %pS bytes %lu",
__entry->trans_fn,
(void *) __entry->caller_ip,
__entry->bytes)
);
TRACE_EVENT(trans_restart_key_cache_key_realloced,
TP_PROTO(struct btree_trans *trans,
unsigned long caller_ip,
struct btree_path *path,
unsigned old_u64s,
unsigned new_u64s),
TP_ARGS(trans, caller_ip, path, old_u64s, new_u64s),
TP_STRUCT__entry(
__array(char, trans_fn, 32 )
__field(unsigned long, caller_ip )
__field(enum btree_id, btree_id )
TRACE_BPOS_entries(pos)
__field(u32, old_u64s )
__field(u32, new_u64s )
),
TP_fast_assign(
strscpy(__entry->trans_fn, trans->fn, sizeof(__entry->trans_fn));
__entry->caller_ip = caller_ip;
__entry->btree_id = path->btree_id;
TRACE_BPOS_assign(pos, path->pos);
__entry->old_u64s = old_u64s;
__entry->new_u64s = new_u64s;
),
TP_printk("%s %pS btree %s pos %llu:%llu:%u old_u64s %u new_u64s %u",
__entry->trans_fn,
(void *) __entry->caller_ip,
bch2_btree_ids[__entry->btree_id],
__entry->pos_inode,
__entry->pos_offset,
__entry->pos_snapshot,
__entry->old_u64s,
__entry->new_u64s)
);
DEFINE_EVENT(transaction_event, trans_restart_write_buffer_flush,
TP_PROTO(struct btree_trans *trans,
unsigned long caller_ip),
TP_ARGS(trans, caller_ip)
);
TRACE_EVENT(write_buffer_flush,
TP_PROTO(struct btree_trans *trans, size_t nr, size_t skipped, size_t fast, size_t size),
TP_ARGS(trans, nr, skipped, fast, size),
TP_STRUCT__entry(
__field(size_t, nr )
__field(size_t, skipped )
__field(size_t, fast )
__field(size_t, size )
),
TP_fast_assign(
__entry->nr = nr;
__entry->skipped = skipped;
__entry->fast = fast;
__entry->size = size;
),
TP_printk("%zu/%zu skipped %zu fast %zu",
__entry->nr, __entry->size, __entry->skipped, __entry->fast)
);
TRACE_EVENT(write_buffer_flush_slowpath,
TP_PROTO(struct btree_trans *trans, size_t nr, size_t size),
TP_ARGS(trans, nr, size),
TP_STRUCT__entry(
__field(size_t, nr )
__field(size_t, size )
),
TP_fast_assign(
__entry->nr = nr;
__entry->size = size;
),
TP_printk("%zu/%zu", __entry->nr, __entry->size)
);
#endif /* _TRACE_BCACHEFS_H */
/* This part must be outside protection */
#undef TRACE_INCLUDE_PATH
#define TRACE_INCLUDE_PATH ../../fs/bcachefs
#undef TRACE_INCLUDE_FILE
#define TRACE_INCLUDE_FILE trace
#include <trace/define_trace.h>