From 5f43b0134e40bd798ba0999a11e90f24a0c65a51 Mon Sep 17 00:00:00 2001 From: Kent Overstreet Date: Mon, 22 Jan 2024 14:25:00 -0500 Subject: bcachefs: btree node prefetching in check_topology btree_and_journal_iter is old code that we want to get rid of, but we're not ready to yet. lack of btree node prefetching is, it turns out, a real performance issue for fsck on spinning rust, so - add it. Signed-off-by: Kent Overstreet --- fs/bcachefs/btree_cache.c | 10 +++++++--- 1 file changed, 7 insertions(+), 3 deletions(-) (limited to 'fs/bcachefs/btree_cache.c') diff --git a/fs/bcachefs/btree_cache.c b/fs/bcachefs/btree_cache.c index d7c81beac14a..a8b393bc7567 100644 --- a/fs/bcachefs/btree_cache.c +++ b/fs/bcachefs/btree_cache.c @@ -711,6 +711,9 @@ static noinline struct btree *bch2_btree_node_fill(struct btree_trans *trans, b = bch2_btree_node_mem_alloc(trans, level != 0); if (bch2_err_matches(PTR_ERR_OR_ZERO(b), ENOMEM)) { + if (!path) + return b; + trans->memory_allocation_failure = true; trace_and_count(c, trans_restart_memory_allocation_failure, trans, _THIS_IP_, path); return ERR_PTR(btree_trans_restart(trans, BCH_ERR_transaction_restart_fill_mem_alloc_fail)); @@ -760,8 +763,9 @@ static noinline struct btree *bch2_btree_node_fill(struct btree_trans *trans, } if (!six_relock_type(&b->c.lock, lock_type, seq)) { - if (path) - trace_and_count(c, trans_restart_relock_after_fill, trans, _THIS_IP_, path); + BUG_ON(!path); + + trace_and_count(c, trans_restart_relock_after_fill, trans, _THIS_IP_, path); return ERR_PTR(btree_trans_restart(trans, BCH_ERR_transaction_restart_relock_after_fill)); } @@ -1096,7 +1100,7 @@ int bch2_btree_node_prefetch(struct btree_trans *trans, struct btree_cache *bc = &c->btree_cache; struct btree *b; - BUG_ON(trans && !btree_node_locked(path, level + 1)); + BUG_ON(path && !btree_node_locked(path, level + 1)); BUG_ON(level >= BTREE_MAX_DEPTH); b = btree_cache_find(bc, k); -- cgit v1.2.3 From cb6fc943b650c4f0ca2ba753531c0803c8afbb5c Mon Sep 17 00:00:00 2001 From: Kent Overstreet Date: Thu, 1 Feb 2024 06:35:46 -0500 Subject: bcachefs: kill kvpmalloc() Signed-off-by: Kent Overstreet --- fs/bcachefs/btree_cache.c | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) (limited to 'fs/bcachefs/btree_cache.c') diff --git a/fs/bcachefs/btree_cache.c b/fs/bcachefs/btree_cache.c index a8b393bc7567..72d24933dc19 100644 --- a/fs/bcachefs/btree_cache.c +++ b/fs/bcachefs/btree_cache.c @@ -60,7 +60,7 @@ static void btree_node_data_free(struct bch_fs *c, struct btree *b) clear_btree_node_just_written(b); - kvpfree(b->data, btree_buf_bytes(b)); + kvfree(b->data); b->data = NULL; #ifdef __KERNEL__ kvfree(b->aux_data); @@ -94,7 +94,7 @@ static int btree_node_data_alloc(struct bch_fs *c, struct btree *b, gfp_t gfp) { BUG_ON(b->data || b->aux_data); - b->data = kvpmalloc(btree_buf_bytes(b), gfp); + b->data = kvmalloc(btree_buf_bytes(b), gfp); if (!b->data) return -BCH_ERR_ENOMEM_btree_node_mem_alloc; #ifdef __KERNEL__ @@ -107,7 +107,7 @@ static int btree_node_data_alloc(struct bch_fs *c, struct btree *b, gfp_t gfp) b->aux_data = NULL; #endif if (!b->aux_data) { - kvpfree(b->data, btree_buf_bytes(b)); + kvfree(b->data); b->data = NULL; return -BCH_ERR_ENOMEM_btree_node_mem_alloc; } @@ -408,7 +408,7 @@ void bch2_fs_btree_cache_exit(struct bch_fs *c) if (c->verify_data) list_move(&c->verify_data->list, &bc->live); - kvpfree(c->verify_ondisk, c->opts.btree_node_size); + kvfree(c->verify_ondisk); for (i = 0; i < btree_id_nr_alive(c); i++) { struct btree_root *r = bch2_btree_id_root(c, i); -- cgit v1.2.3 From 52946d828aac5bc8e20665a8434f87740fad03e3 Mon Sep 17 00:00:00 2001 From: Kent Overstreet Date: Tue, 6 Feb 2024 17:24:18 -0500 Subject: bcachefs: Kill more -EIO error codes This converts -EIOs related to btree node errors to private error codes, which will help with some ongoing debugging by giving us better error messages. Signed-off-by: Kent Overstreet --- fs/bcachefs/btree_cache.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) (limited to 'fs/bcachefs/btree_cache.c') diff --git a/fs/bcachefs/btree_cache.c b/fs/bcachefs/btree_cache.c index 72d24933dc19..e8665258360e 100644 --- a/fs/bcachefs/btree_cache.c +++ b/fs/bcachefs/btree_cache.c @@ -905,7 +905,7 @@ retry: if (unlikely(btree_node_read_error(b))) { six_unlock_type(&b->c.lock, lock_type); - return ERR_PTR(-EIO); + return ERR_PTR(-BCH_ERR_btree_node_read_error); } EBUG_ON(b->c.btree_id != path->btree_id); @@ -996,7 +996,7 @@ struct btree *bch2_btree_node_get(struct btree_trans *trans, struct btree_path * if (unlikely(btree_node_read_error(b))) { six_unlock_type(&b->c.lock, lock_type); - return ERR_PTR(-EIO); + return ERR_PTR(-BCH_ERR_btree_node_read_error); } EBUG_ON(b->c.btree_id != path->btree_id); @@ -1079,7 +1079,7 @@ lock_node: if (unlikely(btree_node_read_error(b))) { six_unlock_read(&b->c.lock); - b = ERR_PTR(-EIO); + b = ERR_PTR(-BCH_ERR_btree_node_read_error); goto out; } -- cgit v1.2.3 From 91dcad18d38849f1702e0d50f5598bb3614c8ff8 Mon Sep 17 00:00:00 2001 From: Kent Overstreet Date: Tue, 23 Jan 2024 00:01:07 -0500 Subject: bcachefs: Pin btree cache in ram for random access in fsck Various phases of fsck involve checking references from one btree to another: this means doing a sequential scan of one btree, and then mostly random access into the second. This is particularly painful for checking extents <-> backpointers; we can prefetch btree node access on the sequential scan, but not on the random access portion, and this is particularly painful on spinning rust, where we'd like to keep the pipeline fairly full of btree node reads so that the elevator can reduce seeking. This patch implements prefetching and pinning of the portion of the btree that we'll be doing random access to. We already calculate how much of the random access btree will fit in memory so it's a fairly straightforward change. This will put more pressure on system memory usage, so we introduce a new option, fsck_memory_usage_percent, which is the percentage of total system ram that fsck is allowed to pin. Signed-off-by: Kent Overstreet --- fs/bcachefs/btree_cache.c | 13 +++++++++++++ 1 file changed, 13 insertions(+) (limited to 'fs/bcachefs/btree_cache.c') diff --git a/fs/bcachefs/btree_cache.c b/fs/bcachefs/btree_cache.c index e8665258360e..562561a9a510 100644 --- a/fs/bcachefs/btree_cache.c +++ b/fs/bcachefs/btree_cache.c @@ -1,6 +1,7 @@ // SPDX-License-Identifier: GPL-2.0 #include "bcachefs.h" +#include "bbpos.h" #include "bkey_buf.h" #include "btree_cache.h" #include "btree_io.h" @@ -208,6 +209,18 @@ static int __btree_node_reclaim(struct bch_fs *c, struct btree *b, bool flush) int ret = 0; lockdep_assert_held(&bc->lock); + + struct bbpos pos = BBPOS(b->c.btree_id, b->key.k.p); + + u64 mask = b->c.level + ? bc->pinned_nodes_interior_mask + : bc->pinned_nodes_leaf_mask; + + if ((mask & BIT_ULL(b->c.btree_id)) && + bbpos_cmp(bc->pinned_nodes_start, pos) < 0 && + bbpos_cmp(bc->pinned_nodes_end, pos) >= 0) + return -BCH_ERR_ENOMEM_btree_node_reclaim; + wait_on_io: if (b->flags & ((1U << BTREE_NODE_dirty)| (1U << BTREE_NODE_read_in_flight)| -- cgit v1.2.3