bcachefs: Kill unnecessary mark_lock usage

We can't hold mark_lock while calling fsck_err() - that's a deadlock,
mark_lock is meant to be a leaf node lock.

It's also unnecessary for gc_bucket() and bucket_gen(); rcu suffices
since the bucket_gens array describes its size, and we can't race with
device removal or resize during gc/fsck since that takes state lock.

Reported-by: syzbot+38641fcbda1aaffefdd4@syzkaller.appspotmail.com
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

fs/bcachefs/alloc_foreground.c

@@ -107,14 +107,10 @@ void __bch2_open_bucket_put(struct bch_fs *c, struct open_bucket *ob)
 		return;
 	}
 
-	percpu_down_read(&c->mark_lock);
 	spin_lock(&ob->lock);
-
 	ob->valid = false;
 	ob->data_type = 0;
-
 	spin_unlock(&ob->lock);
-	percpu_up_read(&c->mark_lock);
 
 	spin_lock(&c->freelist_lock);
 	bch2_open_bucket_hash_remove(c, ob);

fs/bcachefs/bcachefs.h

@@ -547,15 +547,13 @@ struct bch_dev {
 
 	/*
 	 * Buckets:
-	 * Per-bucket arrays are protected by c->mark_lock, bucket_lock and
-	 * gc_gens_lock, for device resize - holding any is sufficient for
-	 * access: Or rcu_read_lock(), but only for dev_ptr_stale():
+	 * Per-bucket arrays are protected by either rcu_read_lock or
+	 * state_lock, for device resize.
 	 */
 	GENRADIX(struct bucket)	buckets_gc;
 	struct bucket_gens __rcu *bucket_gens;
 	u8			*oldest_gen;
 	unsigned long		*buckets_nouse;
-	struct rw_semaphore	bucket_lock;
 
 	struct bch_dev_usage __percpu	*usage;
 

fs/bcachefs/btree_gc.c

@@ -811,7 +811,6 @@ static int bch2_alloc_write_key(struct btree_trans *trans,
 	old = bch2_alloc_to_v4(k, &old_convert);
 	gc = new = *old;
 
-	percpu_down_read(&c->mark_lock);
 	__bucket_m_to_alloc(&gc, *gc_bucket(ca, iter->pos.offset));
 
 	old_gc = gc;
@@ -822,7 +821,6 @@ static int bch2_alloc_write_key(struct btree_trans *trans,
 		gc.data_type = old->data_type;
 		gc.dirty_sectors = old->dirty_sectors;
 	}
-	percpu_up_read(&c->mark_lock);
 
 	/*
 	 * gc.data_type doesn't yet include need_discard & need_gc_gen states -
@@ -840,11 +838,9 @@ static int bch2_alloc_write_key(struct btree_trans *trans,
 		 * safe w.r.t. transaction restarts, so fixup the gc_bucket so
 		 * we don't run it twice:
 		 */
-		percpu_down_read(&c->mark_lock);
 		struct bucket *gc_m = gc_bucket(ca, iter->pos.offset);
 		gc_m->data_type = gc.data_type;
 		gc_m->dirty_sectors = gc.dirty_sectors;
-		percpu_up_read(&c->mark_lock);
 	}
 
 	if (fsck_err_on(new.data_type != gc.data_type,
@@ -1088,7 +1084,6 @@ static int gc_btree_gens_key(struct btree_trans *trans,
 	if (unlikely(test_bit(BCH_FS_going_ro, &c->flags)))
 		return -EROFS;
 
-	percpu_down_read(&c->mark_lock);
 	rcu_read_lock();
 	bkey_for_each_ptr(ptrs, ptr) {
 		struct bch_dev *ca = bch2_dev_rcu(c, ptr->dev);
@@ -1097,7 +1092,6 @@ static int gc_btree_gens_key(struct btree_trans *trans,
 
 		if (dev_ptr_stale(ca, ptr) > 16) {
 			rcu_read_unlock();
-			percpu_up_read(&c->mark_lock);
 			goto update;
 		}
 	}
@@ -1112,7 +1106,6 @@ static int gc_btree_gens_key(struct btree_trans *trans,
 			*gen = ptr->gen;
 	}
 	rcu_read_unlock();
-	percpu_up_read(&c->mark_lock);
 	return 0;
 update:
 	u = bch2_bkey_make_mut(trans, iter, &k, 0);

fs/bcachefs/buckets.c

@@ -262,8 +262,6 @@ int bch2_check_fix_ptrs(struct btree_trans *trans,
 	struct printbuf buf = PRINTBUF;
 	int ret = 0;
 
-	percpu_down_read(&c->mark_lock);
-
 	bkey_for_each_ptr_decode(k.k, ptrs_c, p, entry_c) {
 		ret = bch2_check_fix_ptr(trans, k, p, entry_c, &do_update);
 		if (ret)
@@ -364,7 +362,6 @@ int bch2_check_fix_ptrs(struct btree_trans *trans,
 			bch_info(c, "new key %s", buf.buf);
 		}
 
-		percpu_up_read(&c->mark_lock);
 		struct btree_iter iter;
 		bch2_trans_node_iter_init(trans, &iter, btree, new->k.p, 0, level,
 					  BTREE_ITER_intent|BTREE_ITER_all_snapshots);
@@ -373,8 +370,6 @@ int bch2_check_fix_ptrs(struct btree_trans *trans,
 					  BTREE_UPDATE_internal_snapshot_node|
 					  BTREE_TRIGGER_norun);
 		bch2_trans_iter_exit(trans, &iter);
-		percpu_down_read(&c->mark_lock);
-
 		if (ret)
 			goto err;
 
@@ -382,7 +377,6 @@ int bch2_check_fix_ptrs(struct btree_trans *trans,
 			bch2_btree_node_update_key_early(trans, btree, level - 1, k, new);
 	}
 err:
-	percpu_up_read(&c->mark_lock);
 	printbuf_exit(&buf);
 	return ret;
 }
@@ -603,13 +597,12 @@ static int bch2_trigger_pointer(struct btree_trans *trans,
 	}
 
 	if (flags & BTREE_TRIGGER_gc) {
-		percpu_down_read(&c->mark_lock);
 		struct bucket *g = gc_bucket(ca, bucket.offset);
 		if (bch2_fs_inconsistent_on(!g, c, "reference to invalid bucket on device %u\n  %s",
 					    p.ptr.dev,
 					    (bch2_bkey_val_to_text(&buf, c, k), buf.buf))) {
 			ret = -BCH_ERR_trigger_pointer;
-			goto err_unlock;
+			goto err;
 		}
 
 		bucket_lock(g);
@@ -617,8 +610,6 @@ static int bch2_trigger_pointer(struct btree_trans *trans,
 		ret = __mark_pointer(trans, ca, k, &p, *sectors, bp.v.data_type, &new);
 		alloc_to_bucket(g, new);
 		bucket_unlock(g);
-err_unlock:
-		percpu_up_read(&c->mark_lock);
 
 		if (!ret)
 			ret = bch2_alloc_key_to_dev_counters(trans, ca, &old, &new, flags);
@@ -996,11 +987,10 @@ static int bch2_mark_metadata_bucket(struct btree_trans *trans, struct bch_dev *
 	struct bch_fs *c = trans->c;
 	int ret = 0;
 
-	percpu_down_read(&c->mark_lock);
 	struct bucket *g = gc_bucket(ca, b);
 	if (bch2_fs_inconsistent_on(!g, c, "reference to invalid bucket on device %u when marking metadata type %s",
 				    ca->dev_idx, bch2_data_type_str(data_type)))
-		goto err_unlock;
+		goto err;
 
 	bucket_lock(g);
 	struct bch_alloc_v4 old = bucket_m_to_alloc(*g);
@@ -1010,26 +1000,24 @@ static int bch2_mark_metadata_bucket(struct btree_trans *trans, struct bch_dev *
 			"different types of data in same bucket: %s, %s",
 			bch2_data_type_str(g->data_type),
 			bch2_data_type_str(data_type)))
-		goto err;
+		goto err_unlock;
 
 	if (bch2_fs_inconsistent_on((u64) g->dirty_sectors + sectors > ca->mi.bucket_size, c,
 			"bucket %u:%llu gen %u data type %s sector count overflow: %u + %u > bucket size",
 			ca->dev_idx, b, g->gen,
 			bch2_data_type_str(g->data_type ?: data_type),
 			g->dirty_sectors, sectors))
-		goto err;
+		goto err_unlock;
 
 	g->data_type = data_type;
 	g->dirty_sectors += sectors;
 	struct bch_alloc_v4 new = bucket_m_to_alloc(*g);
 	bucket_unlock(g);
-	percpu_up_read(&c->mark_lock);
 	ret = bch2_alloc_key_to_dev_counters(trans, ca, &old, &new, flags);
 	return ret;
-err:
-	bucket_unlock(g);
 err_unlock:
-	percpu_up_read(&c->mark_lock);
+	bucket_unlock(g);
+err:
 	return -BCH_ERR_metadata_bucket_inconsistency;
 }
 
@@ -1295,7 +1283,11 @@ int bch2_dev_buckets_resize(struct bch_fs *c, struct bch_dev *ca, u64 nbuckets)
 	bool resize = ca->bucket_gens != NULL;
 	int ret;
 
-	BUG_ON(resize && ca->buckets_nouse);
+	if (resize)
+		lockdep_assert_held(&c->state_lock);
+
+	if (resize && ca->buckets_nouse)
+		return -BCH_ERR_no_resize_with_buckets_nouse;
 
 	bucket_gens = kvmalloc(struct_size(bucket_gens, b, nbuckets),
 			       GFP_KERNEL|__GFP_ZERO);
@@ -1309,11 +1301,6 @@ int bch2_dev_buckets_resize(struct bch_fs *c, struct bch_dev *ca, u64 nbuckets)
 	bucket_gens->nbuckets_minus_first =
 		bucket_gens->nbuckets - bucket_gens->first_bucket;
 
-	if (resize) {
-		down_write(&ca->bucket_lock);
-		percpu_down_write(&c->mark_lock);
-	}
-
 	old_bucket_gens = rcu_dereference_protected(ca->bucket_gens, 1);
 
 	if (resize) {
@@ -1331,11 +1318,6 @@ int bch2_dev_buckets_resize(struct bch_fs *c, struct bch_dev *ca, u64 nbuckets)
 
 	nbuckets = ca->mi.nbuckets;
 
-	if (resize) {
-		percpu_up_write(&c->mark_lock);
-		up_write(&ca->bucket_lock);
-	}
-
 	ret = 0;
 err:
 	if (bucket_gens)

fs/bcachefs/buckets.h

@@ -82,16 +82,15 @@ static inline void bucket_lock(struct bucket *b)
 
 static inline struct bucket *gc_bucket(struct bch_dev *ca, size_t b)
 {
-	return genradix_ptr(&ca->buckets_gc, b);
+	return bucket_valid(ca, b)
+		? genradix_ptr(&ca->buckets_gc, b)
+		: NULL;
 }
 
 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->state_lock) ||
-				     lockdep_is_held(&ca->bucket_lock));
+				     lockdep_is_held(&ca->fs->state_lock));
 }
 
 static inline u8 *bucket_gen(struct bch_dev *ca, size_t b)

fs/bcachefs/ec.c

@@ -305,13 +305,12 @@ static int mark_stripe_bucket(struct btree_trans *trans,
 	}
 
 	if (flags & BTREE_TRIGGER_gc) {
-		percpu_down_read(&c->mark_lock);
 		struct bucket *g = gc_bucket(ca, bucket.offset);
 		if (bch2_fs_inconsistent_on(!g, c, "reference to invalid bucket on device %u\n  %s",
 					    ptr->dev,
 					    (bch2_bkey_val_to_text(&buf, c, s.s_c), buf.buf))) {
 			ret = -BCH_ERR_mark_stripe;
-			goto err_unlock;
+			goto err;
 		}
 
 		bucket_lock(g);
@@ -319,8 +318,7 @@ static int mark_stripe_bucket(struct btree_trans *trans,
 		ret = __mark_stripe_bucket(trans, ca, s, ptr_idx, deleting, bucket, &new, flags);
 		alloc_to_bucket(g, new);
 		bucket_unlock(g);
-err_unlock:
-		percpu_up_read(&c->mark_lock);
+
 		if (!ret)
 			ret = bch2_alloc_key_to_dev_counters(trans, ca, &old, &new, flags);
 	}

fs/bcachefs/errcode.h

@@ -195,6 +195,7 @@
 	x(EINVAL,			opt_parse_error)			\
 	x(EINVAL,			remove_with_metadata_missing_unimplemented)\
 	x(EINVAL,			remove_would_lose_data)			\
+	x(EINVAL,			no_resize_with_buckets_nouse)		\
 	x(EROFS,			erofs_trans_commit)			\
 	x(EROFS,			erofs_no_writes)			\
 	x(EROFS,			erofs_journal_err)			\

fs/bcachefs/super.c

@@ -1311,8 +1311,6 @@ static struct bch_dev *__bch2_dev_alloc(struct bch_fs *c,
 	init_completion(&ca->ref_completion);
 	init_completion(&ca->io_ref_completion);
 
-	init_rwsem(&ca->bucket_lock);
-
 	INIT_WORK(&ca->io_error_work, bch2_io_error_work);
 
 	bch2_time_stats_quantiles_init(&ca->io_latency[READ]);