1 files changed, 364 insertions, 307 deletions

diff --git a/libbcachefs/btree_update_leaf.c b/libbcachefs/btree_update_leaf.c
index 7e9909e2..9c8c5cac 100644
--- a/libbcachefs/btree_update_leaf.c
+++ b/libbcachefs/btree_update_leaf.c
@@ -29,40 +29,59 @@ static inline int btree_insert_entry_cmp(const struct btree_insert_entry *l,
 		 bpos_cmp(l->k->k.p,	r->k->k.p);
 }
 
+static inline struct btree_path_level *insert_l(struct btree_insert_entry *i)
+{
+	return i->path->l + i->level;
+}
+
 static inline bool same_leaf_as_prev(struct btree_trans *trans,
 				     struct btree_insert_entry *i)
 {
 	return i != trans->updates &&
-		iter_l(i[0].iter)->b == iter_l(i[-1].iter)->b;
+		insert_l(&i[0])->b == insert_l(&i[-1])->b;
+}
+
+static inline bool same_leaf_as_next(struct btree_trans *trans,
+				     struct btree_insert_entry *i)
+{
+	return i + 1 < trans->updates + trans->nr_updates &&
+		insert_l(&i[0])->b == insert_l(&i[1])->b;
 }
 
-inline void bch2_btree_node_lock_for_insert(struct btree_trans *trans,
-					    struct btree_iter *iter,
-					    struct btree *b)
+static inline void bch2_btree_node_prep_for_write(struct btree_trans *trans,
+						  struct btree_path *path,
+						  struct btree *b)
 {
 	struct bch_fs *c = trans->c;
 
-	bch2_btree_node_lock_write(b, iter);
-
-	if (btree_iter_type(iter) == BTREE_ITER_CACHED)
+	if (path->cached)
 		return;
 
 	if (unlikely(btree_node_just_written(b)) &&
 	    bch2_btree_post_write_cleanup(c, b))
-		bch2_btree_iter_reinit_node(iter, b);
+		bch2_trans_node_reinit_iter(trans, b);
 
 	/*
 	 * If the last bset has been written, or if it's gotten too big - start
 	 * a new bset to insert into:
 	 */
 	if (want_new_bset(c, b))
-		bch2_btree_init_next(trans, iter, b);
+		bch2_btree_init_next(trans, b);
+}
+
+void bch2_btree_node_lock_for_insert(struct btree_trans *trans,
+				     struct btree_path *path,
+				     struct btree *b)
+{
+	bch2_btree_node_lock_write(trans, path, b);
+	bch2_btree_node_prep_for_write(trans, path, b);
 }
 
 /* Inserting into a given leaf node (last stage of insert): */
 
 /* Handle overwrites and do insert, for non extents: */
-bool bch2_btree_bset_insert_key(struct btree_iter *iter,
+bool bch2_btree_bset_insert_key(struct btree_trans *trans,
+				struct btree_path *path,
 				struct btree *b,
 				struct btree_node_iter *node_iter,
 				struct bkey_i *insert)
@@ -76,8 +95,7 @@ bool bch2_btree_bset_insert_key(struct btree_iter *iter,
 	EBUG_ON(bpos_cmp(insert->k.p, b->data->min_key) < 0);
 	EBUG_ON(bpos_cmp(insert->k.p, b->data->max_key) > 0);
 	EBUG_ON(insert->k.u64s >
-		bch_btree_keys_u64s_remaining(iter->trans->c, b));
-	EBUG_ON(iter->flags & BTREE_ITER_IS_EXTENTS);
+		bch_btree_keys_u64s_remaining(trans->c, b));
 
 	k = bch2_btree_node_iter_peek_all(node_iter, b);
 	if (k && bkey_cmp_left_packed(b, k, &insert->k.p))
@@ -96,7 +114,7 @@ bool bch2_btree_bset_insert_key(struct btree_iter *iter,
 		k->type = KEY_TYPE_deleted;
 
 		if (k->needs_whiteout)
-			push_whiteout(iter->trans->c, b, insert->k.p);
+			push_whiteout(trans->c, b, insert->k.p);
 		k->needs_whiteout = false;
 
 		if (k >= btree_bset_last(b)->start) {
@@ -104,7 +122,7 @@ bool bch2_btree_bset_insert_key(struct btree_iter *iter,
 			bch2_bset_delete(b, k, clobber_u64s);
 			goto fix_iter;
 		} else {
-			bch2_btree_iter_fix_key_modified(iter, b, k);
+			bch2_btree_path_fix_key_modified(trans, b, k);
 		}
 
 		return true;
@@ -122,7 +140,7 @@ bool bch2_btree_bset_insert_key(struct btree_iter *iter,
 			clobber_u64s = k->u64s;
 			goto overwrite;
 		} else {
-			bch2_btree_iter_fix_key_modified(iter, b, k);
+			bch2_btree_path_fix_key_modified(trans, b, k);
 		}
 	}
 
@@ -132,7 +150,7 @@ overwrite:
 	new_u64s = k->u64s;
 fix_iter:
 	if (clobber_u64s != new_u64s)
-		bch2_btree_node_iter_fix(iter, b, node_iter, k,
+		bch2_btree_node_iter_fix(trans, path, b, node_iter, k,
 					 clobber_u64s, new_u64s);
 	return true;
 }
@@ -176,22 +194,21 @@ inline void bch2_btree_add_journal_pin(struct bch_fs *c,
  * btree_insert_key - insert a key one key into a leaf node
  */
 static bool btree_insert_key_leaf(struct btree_trans *trans,
-				  struct btree_iter *iter,
-				  struct bkey_i *insert)
+				  struct btree_insert_entry *insert)
 {
 	struct bch_fs *c = trans->c;
-	struct btree *b = iter_l(iter)->b;
+	struct btree *b = insert_l(insert)->b;
 	struct bset_tree *t = bset_tree_last(b);
 	struct bset *i = bset(b, t);
 	int old_u64s = bset_u64s(t);
 	int old_live_u64s = b->nr.live_u64s;
 	int live_u64s_added, u64s_added;
 
-	EBUG_ON(!iter->level &&
+	EBUG_ON(!insert->level &&
 		!test_bit(BCH_FS_BTREE_INTERIOR_REPLAY_DONE, &c->flags));
 
-	if (unlikely(!bch2_btree_bset_insert_key(iter, b,
-					&iter_l(iter)->iter, insert)))
+	if (unlikely(!bch2_btree_bset_insert_key(trans, insert->path, b,
+					&insert_l(insert)->iter, insert->k)))
 		return false;
 
 	i->journal_seq = cpu_to_le64(max(trans->journal_res.seq,
@@ -212,9 +229,8 @@ static bool btree_insert_key_leaf(struct btree_trans *trans,
 
 	if (u64s_added > live_u64s_added &&
 	    bch2_maybe_compact_whiteouts(c, b))
-		bch2_btree_iter_reinit_node(iter, b);
+		bch2_trans_node_reinit_iter(trans, b);
 
-	trace_btree_insert_key(c, b, insert);
 	return true;
 }
 
@@ -225,9 +241,10 @@ static bool btree_insert_key_leaf(struct btree_trans *trans,
 static inline void btree_insert_entry_checks(struct btree_trans *trans,
 					     struct btree_insert_entry *i)
 {
-	BUG_ON(bpos_cmp(i->k->k.p, i->iter->real_pos));
-	BUG_ON(i->level		!= i->iter->level);
-	BUG_ON(i->btree_id	!= i->iter->btree_id);
+	BUG_ON(bpos_cmp(i->k->k.p, i->path->pos));
+	BUG_ON(i->cached	!= i->path->cached);
+	BUG_ON(i->level		!= i->path->level);
+	BUG_ON(i->btree_id	!= i->path->btree_id);
 }
 
 static noinline int
@@ -267,13 +284,12 @@ static inline int bch2_trans_journal_res_get(struct btree_trans *trans,
 	return ret == -EAGAIN ? BTREE_INSERT_NEED_JOURNAL_RES : ret;
 }
 
-static enum btree_insert_ret
+static inline enum btree_insert_ret
 btree_key_can_insert(struct btree_trans *trans,
-		     struct btree_iter *iter,
+		     struct btree *b,
 		     unsigned u64s)
 {
 	struct bch_fs *c = trans->c;
-	struct btree *b = iter_l(iter)->b;
 
 	if (!bch2_btree_node_insert_fits(c, b, u64s))
 		return BTREE_INSERT_BTREE_NODE_FULL;
@@ -283,14 +299,14 @@ btree_key_can_insert(struct btree_trans *trans,
 
 static enum btree_insert_ret
 btree_key_can_insert_cached(struct btree_trans *trans,
-			    struct btree_iter *iter,
+			    struct btree_path *path,
 			    unsigned u64s)
 {
-	struct bkey_cached *ck = (void *) iter->l[0].b;
+	struct bkey_cached *ck = (void *) path->l[0].b;
 	unsigned new_u64s;
 	struct bkey_i *new_k;
 
-	BUG_ON(iter->level);
+	EBUG_ON(path->level);
 
 	if (!test_bit(BKEY_CACHED_DIRTY, &ck->flags) &&
 	    bch2_btree_key_cache_must_wait(trans->c) &&
@@ -328,9 +344,9 @@ static inline void do_btree_insert_one(struct btree_trans *trans,
 
 	i->k->k.needs_whiteout = false;
 
-	did_work = (btree_iter_type(i->iter) != BTREE_ITER_CACHED)
-		? btree_insert_key_leaf(trans, i->iter, i->k)
-		: bch2_btree_insert_key_cached(trans, i->iter, i->k);
+	did_work = !i->cached
+		? btree_insert_key_leaf(trans, i)
+		: bch2_btree_insert_key_cached(trans, i->path, i->k);
 	if (!did_work)
 		return;
 
@@ -356,11 +372,12 @@ static noinline void bch2_trans_mark_gc(struct btree_trans *trans)
 	trans_for_each_update(trans, i) {
 		/*
 		 * XXX: synchronization of cached update triggers with gc
+		 * XXX: synchronization of interior node updates with gc
 		 */
-		BUG_ON(btree_iter_type(i->iter) == BTREE_ITER_CACHED);
+		BUG_ON(i->cached || i->level);
 
-		if (gc_visited(c, gc_pos_btree_node(i->iter->l[0].b)))
-			bch2_mark_update(trans, i->iter, i->k,
+		if (gc_visited(c, gc_pos_btree_node(insert_l(i)->b)))
+			bch2_mark_update(trans, i->path, i->k,
 					 i->flags|BTREE_TRIGGER_GC);
 	}
 }
@@ -405,9 +422,9 @@ bch2_trans_commit_write_locked(struct btree_trans *trans,
 			u64s = 0;
 
 		u64s += i->k->k.u64s;
-		ret = btree_iter_type(i->iter) != BTREE_ITER_CACHED
-			? btree_key_can_insert(trans, i->iter, u64s)
-			: btree_key_can_insert_cached(trans, i->iter, u64s);
+		ret = !i->cached
+			? btree_key_can_insert(trans, insert_l(i)->b, u64s)
+			: btree_key_can_insert_cached(trans, i->path, u64s);
 		if (ret) {
 			*stopped_at = i;
 			return ret;
@@ -466,8 +483,7 @@ bch2_trans_commit_write_locked(struct btree_trans *trans,
 
 	trans_for_each_update(trans, i)
 		if (BTREE_NODE_TYPE_HAS_MEM_TRIGGERS & (1U << i->bkey_type))
-			bch2_mark_update(trans, i->iter, i->k,
-					 i->flags);
+			bch2_mark_update(trans, i->path, i->k, i->flags);
 
 	if (marking && trans->fs_usage_deltas)
 		bch2_trans_fs_usage_apply(trans, trans->fs_usage_deltas);
@@ -485,42 +501,96 @@ err:
 	return ret;
 }
 
-static noinline int maybe_do_btree_merge(struct btree_trans *trans, struct btree_iter *iter)
+static inline void upgrade_readers(struct btree_trans *trans, struct btree_path *path)
 {
-	struct btree_insert_entry *i;
-	struct btree *b = iter_l(iter)->b;
-	struct bkey_s_c old;
-	int u64s_delta = 0;
-	int ret;
+	struct btree *b = path_l(path)->b;
 
-	/*
-	 * Inserting directly into interior nodes is an uncommon operation with
-	 * various weird edge cases: also, a lot of things about
-	 * BTREE_ITER_NODES iters need to be audited
-	 */
-	if (unlikely(btree_iter_type(iter) != BTREE_ITER_KEYS))
-		return 0;
+	do {
+		if (path->nodes_locked &&
+		    path->nodes_locked != path->nodes_intent_locked)
+			BUG_ON(!bch2_btree_path_upgrade(trans, path, path->level + 1));
+	} while ((path = prev_btree_path(trans, path)) &&
+		 path_l(path)->b == b);
+}
+
+/*
+ * Check for nodes that we have both read and intent locks on, and upgrade the
+ * readers to intent:
+ */
+static inline void normalize_read_intent_locks(struct btree_trans *trans)
+{
+	struct btree_path *path;
+	unsigned i, nr_read = 0, nr_intent = 0;
+
+	trans_for_each_path_inorder(trans, path, i) {
+		struct btree_path *next = i + 1 < trans->nr_sorted
+			? trans->paths + trans->sorted[i + 1]
+			: NULL;
+
+		if (path->nodes_locked) {
+			if (path->nodes_intent_locked)
+				nr_intent++;
+			else
+				nr_read++;
+		}
+
+		if (!next || path_l(path)->b != path_l(next)->b) {
+			if (nr_read && nr_intent)
+				upgrade_readers(trans, path);
+
+			nr_read = nr_intent = 0;
+		}
+	}
+
+	bch2_trans_verify_locks(trans);
+}
+
+static inline bool have_conflicting_read_lock(struct btree_trans *trans, struct btree_path *pos)
+{
+	struct btree_path *path;
+	unsigned i;
 
-	BUG_ON(iter->level);
+	trans_for_each_path_inorder(trans, path, i) {
+		//if (path == pos)
+		//	break;
+
+		if (path->nodes_locked != path->nodes_intent_locked)
+			return true;
+	}
+
+	return false;
+}
+
+static inline int trans_lock_write(struct btree_trans *trans)
+{
+	struct btree_insert_entry *i;
 
 	trans_for_each_update(trans, i) {
-		if (iter_l(i->iter)->b != b)
+		if (same_leaf_as_prev(trans, i))
 			continue;
 
-		old = bch2_btree_iter_peek_slot(i->iter);
-		ret = bkey_err(old);
-		if (ret)
-			return ret;
+		if (!six_trylock_write(&insert_l(i)->b->c.lock)) {
+			if (have_conflicting_read_lock(trans, i->path))
+				goto fail;
 
-		u64s_delta += !bkey_deleted(&i->k->k) ? i->k->k.u64s : 0;
-		u64s_delta -= !bkey_deleted(old.k) ? old.k->u64s : 0;
+			__btree_node_lock_type(trans->c, insert_l(i)->b,
+					       SIX_LOCK_write);
+		}
+
+		bch2_btree_node_prep_for_write(trans, i->path, insert_l(i)->b);
 	}
 
-	if (u64s_delta > 0)
-		return 0;
+	return 0;
+fail:
+	while (--i >= trans->updates) {
+		if (same_leaf_as_prev(trans, i))
+			continue;
+
+		bch2_btree_node_unlock_write_inlined(trans, i->path, insert_l(i)->b);
+	}
 
-	return bch2_foreground_maybe_merge(trans, iter,
-				iter->level, trans->flags);
+	trace_trans_restart_would_deadlock_write(trans->ip);
+	return btree_trans_restart(trans);
 }
 
 /*
@@ -532,29 +602,55 @@ static inline int do_bch2_trans_commit(struct btree_trans *trans,
 {
 	struct bch_fs *c = trans->c;
 	struct btree_insert_entry *i;
-	struct btree_iter *iter;
-	int ret;
+	struct bkey_s_c old;
+	int ret, u64s_delta = 0;
 
 	trans_for_each_update(trans, i) {
-		struct btree *b;
+		const char *invalid = bch2_bkey_invalid(c,
+				bkey_i_to_s_c(i->k), i->bkey_type);
+		if (invalid) {
+			char buf[200];
 
-		BUG_ON(!btree_node_intent_locked(i->iter, i->level));
+			bch2_bkey_val_to_text(&PBUF(buf), c, bkey_i_to_s_c(i->k));
+			bch_err(c, "invalid bkey %s on insert from %ps -> %ps: %s\n",
+				buf, (void *) trans->ip,
+				(void *) i->ip_allocated, invalid);
+			bch2_fatal_error(c);
+			return -EINVAL;
+		}
+		btree_insert_entry_checks(trans, i);
+	}
+
+	trans_for_each_update(trans, i) {
+		struct bkey u;
 
-		if (btree_iter_type(i->iter) == BTREE_ITER_CACHED)
+		/*
+		 * peek_slot() doesn't yet work on iterators that point to
+		 * interior nodes:
+		 */
+		if (i->cached || i->level)
 			continue;
 
-		b = iter_l(i->iter)->b;
-		if (b->sib_u64s[0] < c->btree_foreground_merge_threshold ||
-		    b->sib_u64s[1] < c->btree_foreground_merge_threshold) {
-			ret = maybe_do_btree_merge(trans, i->iter);
-			if (unlikely(ret))
-				return ret;
+		old = bch2_btree_path_peek_slot(i->path, &u);
+		ret = bkey_err(old);
+		if (unlikely(ret))
+			return ret;
+
+		u64s_delta += !bkey_deleted(&i->k->k) ? i->k->k.u64s : 0;
+		u64s_delta -= !bkey_deleted(old.k) ? old.k->u64s : 0;
+
+		if (!same_leaf_as_next(trans, i)) {
+			if (u64s_delta <= 0) {
+				ret = bch2_foreground_maybe_merge(trans, i->path,
+							i->level, trans->flags);
+				if (unlikely(ret))
+					return ret;
+			}
+
+			u64s_delta = 0;
 		}
 	}
 
-	trans_for_each_update(trans, i)
-		BUG_ON(!btree_node_intent_locked(i->iter, i->level));
-
 	ret = bch2_journal_preres_get(&c->journal,
 			&trans->journal_preres, trans->journal_preres_u64s,
 			JOURNAL_RES_GET_NONBLOCK|
@@ -566,52 +662,18 @@ static inline int do_bch2_trans_commit(struct btree_trans *trans,
 	if (unlikely(ret))
 		return ret;
 
-	/*
-	 * Can't be holding any read locks when we go to take write locks:
-	 * another thread could be holding an intent lock on the same node we
-	 * have a read lock on, and it'll block trying to take a write lock
-	 * (because we hold a read lock) and it could be blocking us by holding
-	 * its own read lock (while we're trying to to take write locks).
-	 *
-	 * note - this must be done after bch2_trans_journal_preres_get_cold()
-	 * or anything else that might call bch2_trans_relock(), since that
-	 * would just retake the read locks:
-	 */
-	trans_for_each_iter(trans, iter)
-		if (iter->nodes_locked != iter->nodes_intent_locked &&
-		    !bch2_btree_iter_upgrade(iter, 1)) {
-			trace_trans_restart_upgrade(trans->ip, trace_ip,
-						    iter->btree_id,
-						    &iter->real_pos);
-			trans->restarted = true;
-			return -EINTR;
-		}
-
-	trans_for_each_update(trans, i) {
-		const char *invalid = bch2_bkey_invalid(c,
-				bkey_i_to_s_c(i->k), i->bkey_type);
-		if (invalid) {
-			char buf[200];
-
-			bch2_bkey_val_to_text(&PBUF(buf), c, bkey_i_to_s_c(i->k));
-			bch_err(c, "invalid bkey %s on insert: %s\n", buf, invalid);
-			bch2_fatal_error(c);
-		}
-		btree_insert_entry_checks(trans, i);
-	}
-	bch2_btree_trans_verify_locks(trans);
+	normalize_read_intent_locks(trans);
 
-	trans_for_each_update(trans, i)
-		if (!same_leaf_as_prev(trans, i))
-			bch2_btree_node_lock_for_insert(trans, i->iter,
-					iter_l(i->iter)->b);
+	ret = trans_lock_write(trans);
+	if (unlikely(ret))
+		return ret;
 
 	ret = bch2_trans_commit_write_locked(trans, stopped_at, trace_ip);
 
 	trans_for_each_update(trans, i)
 		if (!same_leaf_as_prev(trans, i))
-			bch2_btree_node_unlock_write_inlined(iter_l(i->iter)->b,
-							     i->iter);
+			bch2_btree_node_unlock_write_inlined(trans, i->path,
+							insert_l(i)->b);
 
 	if (!ret && trans->journal_pin)
 		bch2_journal_pin_add(&c->journal, trans->journal_res.seq,
@@ -650,14 +712,13 @@ int bch2_trans_commit_error(struct btree_trans *trans,
 
 	switch (ret) {
 	case BTREE_INSERT_BTREE_NODE_FULL:
-		ret = bch2_btree_split_leaf(trans, i->iter, trans->flags);
+		ret = bch2_btree_split_leaf(trans, i->path, trans->flags);
 		if (!ret)
 			return 0;
 
 		if (ret == -EINTR)
 			trace_trans_restart_btree_node_split(trans->ip, trace_ip,
-							     i->iter->btree_id,
-							     &i->iter->real_pos);
+						i->btree_id, &i->path->pos);
 		break;
 	case BTREE_INSERT_NEED_MARK_REPLICAS:
 		bch2_trans_unlock(trans);
@@ -738,120 +799,9 @@ bch2_trans_commit_get_rw_cold(struct btree_trans *trans)
 	return 0;
 }
 
-static int extent_handle_overwrites(struct btree_trans *trans,
-				    struct btree_insert_entry *i)
-{
-	struct bch_fs *c = trans->c;
-	struct btree_iter *iter, *update_iter;
-	struct bpos start = bkey_start_pos(&i->k->k);
-	struct bkey_i *update;
-	struct bkey_s_c k;
-	int ret = 0, compressed_sectors;
-
-	iter = bch2_trans_get_iter(trans, i->btree_id, start,
-				   BTREE_ITER_INTENT|
-				   BTREE_ITER_WITH_UPDATES|
-				   BTREE_ITER_NOT_EXTENTS);
-	k = bch2_btree_iter_peek(iter);
-	if (!k.k || (ret = bkey_err(k)))
-		goto out;
-
-	if (bch2_bkey_maybe_mergable(k.k, &i->k->k)) {
-		update = bch2_trans_kmalloc(trans, bkey_bytes(k.k));
-		if ((ret = PTR_ERR_OR_ZERO(update)))
-			goto out;
-
-		bkey_reassemble(update, k);
-
-		if (bch2_bkey_merge(c, bkey_i_to_s(update), bkey_i_to_s_c(i->k))) {
-			update_iter = bch2_trans_copy_iter(trans, iter);
-			ret = bch2_btree_delete_at(trans, update_iter, i->flags);
-			bch2_trans_iter_put(trans, update_iter);
-
-			if (ret)
-				goto out;
-
-			i->k = update;
-			goto next;
-		}
-	}
-
-	if (!bkey_cmp(k.k->p, bkey_start_pos(&i->k->k)))
-		goto next;
-
-	while (bkey_cmp(i->k->k.p, bkey_start_pos(k.k)) > 0) {
-		/*
-		 * If we're going to be splitting a compressed extent, note it
-		 * so that __bch2_trans_commit() can increase our disk
-		 * reservation:
-		 */
-		if (bkey_cmp(bkey_start_pos(k.k), start) < 0 &&
-		    bkey_cmp(k.k->p, i->k->k.p) > 0 &&
-		    (compressed_sectors = bch2_bkey_sectors_compressed(k)))
-			trans->extra_journal_res += compressed_sectors;
-
-		if (bkey_cmp(bkey_start_pos(k.k), start) < 0) {
-			update = bch2_trans_kmalloc(trans, bkey_bytes(k.k));
-			if ((ret = PTR_ERR_OR_ZERO(update)))
-				goto out;
-
-			bkey_reassemble(update, k);
-
-			bch2_cut_back(start, update);
-
-			update_iter = bch2_trans_get_iter(trans, i->btree_id, update->k.p,
-							  BTREE_ITER_NOT_EXTENTS|
-							  BTREE_ITER_INTENT);
-			ret = bch2_btree_iter_traverse(update_iter);
-			if (ret) {
-				bch2_trans_iter_put(trans, update_iter);
-				goto out;
-			}
-
-			bch2_trans_update(trans, update_iter, update,
-					  BTREE_UPDATE_INTERNAL_SNAPSHOT_NODE|
-					  i->flags);
-			bch2_trans_iter_put(trans, update_iter);
-		}
-
-		if (bkey_cmp(k.k->p, i->k->k.p) <= 0) {
-			update_iter = bch2_trans_copy_iter(trans, iter);
-			ret = bch2_btree_delete_at(trans, update_iter,
-						   i->flags);
-			bch2_trans_iter_put(trans, update_iter);
-
-			if (ret)
-				goto out;
-		}
-
-		if (bkey_cmp(k.k->p, i->k->k.p) > 0) {
-			update = bch2_trans_kmalloc(trans, bkey_bytes(k.k));
-			if ((ret = PTR_ERR_OR_ZERO(update)))
-				goto out;
-
-			bkey_reassemble(update, k);
-			bch2_cut_front(i->k->k.p, update);
-
-			bch2_trans_update(trans, iter, update, i->flags);
-			goto out;
-		}
-next:
-		k = bch2_btree_iter_next(iter);
-		if (!k.k || (ret = bkey_err(k)))
-			goto out;
-	}
-
-	bch2_bkey_merge(c, bkey_i_to_s(i->k), k);
-out:
-	bch2_trans_iter_put(trans, iter);
-
-	return ret;
-}
-
 int __bch2_trans_commit(struct btree_trans *trans)
 {
 	struct btree_insert_entry *i = NULL;
-	struct btree_iter *iter;
 	bool trans_trigger_run;
 	unsigned u64s;
 	int ret = 0;
@@ -876,8 +826,12 @@ int __bch2_trans_commit(struct btree_trans *trans)
 	}
 
 #ifdef CONFIG_BCACHEFS_DEBUG
+	/*
+	 * if BTREE_TRIGGER_NORUN is set, it means we're probably being called
+	 * from the key cache flush code:
+	 */
 	trans_for_each_update(trans, i)
-		if (btree_iter_type(i->iter) != BTREE_ITER_CACHED &&
+		if (!i->cached &&
 		    !(i->flags & BTREE_TRIGGER_NORUN))
 			bch2_btree_key_cache_verify_clean(trans,
 					i->btree_id, i->k->k.p);
@@ -896,13 +850,12 @@ int __bch2_trans_commit(struct btree_trans *trans)
 				i->trans_triggers_run = true;
 				trans_trigger_run = true;
 
-				ret = bch2_trans_mark_update(trans, i->iter,
+				ret = bch2_trans_mark_update(trans, i->path,
 							     i->k, i->flags);
 				if (unlikely(ret)) {
 					if (ret == -EINTR)
 						trace_trans_restart_mark(trans->ip, _RET_IP_,
-									 i->iter->btree_id,
-									 &i->iter->pos);
+								i->btree_id, &i->path->pos);
 					goto out;
 				}
 			}
@@ -910,21 +863,19 @@ int __bch2_trans_commit(struct btree_trans *trans)
 	} while (trans_trigger_run);
 
 	trans_for_each_update(trans, i) {
-		BUG_ON(!i->iter->should_be_locked);
+		BUG_ON(!i->path->should_be_locked);
 
-		if (unlikely(!bch2_btree_iter_upgrade(i->iter, i->level + 1))) {
+		if (unlikely(!bch2_btree_path_upgrade(trans, i->path, i->level + 1))) {
 			trace_trans_restart_upgrade(trans->ip, _RET_IP_,
-						    i->iter->btree_id,
-						    &i->iter->pos);
-			trans->restarted = true;
-			ret = -EINTR;
+						    i->btree_id, &i->path->pos);
+			ret = btree_trans_restart(trans);
 			goto out;
 		}
 
-		BUG_ON(!btree_node_intent_locked(i->iter, i->level));
+		BUG_ON(!btree_node_intent_locked(i->path, i->level));
 
 		u64s = jset_u64s(i->k->k.u64s);
-		if (btree_iter_type(i->iter) == BTREE_ITER_CACHED &&
+		if (i->cached &&
 		    likely(!(trans->flags & BTREE_INSERT_JOURNAL_REPLAY)))
 			trans->journal_preres_u64s += u64s;
 		trans->journal_u64s += u64s;
@@ -945,21 +896,19 @@ retry:
 	ret = do_bch2_trans_commit(trans, &i, _RET_IP_);
 
 	/* make sure we didn't drop or screw up locks: */
-	bch2_btree_trans_verify_locks(trans);
+	bch2_trans_verify_locks(trans);
 
 	if (ret)
 		goto err;
-
-	trans_for_each_iter(trans, iter)
-		if (btree_iter_live(trans, iter) &&
-		    (iter->flags & BTREE_ITER_SET_POS_AFTER_COMMIT))
-			bch2_btree_iter_set_pos(iter, iter->pos_after_commit);
 out:
 	bch2_journal_preres_put(&trans->c->journal, &trans->journal_preres);
 
 	if (likely(!(trans->flags & BTREE_INSERT_NOCHECK_RW)))
 		percpu_ref_put(&trans->c->writes);
 out_reset:
+	trans_for_each_update(trans, i)
+		bch2_path_put(trans, i->path, true);
+
 	trans->extra_journal_res	= 0;
 	trans->nr_updates		= 0;
 	trans->hooks			= NULL;
@@ -982,22 +931,154 @@ err:
 	goto retry;
 }
 
+static int bch2_trans_update_extent(struct btree_trans *trans,
+				    struct btree_iter *orig_iter,
+				    struct bkey_i *insert,
+				    enum btree_update_flags flags)
+{
+	struct bch_fs *c = trans->c;
+	struct btree_iter iter, update_iter;
+	struct bpos start = bkey_start_pos(&insert->k);
+	struct bkey_i *update;
+	struct bkey_s_c k;
+	enum btree_id btree_id = orig_iter->btree_id;
+	int ret = 0, compressed_sectors;
+
+	bch2_trans_iter_init(trans, &iter, btree_id, start,
+			     BTREE_ITER_INTENT|
+			     BTREE_ITER_WITH_UPDATES|
+			     BTREE_ITER_NOT_EXTENTS);
+	k = bch2_btree_iter_peek(&iter);
+	if ((ret = bkey_err(k)))
+		goto err;
+	if (!k.k)
+		goto out;
+
+	if (bch2_bkey_maybe_mergable(k.k, &insert->k)) {
+		update = bch2_trans_kmalloc(trans, bkey_bytes(k.k));
+		if ((ret = PTR_ERR_OR_ZERO(update)))
+			goto err;
+
+		bkey_reassemble(update, k);
+
+		if (bch2_bkey_merge(c, bkey_i_to_s(update), bkey_i_to_s_c(insert))) {
+			ret = bch2_btree_delete_at(trans, &iter, flags);
+			if (ret)
+				goto err;
+
+			insert = update;
+			goto next;
+		}
+	}
+
+	if (!bkey_cmp(k.k->p, bkey_start_pos(&insert->k)))
+		goto next;
+
+	while (bkey_cmp(insert->k.p, bkey_start_pos(k.k)) > 0) {
+		/*
+		 * If we're going to be splitting a compressed extent, note it
+		 * so that __bch2_trans_commit() can increase our disk
+		 * reservation:
+		 */
+		if (bkey_cmp(bkey_start_pos(k.k), start) < 0 &&
+		    bkey_cmp(k.k->p, insert->k.p) > 0 &&
+		    (compressed_sectors = bch2_bkey_sectors_compressed(k)))
+			trans->extra_journal_res += compressed_sectors;
+
+		if (bkey_cmp(bkey_start_pos(k.k), start) < 0) {
+			update = bch2_trans_kmalloc(trans, bkey_bytes(k.k));
+			if ((ret = PTR_ERR_OR_ZERO(update)))
+				goto err;
+
+			bkey_reassemble(update, k);
+
+			bch2_cut_back(start, update);
+
+			bch2_trans_iter_init(trans, &update_iter, btree_id, update->k.p,
+					     BTREE_ITER_NOT_EXTENTS|
+					     BTREE_ITER_INTENT);
+			ret   = bch2_btree_iter_traverse(&update_iter) ?:
+				bch2_trans_update(trans, &update_iter, update,
+						  BTREE_UPDATE_INTERNAL_SNAPSHOT_NODE|
+						  flags);
+			bch2_trans_iter_exit(trans, &update_iter);
+			if (ret)
+				goto err;
+		}
+
+		if (bkey_cmp(k.k->p, insert->k.p) <= 0) {
+			ret = bch2_btree_delete_at(trans, &iter, flags);
+			if (ret)
+				goto err;
+		}
+
+		if (bkey_cmp(k.k->p, insert->k.p) > 0) {
+			update = bch2_trans_kmalloc(trans, bkey_bytes(k.k));
+			if ((ret = PTR_ERR_OR_ZERO(update)))
+				goto err;
+
+			bkey_reassemble(update, k);
+			bch2_cut_front(insert->k.p, update);
+
+			ret = bch2_trans_update(trans, &iter, update, flags);
+			if (ret)
+				goto err;
+
+			goto out;
+		}
+next:
+		k = bch2_btree_iter_next(&iter);
+		if ((ret = bkey_err(k)))
+			goto err;
+		if (!k.k)
+			goto out;
+	}
+
+	bch2_bkey_merge(c, bkey_i_to_s(insert), k);
+out:
+	if (!bkey_deleted(&insert->k)) {
+		/*
+		 * Rewinding iterators is expensive: get a new one and the one
+		 * that points to the start of insert will be cloned from:
+		 */
+		bch2_trans_iter_exit(trans, &iter);
+		bch2_trans_iter_init(trans, &iter, btree_id, insert->k.p,
+				     BTREE_ITER_NOT_EXTENTS|
+				     BTREE_ITER_INTENT);
+		ret   = bch2_btree_iter_traverse(&iter) ?:
+			bch2_trans_update(trans, &iter, insert, flags);
+	}
+err:
+	bch2_trans_iter_exit(trans, &iter);
+
+	return ret;
+}
+
 int bch2_trans_update(struct btree_trans *trans, struct btree_iter *iter,
 		      struct bkey_i *k, enum btree_update_flags flags)
 {
-	struct btree_insert_entry *i, n = (struct btree_insert_entry) {
+	struct btree_insert_entry *i, n;
+
+	BUG_ON(!iter->path->should_be_locked);
+
+	if (iter->flags & BTREE_ITER_IS_EXTENTS)
+		return bch2_trans_update_extent(trans, iter, k, flags);
+
+	BUG_ON(trans->nr_updates >= BTREE_ITER_MAX);
+	BUG_ON(bpos_cmp(k->k.p, iter->path->pos));
+
+	n = (struct btree_insert_entry) {
 		.flags		= flags,
-		.bkey_type	= __btree_node_type(iter->level, iter->btree_id),
+		.bkey_type	= __btree_node_type(iter->path->level, iter->btree_id),
 		.btree_id	= iter->btree_id,
-		.level		= iter->level,
-		.iter		= iter,
-		.k		= k
+		.level		= iter->path->level,
+		.cached		= iter->flags & BTREE_ITER_CACHED,
+		.path		= iter->path,
+		.k		= k,
+		.ip_allocated	= _RET_IP_,
 	};
-	bool is_extent = (iter->flags & BTREE_ITER_IS_EXTENTS) != 0;
-	int ret = 0;
 
-	BUG_ON(trans->nr_updates >= BTREE_ITER_MAX);
-	BUG_ON(!iter->should_be_locked);
+	__btree_path_get(n.path, true);
 
 #ifdef CONFIG_BCACHEFS_DEBUG
 	trans_for_each_update(trans, i)
@@ -1005,31 +1086,6 @@ int bch2_trans_update(struct btree_trans *trans, struct btree_iter *iter,
 		       btree_insert_entry_cmp(i - 1, i) >= 0);
 #endif
 
-	if (is_extent) {
-		ret = extent_handle_overwrites(trans, &n);
-		if (ret)
-			return ret;
-
-		iter->pos_after_commit = k->k.p;
-		iter->flags |= BTREE_ITER_SET_POS_AFTER_COMMIT;
-
-		if (bkey_deleted(&n.k->k))
-			return 0;
-
-		n.iter = bch2_trans_get_iter(trans, n.btree_id, n.k->k.p,
-					     BTREE_ITER_INTENT|
-					     BTREE_ITER_NOT_EXTENTS);
-		ret = bch2_btree_iter_traverse(n.iter);
-		bch2_trans_iter_put(trans, n.iter);
-
-		if (ret)
-			return ret;
-	}
-
-	BUG_ON(n.iter->flags & BTREE_ITER_IS_EXTENTS);
-
-	n.iter->flags |= BTREE_ITER_KEEP_UNTIL_COMMIT;
-
 	/*
 	 * Pending updates are kept sorted: first, find position of new update,
 	 * then delete/trim any updates the new update overwrites:
@@ -1047,11 +1103,13 @@ int bch2_trans_update(struct btree_trans *trans, struct btree_iter *iter,
 		 * not the key cache, which helps with cache coherency issues in
 		 * other areas:
 		 */
-		if (btree_iter_type(n.iter) == BTREE_ITER_CACHED &&
-		    btree_iter_type(i->iter) != BTREE_ITER_CACHED) {
+		if (n.cached && !i->cached) {
 			i->k = n.k;
 			i->flags = n.flags;
+
+			__btree_path_get(n.path, false);
 		} else {
+			bch2_path_put(trans, i->path, true);
 			*i = n;
 		}
 	} else
@@ -1071,15 +1129,14 @@ void bch2_trans_commit_hook(struct btree_trans *trans,
 int __bch2_btree_insert(struct btree_trans *trans,
 			enum btree_id id, struct bkey_i *k)
 {
-	struct btree_iter *iter;
+	struct btree_iter iter;
 	int ret;
 
-	iter = bch2_trans_get_iter(trans, id, bkey_start_pos(&k->k),
-				   BTREE_ITER_INTENT);
-
-	ret   = bch2_btree_iter_traverse(iter) ?:
-		bch2_trans_update(trans, iter, k, 0);
-	bch2_trans_iter_put(trans, iter);
+	bch2_trans_iter_init(trans, &iter, id, bkey_start_pos(&k->k),
+			     BTREE_ITER_INTENT);
+	ret   = bch2_btree_iter_traverse(&iter) ?:
+		bch2_trans_update(trans, &iter, k, 0);
+	bch2_trans_iter_exit(trans, &iter);
 	return ret;
 }
 
@@ -1117,16 +1174,16 @@ int bch2_btree_delete_range_trans(struct btree_trans *trans, enum btree_id id,
 				  struct bpos start, struct bpos end,
 				  u64 *journal_seq)
 {
-	struct btree_iter *iter;
+	struct btree_iter iter;
 	struct bkey_s_c k;
 	int ret = 0;
 
-	iter = bch2_trans_get_iter(trans, id, start, BTREE_ITER_INTENT);
+	bch2_trans_iter_init(trans, &iter, id, start, BTREE_ITER_INTENT);
 retry:
 	while ((bch2_trans_begin(trans),
-	       (k = bch2_btree_iter_peek(iter)).k) &&
+	       (k = bch2_btree_iter_peek(&iter)).k) &&
 	       !(ret = bkey_err(k)) &&
-	       bkey_cmp(iter->pos, end) < 0) {
+	       bkey_cmp(iter.pos, end) < 0) {
 		struct bkey_i delete;
 
 		bkey_init(&delete.k);
@@ -1145,9 +1202,9 @@ retry:
 		 * (bch2_btree_iter_peek() does guarantee that iter.pos >=
 		 * bkey_start_pos(k.k)).
 		 */
-		delete.k.p = iter->pos;
+		delete.k.p = iter.pos;
 
-		if (btree_node_type_is_extents(iter->btree_id)) {
+		if (btree_node_type_is_extents(id)) {
 			unsigned max_sectors =
 				KEY_SIZE_MAX & (~0 << trans->c->block_bits);
 
@@ -1155,12 +1212,12 @@ retry:
 			bch2_key_resize(&delete.k, max_sectors);
 			bch2_cut_back(end, &delete);
 
-			ret = bch2_extent_trim_atomic(&delete, iter);
+			ret = bch2_extent_trim_atomic(trans, &iter, &delete);
 			if (ret)
 				break;
 		}
 
-		ret   = bch2_trans_update(trans, iter, &delete, 0) ?:
+		ret   = bch2_trans_update(trans, &iter, &delete, 0) ?:
 			bch2_trans_commit(trans, NULL, journal_seq,
 					BTREE_INSERT_NOFAIL);
 		if (ret)
@@ -1174,7 +1231,7 @@ retry:
 		goto retry;
 	}
 
-	bch2_trans_iter_free(trans, iter);
+	bch2_trans_iter_exit(trans, &iter);
 	return ret;
 }