X-Git-Url: https://git.sesse.net/?a=blobdiff_plain;f=libbcachefs%2Fbtree_update_leaf.c;h=7eca9203be01a1e54c8e62c2951d56a410aebc3c;hb=a10a41fa2b1a917b0f3b34d20175867f968b2d12;hp=53b39de52c6bb3e052aff5f3823d6d8f387407b1;hpb=c598d91dcb0c7e95abdacb2711898ae14ab52ca1;p=bcachefs-tools-debian

diff --git a/libbcachefs/btree_update_leaf.c b/libbcachefs/btree_update_leaf.c
index 53b39de..7eca920 100644
--- a/libbcachefs/btree_update_leaf.c
+++ b/libbcachefs/btree_update_leaf.c
@@ -5,10 +5,14 @@
 #include "btree_io.h"
 #include "btree_iter.h"
 #include "btree_locking.h"
+#include "buckets.h"
 #include "debug.h"
+#include "error.h"
 #include "extents.h"
 #include "journal.h"
+#include "journal_reclaim.h"
 #include "keylist.h"
+#include "replicas.h"
 
 #include <linux/sort.h>
 #include <trace/events/bcachefs.h>
@@ -23,7 +27,6 @@ bool bch2_btree_bset_insert_key(struct btree_iter *iter,
 {
 	const struct bkey_format *f = &b->format;
 	struct bkey_packed *k;
-	struct bset_tree *t;
 	unsigned clobber_u64s;
 
 	EBUG_ON(btree_node_just_written(b));
@@ -36,9 +39,7 @@ bool bch2_btree_bset_insert_key(struct btree_iter *iter,
 	if (k && !bkey_cmp_packed(b, k, &insert->k)) {
 		BUG_ON(bkey_whiteout(k));
 
-		t = bch2_bkey_to_bset(b, k);
-
-		if (bset_unwritten(b, bset(b, t)) &&
+		if (!bkey_written(b, k) &&
 		    bkey_val_u64s(&insert->k) == bkeyp_val_u64s(f, k) &&
 		    !bkey_whiteout(&insert->k)) {
 			k->type = insert->k.type;
@@ -49,9 +50,9 @@ bool bch2_btree_bset_insert_key(struct btree_iter *iter,
 
 		insert->k.needs_whiteout = k->needs_whiteout;
 
-		btree_keys_account_key_drop(&b->nr, t - b->set, k);
+		btree_account_key_drop(b, k);
 
-		if (t == bset_tree_last(b)) {
+		if (k >= btree_bset_last(b)->start) {
 			clobber_u64s = k->u64s;
 
 			/*
@@ -61,20 +62,22 @@ bool bch2_btree_bset_insert_key(struct btree_iter *iter,
 			 */
 			if (bkey_whiteout(&insert->k) && !k->needs_whiteout) {
 				bch2_bset_delete(b, k, clobber_u64s);
-				bch2_btree_node_iter_fix(iter, b, node_iter, t,
-							k, clobber_u64s, 0);
+				bch2_btree_node_iter_fix(iter, b, node_iter,
+							 k, clobber_u64s, 0);
+				bch2_btree_iter_verify(iter, b);
 				return true;
 			}
 
 			goto overwrite;
 		}
 
-		k->type = KEY_TYPE_DELETED;
-		bch2_btree_node_iter_fix(iter, b, node_iter, t, k,
-					k->u64s, k->u64s);
+		k->type = KEY_TYPE_deleted;
+		bch2_btree_node_iter_fix(iter, b, node_iter, k,
+					 k->u64s, k->u64s);
+		bch2_btree_iter_verify(iter, b);
 
 		if (bkey_whiteout(&insert->k)) {
-			reserve_whiteout(b, t, k);
+			reserve_whiteout(b, k);
 			return true;
 		} else {
 			k->needs_whiteout = false;
@@ -89,14 +92,14 @@ bool bch2_btree_bset_insert_key(struct btree_iter *iter,
 		insert->k.needs_whiteout = false;
 	}
 
-	t = bset_tree_last(b);
-	k = bch2_btree_node_iter_bset_pos(node_iter, b, t);
+	k = bch2_btree_node_iter_bset_pos(node_iter, b, bset_tree_last(b));
 	clobber_u64s = 0;
 overwrite:
 	bch2_bset_insert(b, node_iter, k, insert, clobber_u64s);
 	if (k->u64s != clobber_u64s || bkey_whiteout(&insert->k))
-		bch2_btree_node_iter_fix(iter, b, node_iter, t, k,
-					clobber_u64s, k->u64s);
+		bch2_btree_node_iter_fix(iter, b, node_iter, k,
+					 clobber_u64s, k->u64s);
+	bch2_btree_iter_verify(iter, b);
 	return true;
 }
 
@@ -107,10 +110,9 @@ static void __btree_node_flush(struct journal *j, struct journal_entry_pin *pin,
 	struct btree_write *w = container_of(pin, struct btree_write, journal);
 	struct btree *b = container_of(w, struct btree, writes[i]);
 
-	six_lock_read(&b->lock);
+	btree_node_lock_type(c, b, SIX_LOCK_read);
 	bch2_btree_node_write_cond(c, b,
-			(btree_current_write(b) == w &&
-			 w->journal.pin_list == journal_seq_pin(j, seq)));
+		(btree_current_write(b) == w && w->journal.seq == seq));
 	six_unlock_read(&b->lock);
 }
 
@@ -124,6 +126,27 @@ static void btree_node_flush1(struct journal *j, struct journal_entry_pin *pin,
 	return __btree_node_flush(j, pin, 1, seq);
 }
 
+static inline void __btree_journal_key(struct btree_insert *trans,
+				       enum btree_id btree_id,
+				       struct bkey_i *insert)
+{
+	struct journal *j = &trans->c->journal;
+	u64 seq = trans->journal_res.seq;
+	bool needs_whiteout = insert->k.needs_whiteout;
+
+	/* ick */
+	insert->k.needs_whiteout = false;
+	bch2_journal_add_keys(j, &trans->journal_res,
+			      btree_id, insert);
+	insert->k.needs_whiteout = needs_whiteout;
+
+	bch2_journal_set_has_inode(j, &trans->journal_res,
+				   insert->k.p.inode);
+
+	if (trans->journal_seq)
+		*trans->journal_seq = seq;
+}
+
 void bch2_btree_journal_key(struct btree_insert *trans,
 			   struct btree_iter *iter,
 			   struct bkey_i *insert)
@@ -137,30 +160,22 @@ void bch2_btree_journal_key(struct btree_insert *trans,
 	EBUG_ON(trans->journal_res.ref !=
 		!(trans->flags & BTREE_INSERT_JOURNAL_REPLAY));
 
-	if (likely(trans->journal_res.ref)) {
-		u64 seq = trans->journal_res.seq;
-		bool needs_whiteout = insert->k.needs_whiteout;
-
-		/* ick */
-		insert->k.needs_whiteout = false;
-		bch2_journal_add_keys(j, &trans->journal_res,
-				      iter->btree_id, insert);
-		insert->k.needs_whiteout = needs_whiteout;
-
-		bch2_journal_set_has_inode(j, &trans->journal_res,
-					   insert->k.p.inode);
-
-		if (trans->journal_seq)
-			*trans->journal_seq = seq;
-		btree_bset_last(b)->journal_seq = cpu_to_le64(seq);
+	if (likely(!(trans->flags & BTREE_INSERT_JOURNAL_REPLAY))) {
+		__btree_journal_key(trans, iter->btree_id, insert);
+		btree_bset_last(b)->journal_seq =
+			cpu_to_le64(trans->journal_res.seq);
 	}
 
-	if (unlikely(!journal_pin_active(&w->journal)))
-		bch2_journal_pin_add(j, &trans->journal_res,
-				     &w->journal,
+	if (unlikely(!journal_pin_active(&w->journal))) {
+		u64 seq = likely(!(trans->flags & BTREE_INSERT_JOURNAL_REPLAY))
+			? trans->journal_res.seq
+			: j->replay_journal_seq;
+
+		bch2_journal_pin_add(j, seq, &w->journal,
 				     btree_node_write_idx(b) == 0
 				     ? btree_node_flush0
 				     : btree_node_flush1);
+	}
 
 	if (unlikely(!btree_node_dirty(b)))
 		set_btree_node_dirty(b);
@@ -181,7 +196,6 @@ bch2_insert_fixup_key(struct btree_insert *trans,
 				       insert->k))
 		bch2_btree_journal_key(trans, iter, insert->k);
 
-	trans->did_work = true;
 	return BTREE_INSERT_OK;
 }
 
@@ -200,7 +214,7 @@ btree_insert_key_leaf(struct btree_insert *trans,
 	int old_live_u64s = b->nr.live_u64s;
 	int live_u64s_added, u64s_added;
 
-	btree_iter_set_dirty(iter, BTREE_ITER_NEED_PEEK);
+	bch2_mark_update(trans, insert);
 
 	ret = !btree_node_is_extents(b)
 		? bch2_insert_fixup_key(trans, insert)
@@ -222,19 +236,144 @@ btree_insert_key_leaf(struct btree_insert *trans,
 	return ret;
 }
 
+/* Deferred btree updates: */
+
+static void deferred_update_flush(struct journal *j,
+					struct journal_entry_pin *pin,
+					u64 seq)
+{
+	struct bch_fs *c = container_of(j, struct bch_fs, journal);
+	struct deferred_update *d =
+		container_of(pin, struct deferred_update, journal);
+	u64 tmp[32];
+	struct bkey_i *k = (void *) tmp;
+	unsigned gen;
+	int ret;
+
+	if (d->allocated_u64s > ARRAY_SIZE(tmp)) {
+		k = kmalloc(d->allocated_u64s * sizeof(u64), GFP_NOFS);
+
+		BUG_ON(!k); /* XXX */
+	}
+
+	spin_lock(&d->lock);
+	gen = d->gen;
+
+	if (journal_pin_active(&d->journal)) {
+		BUG_ON(d->k.k.u64s > d->allocated_u64s);
+		bkey_copy(k, &d->k);
+
+		spin_unlock(&d->lock);
+
+		ret = bch2_btree_insert(c, d->btree_id, k, NULL, NULL,
+					BTREE_INSERT_NOFAIL);
+		bch2_fs_fatal_err_on(ret && !bch2_journal_error(j),
+			c, "error flushing deferred btree update: %i", ret);
+
+		spin_lock(&d->lock);
+	}
+
+	if (gen == d->gen)
+		bch2_journal_pin_drop(j, &d->journal);
+	spin_unlock(&d->lock);
+
+	if (k != (void *) tmp)
+		kfree(k);
+}
+
+static enum btree_insert_ret
+btree_insert_key_deferred(struct btree_insert *trans,
+			  struct btree_insert_entry *insert)
+{
+	struct bch_fs *c = trans->c;
+	struct journal *j = &c->journal;
+	struct deferred_update *d = insert->d;
+
+	BUG_ON(trans->flags & BTREE_INSERT_JOURNAL_REPLAY);
+	BUG_ON(insert->k->u64s > d->allocated_u64s);
+
+	__btree_journal_key(trans, d->btree_id, insert->k);
+
+	spin_lock(&d->lock);
+	d->gen++;
+	bkey_copy(&d->k, insert->k);
+	spin_unlock(&d->lock);
+
+	bch2_journal_pin_update(j, trans->journal_res.seq, &d->journal,
+				deferred_update_flush);
+
+	return BTREE_INSERT_OK;
+}
+
+void bch2_deferred_update_free(struct bch_fs *c,
+			       struct deferred_update *d)
+{
+	deferred_update_flush(&c->journal, &d->journal, 0);
+
+	BUG_ON(journal_pin_active(&d->journal));
+
+	bch2_journal_pin_flush(&c->journal, &d->journal);
+	kfree(d);
+}
+
+struct deferred_update *
+bch2_deferred_update_alloc(struct bch_fs *c,
+			   enum btree_id btree_id,
+			   unsigned u64s)
+{
+	struct deferred_update *d;
+
+	BUG_ON(u64s > U8_MAX);
+
+	d = kmalloc(offsetof(struct deferred_update, k) +
+		    u64s * sizeof(u64), GFP_NOFS);
+	BUG_ON(!d);
+
+	memset(d, 0, offsetof(struct deferred_update, k));
+
+	spin_lock_init(&d->lock);
+	d->allocated_u64s	= u64s;
+	d->btree_id		= btree_id;
+
+	return d;
+}
+
+/* struct btree_insert operations: */
+
+/*
+ * We sort transaction entries so that if multiple iterators point to the same
+ * leaf node they'll be adjacent:
+ */
 static bool same_leaf_as_prev(struct btree_insert *trans,
 			      struct btree_insert_entry *i)
 {
-	/*
-	 * Because we sorted the transaction entries, if multiple iterators
-	 * point to the same leaf node they'll always be adjacent now:
-	 */
 	return i != trans->entries &&
+		!i->deferred &&
 		i[0].iter->l[0].b == i[-1].iter->l[0].b;
 }
 
+#define __trans_next_entry(_trans, _i, _filter)				\
+({									\
+	while ((_i) < (_trans)->entries + (_trans->nr) && !(_filter))	\
+		(_i)++;							\
+									\
+	(_i) < (_trans)->entries + (_trans->nr);			\
+})
+
+#define __trans_for_each_entry(_trans, _i, _filter)			\
+	for ((_i) = (_trans)->entries;					\
+	     __trans_next_entry(_trans, _i, _filter);			\
+	     (_i)++)
+
 #define trans_for_each_entry(trans, i)					\
-	for ((i) = (trans)->entries; (i) < (trans)->entries + (trans)->nr; (i)++)
+	__trans_for_each_entry(trans, i, true)
+
+#define trans_for_each_iter(trans, i)					\
+	__trans_for_each_entry(trans, i, !(i)->deferred)
+
+#define trans_for_each_leaf(trans, i)					\
+	__trans_for_each_entry(trans, i, !(i)->deferred &&		\
+			       !same_leaf_as_prev(trans, i))
 
 inline void bch2_btree_node_lock_for_insert(struct bch_fs *c, struct btree *b,
 					    struct btree_iter *iter)
@@ -257,238 +396,405 @@ static void multi_lock_write(struct bch_fs *c, struct btree_insert *trans)
 {
 	struct btree_insert_entry *i;
 
-	trans_for_each_entry(trans, i)
-		if (!same_leaf_as_prev(trans, i))
-			bch2_btree_node_lock_for_insert(c, i->iter->l[0].b,
-							i->iter);
+	trans_for_each_leaf(trans, i)
+		bch2_btree_node_lock_for_insert(c, i->iter->l[0].b, i->iter);
 }
 
 static void multi_unlock_write(struct btree_insert *trans)
 {
 	struct btree_insert_entry *i;
 
-	trans_for_each_entry(trans, i)
-		if (!same_leaf_as_prev(trans, i))
-			bch2_btree_node_unlock_write(i->iter->l[0].b, i->iter);
+	trans_for_each_leaf(trans, i)
+		bch2_btree_node_unlock_write(i->iter->l[0].b, i->iter);
 }
 
 static inline int btree_trans_cmp(struct btree_insert_entry l,
 				  struct btree_insert_entry r)
 {
-	return btree_iter_cmp(l.iter, r.iter);
+	return (l.deferred > r.deferred) - (l.deferred < r.deferred) ?:
+		btree_iter_cmp(l.iter, r.iter);
 }
 
 /* Normal update interface: */
 
-/**
- * __bch_btree_insert_at - insert keys at given iterator positions
- *
- * This is main entry point for btree updates.
- *
- * Return values:
- * -EINTR: locking changed, this function should be called again. Only returned
- *  if passed BTREE_INSERT_ATOMIC.
- * -EROFS: filesystem read only
- * -EIO: journal or btree node IO error
+static enum btree_insert_ret
+btree_key_can_insert(struct btree_insert *trans,
+		     struct btree_insert_entry *insert,
+		     unsigned *u64s)
+{
+	struct bch_fs *c = trans->c;
+	struct btree *b = insert->iter->l[0].b;
+	static enum btree_insert_ret ret;
+
+	if (unlikely(btree_node_fake(b)))
+		return BTREE_INSERT_BTREE_NODE_FULL;
+
+	if (!bch2_bkey_replicas_marked(c,
+			bkey_i_to_s_c(insert->k),
+			true))
+		return BTREE_INSERT_NEED_MARK_REPLICAS;
+
+	ret = !btree_node_is_extents(b)
+		? BTREE_INSERT_OK
+		: bch2_extent_can_insert(trans, insert, u64s);
+	if (ret)
+		return ret;
+
+	if (*u64s > bch_btree_keys_u64s_remaining(c, b))
+		return BTREE_INSERT_BTREE_NODE_FULL;
+
+	return BTREE_INSERT_OK;
+}
+
+static inline enum btree_insert_ret
+do_btree_insert_one(struct btree_insert *trans,
+		    struct btree_insert_entry *insert)
+{
+	return likely(!insert->deferred)
+		? btree_insert_key_leaf(trans, insert)
+		: btree_insert_key_deferred(trans, insert);
+}
+
+/*
+ * Get journal reservation, take write locks, and attempt to do btree update(s):
  */
-int __bch2_btree_insert_at(struct btree_insert *trans)
+static inline int do_btree_insert_at(struct btree_insert *trans,
+				     struct btree_insert_entry **stopped_at)
 {
 	struct bch_fs *c = trans->c;
 	struct btree_insert_entry *i;
-	struct btree_iter *split = NULL;
-	bool cycle_gc_lock = false;
+	struct btree_iter *linked;
 	unsigned u64s;
 	int ret;
 
-	trans_for_each_entry(trans, i) {
-		BUG_ON(i->iter->level);
-		BUG_ON(bkey_cmp(bkey_start_pos(&i->k->k), i->iter->pos));
-		BUG_ON(debug_check_bkeys(c) &&
-		       bch2_bkey_invalid(c, i->iter->btree_id,
-					 bkey_i_to_s_c(i->k)));
+	trans_for_each_iter(trans, i)
+		BUG_ON(i->iter->uptodate >= BTREE_ITER_NEED_RELOCK);
+
+	/* reserve space for deferred updates */
+	__trans_for_each_entry(trans, i, i->deferred) {
+
 	}
 
-	bubble_sort(trans->entries, trans->nr, btree_trans_cmp);
+	memset(&trans->journal_res, 0, sizeof(trans->journal_res));
 
-	if (unlikely(!percpu_ref_tryget(&c->writes)))
-		return -EROFS;
-retry_locks:
-	ret = -EINTR;
-	trans_for_each_entry(trans, i) {
-		if (!bch2_btree_iter_set_locks_want(i->iter, 1))
-			goto err;
+	if (likely(!(trans->flags & BTREE_INSERT_JOURNAL_REPLAY))) {
+		u64s = 0;
+		trans_for_each_entry(trans, i)
+			u64s += jset_u64s(i->k->k.u64s);
+
+		while ((ret = bch2_journal_res_get(&c->journal,
+					&trans->journal_res, u64s,
+					JOURNAL_RES_GET_NONBLOCK)) == -EAGAIN) {
+			struct btree_iter *iter = NULL;
+
+			trans_for_each_iter(trans, i)
+				iter = i->iter;
 
-		if (i->iter->uptodate == BTREE_ITER_NEED_TRAVERSE) {
-			ret = bch2_btree_iter_traverse(i->iter);
+			if (iter)
+				bch2_btree_iter_unlock(iter);
+
+			ret = bch2_journal_res_get(&c->journal,
+					&trans->journal_res, u64s,
+					JOURNAL_RES_GET_CHECK);
 			if (ret)
-				goto err;
-		}
-	}
-retry:
-	trans->did_work = false;
-	u64s = 0;
-	trans_for_each_entry(trans, i)
-		if (!i->done)
-			u64s += jset_u64s(i->k->k.u64s + i->extra_res);
+				return ret;
 
-	memset(&trans->journal_res, 0, sizeof(trans->journal_res));
+			if (iter && !bch2_btree_iter_relock(iter)) {
+				trans_restart(" (iter relock after journal res get blocked)");
+				return -EINTR;
+			}
+		}
 
-	ret = !(trans->flags & BTREE_INSERT_JOURNAL_REPLAY)
-		? bch2_journal_res_get(&c->journal,
-				      &trans->journal_res,
-				      u64s, u64s)
-		: 0;
-	if (ret)
-		goto err;
+		if (ret)
+			return ret;
+	}
 
 	multi_lock_write(c, trans);
 
 	if (race_fault()) {
 		ret = -EINTR;
-		goto unlock;
+		trans_restart(" (race)");
+		goto out;
 	}
 
+	/*
+	 * Check if the insert will fit in the leaf node with the write lock
+	 * held, otherwise another thread could write the node changing the
+	 * amount of space available:
+	 */
 	u64s = 0;
-	trans_for_each_entry(trans, i) {
+	trans_for_each_iter(trans, i) {
 		/* Multiple inserts might go to same leaf: */
 		if (!same_leaf_as_prev(trans, i))
 			u64s = 0;
 
+		u64s += i->k->k.u64s;
+		ret = btree_key_can_insert(trans, i, &u64s);
+		if (ret) {
+			*stopped_at = i;
+			goto out;
+		}
+	}
+
+	if (!(trans->flags & BTREE_INSERT_JOURNAL_REPLAY)) {
+		if (journal_seq_verify(c))
+			trans_for_each_entry(trans, i)
+				i->k->k.version.lo = trans->journal_res.seq;
+		else if (inject_invalid_keys(c))
+			trans_for_each_entry(trans, i)
+				i->k->k.version = MAX_VERSION;
+	}
+
+	if (trans->flags & BTREE_INSERT_NOUNLOCK) {
 		/*
-		 * bch2_btree_node_insert_fits() must be called under write lock:
-		 * with only an intent lock, another thread can still call
-		 * bch2_btree_node_write(), converting an unwritten bset to a
-		 * written one
+		 * linked iterators that weren't being updated may or may not
+		 * have been traversed/locked, depending on what the caller was
+		 * doing:
 		 */
-		if (!i->done) {
-			u64s += i->k->k.u64s + i->extra_res;
-			if (!bch2_btree_node_insert_fits(c,
-					i->iter->l[0].b, u64s)) {
-				split = i->iter;
-				goto unlock;
-			}
+		trans_for_each_iter(trans, i) {
+			for_each_btree_iter(i->iter, linked)
+				if (linked->uptodate < BTREE_ITER_NEED_RELOCK)
+					linked->flags |= BTREE_ITER_NOUNLOCK;
+			break;
 		}
 	}
-
-	ret = 0;
-	split = NULL;
-	cycle_gc_lock = false;
+	trans->did_work = true;
 
 	trans_for_each_entry(trans, i) {
-		if (i->done)
-			continue;
-
-		switch (btree_insert_key_leaf(trans, i)) {
+		switch (do_btree_insert_one(trans, i)) {
 		case BTREE_INSERT_OK:
-			i->done = true;
 			break;
-		case BTREE_INSERT_JOURNAL_RES_FULL:
 		case BTREE_INSERT_NEED_TRAVERSE:
+			BUG_ON((trans->flags &
+				(BTREE_INSERT_ATOMIC|BTREE_INSERT_NOUNLOCK)));
 			ret = -EINTR;
-			break;
-		case BTREE_INSERT_NEED_RESCHED:
-			ret = -EAGAIN;
-			break;
-		case BTREE_INSERT_BTREE_NODE_FULL:
-			split = i->iter;
-			break;
-		case BTREE_INSERT_ENOSPC:
-			ret = -ENOSPC;
-			break;
-		case BTREE_INSERT_NEED_GC_LOCK:
-			cycle_gc_lock = true;
-			ret = -EINTR;
-			break;
+			goto out;
 		default:
 			BUG();
 		}
-
-		if (!trans->did_work && (ret || split))
-			break;
 	}
-unlock:
+out:
 	multi_unlock_write(trans);
 	bch2_journal_res_put(&c->journal, &trans->journal_res);
 
-	if (split)
-		goto split;
-	if (ret)
-		goto err;
+	return ret;
+}
+
+static inline void btree_insert_entry_checks(struct bch_fs *c,
+					     struct btree_insert_entry *i)
+{
+	enum btree_id btree_id = !i->deferred
+		? i->iter->btree_id
+		: i->d->btree_id;
+
+	if (!i->deferred) {
+		BUG_ON(i->iter->level);
+		BUG_ON(bkey_cmp(bkey_start_pos(&i->k->k), i->iter->pos));
+
+		bch2_btree_iter_verify_locks(i->iter);
+	}
+
+	BUG_ON(debug_check_bkeys(c) &&
+	       !bkey_deleted(&i->k->k) &&
+	       bch2_bkey_invalid(c, bkey_i_to_s_c(i->k), btree_id));
+}
+
+/**
+ * __bch_btree_insert_at - insert keys at given iterator positions
+ *
+ * This is main entry point for btree updates.
+ *
+ * Return values:
+ * -EINTR: locking changed, this function should be called again. Only returned
+ *  if passed BTREE_INSERT_ATOMIC.
+ * -EROFS: filesystem read only
+ * -EIO: journal or btree node IO error
+ */
+int __bch2_btree_insert_at(struct btree_insert *trans)
+{
+	struct bch_fs *c = trans->c;
+	struct btree_insert_entry *i;
+	struct btree_iter *linked;
+	unsigned flags;
+	int ret;
+
+	BUG_ON(!trans->nr);
+
+	/* for the sake of sanity: */
+	BUG_ON(trans->nr > 1 && !(trans->flags & BTREE_INSERT_ATOMIC));
+
+	bubble_sort(trans->entries, trans->nr, btree_trans_cmp);
 
 	trans_for_each_entry(trans, i)
-		if (i->iter->flags & BTREE_ITER_AT_END_OF_LEAF)
-			goto out;
+		btree_insert_entry_checks(c, i);
 
-	trans_for_each_entry(trans, i) {
-		/*
-		 * iterators are inconsistent when they hit end of leaf, until
-		 * traversed again
-		 */
-		if (i->iter->uptodate < BTREE_ITER_NEED_TRAVERSE &&
-		    !same_leaf_as_prev(trans, i))
-			bch2_foreground_maybe_merge(c, i->iter, 0);
+	if (unlikely(!percpu_ref_tryget(&c->writes)))
+		return -EROFS;
+retry:
+	trans_for_each_iter(trans, i) {
+		unsigned old_locks_want = i->iter->locks_want;
+		unsigned old_uptodate = i->iter->uptodate;
+
+		if (!bch2_btree_iter_upgrade(i->iter, 1, true)) {
+			trans_restart(" (failed upgrade, locks_want %u uptodate %u)",
+				      old_locks_want, old_uptodate);
+			ret = -EINTR;
+			goto err;
+		}
+
+		if (i->iter->flags & BTREE_ITER_ERROR) {
+			ret = -EIO;
+			goto err;
+		}
 	}
-out:
-	/* make sure we didn't lose an error: */
-	if (!ret && IS_ENABLED(CONFIG_BCACHEFS_DEBUG))
-		trans_for_each_entry(trans, i)
-			BUG_ON(!i->done);
 
+	ret = do_btree_insert_at(trans, &i);
+	if (unlikely(ret))
+		goto err;
+
+	trans_for_each_leaf(trans, i)
+		bch2_foreground_maybe_merge(c, i->iter, 0, trans->flags);
+
+	trans_for_each_iter(trans, i)
+		bch2_btree_iter_downgrade(i->iter);
+out:
 	percpu_ref_put(&c->writes);
+
+	/* make sure we didn't drop or screw up locks: */
+	trans_for_each_iter(trans, i) {
+		bch2_btree_iter_verify_locks(i->iter);
+		break;
+	}
+
+	trans_for_each_iter(trans, i) {
+		for_each_btree_iter(i->iter, linked)
+			linked->flags &= ~BTREE_ITER_NOUNLOCK;
+		break;
+	}
+
+	BUG_ON(!(trans->flags & BTREE_INSERT_ATOMIC) && ret == -EINTR);
+
 	return ret;
-split:
-	/*
-	 * have to drop journal res before splitting, because splitting means
-	 * allocating new btree nodes, and holding a journal reservation
-	 * potentially blocks the allocator:
-	 */
-	ret = bch2_btree_split_leaf(c, split, trans->flags);
+err:
+	flags = trans->flags;
 
 	/*
-	 * This can happen when we insert part of an extent - with an update
-	 * with multiple keys, we don't want to redo the entire update - that's
-	 * just too confusing:
+	 * BTREE_INSERT_NOUNLOCK means don't unlock _after_ successful btree
+	 * update; if we haven't done anything yet it doesn't apply
 	 */
-	if (!ret &&
-	    (trans->flags & BTREE_INSERT_ATOMIC) &&
-	    trans->did_work)
+	if (!trans->did_work)
+		flags &= ~BTREE_INSERT_NOUNLOCK;
+
+	switch (ret) {
+	case BTREE_INSERT_BTREE_NODE_FULL:
+		ret = bch2_btree_split_leaf(c, i->iter, flags);
+
+		/*
+		 * if the split succeeded without dropping locks the insert will
+		 * still be atomic (in the BTREE_INSERT_ATOMIC sense, what the
+		 * caller peeked() and is overwriting won't have changed)
+		 */
+#if 0
+		/*
+		 * XXX:
+		 * split -> btree node merging (of parent node) might still drop
+		 * locks when we're not passing it BTREE_INSERT_NOUNLOCK
+		 */
+		if (!ret && !trans->did_work)
+			goto retry;
+#endif
+
+		/*
+		 * don't care if we got ENOSPC because we told split it
+		 * couldn't block:
+		 */
+		if (!ret || (flags & BTREE_INSERT_NOUNLOCK)) {
+			trans_restart(" (split)");
+			ret = -EINTR;
+		}
+		break;
+	case BTREE_INSERT_NEED_GC_LOCK:
 		ret = -EINTR;
 
-	if (ret)
-		goto err;
+		if (!down_read_trylock(&c->gc_lock)) {
+			if (flags & BTREE_INSERT_NOUNLOCK)
+				goto out;
 
-	/*
-	 * if the split didn't have to drop locks the insert will still be
-	 * atomic (in the BTREE_INSERT_ATOMIC sense, what the caller peeked()
-	 * and is overwriting won't have changed)
-	 */
-	goto retry_locks;
-err:
-	if (cycle_gc_lock) {
-		down_read(&c->gc_lock);
+			bch2_btree_iter_unlock(trans->entries[0].iter);
+			down_read(&c->gc_lock);
+		}
 		up_read(&c->gc_lock);
+		break;
+	case BTREE_INSERT_ENOSPC:
+		ret = -ENOSPC;
+		break;
+	case BTREE_INSERT_NEED_MARK_REPLICAS:
+		if (flags & BTREE_INSERT_NOUNLOCK) {
+			ret = -EINTR;
+			goto out;
+		}
+
+		bch2_btree_iter_unlock(trans->entries[0].iter);
+		ret = bch2_mark_bkey_replicas(c, bkey_i_to_s_c(i->k))
+			?: -EINTR;
+		break;
+	default:
+		BUG_ON(ret >= 0);
+		break;
 	}
 
 	if (ret == -EINTR) {
-		trans_for_each_entry(trans, i) {
+		if (flags & BTREE_INSERT_NOUNLOCK) {
+			trans_restart(" (can't unlock)");
+			goto out;
+		}
+
+		trans_for_each_iter(trans, i) {
 			int ret2 = bch2_btree_iter_traverse(i->iter);
 			if (ret2) {
 				ret = ret2;
+				trans_restart(" (traverse)");
 				goto out;
 			}
+
+			BUG_ON(i->iter->uptodate > BTREE_ITER_NEED_PEEK);
 		}
 
 		/*
 		 * BTREE_ITER_ATOMIC means we have to return -EINTR if we
 		 * dropped locks:
 		 */
-		if (!(trans->flags & BTREE_INSERT_ATOMIC))
+		if (!(flags & BTREE_INSERT_ATOMIC))
 			goto retry;
+
+		trans_restart(" (atomic)");
 	}
 
 	goto out;
 }
 
+int bch2_trans_commit(struct btree_trans *trans,
+		      struct disk_reservation *disk_res,
+		      u64 *journal_seq,
+		      unsigned flags)
+{
+	struct btree_insert insert = {
+		.c		= trans->c,
+		.disk_res	= disk_res,
+		.journal_seq	= journal_seq,
+		.flags		= flags,
+		.nr		= trans->nr_updates,
+		.entries	= trans->updates,
+	};
+
+	if (!trans->nr_updates)
+		return 0;
+
+	trans->nr_updates = 0;
+
+	return __bch2_btree_insert_at(&insert);
+}
+
 int bch2_btree_delete_at(struct btree_iter *iter, unsigned flags)
 {
 	struct bkey_i k;
@@ -496,7 +802,7 @@ int bch2_btree_delete_at(struct btree_iter *iter, unsigned flags)
 	bkey_init(&k.k);
 	k.k.p = iter->pos;
 
-	return bch2_btree_insert_at(iter->c, NULL, NULL, NULL,
+	return bch2_btree_insert_at(iter->c, NULL, NULL,
 				    BTREE_INSERT_NOFAIL|
 				    BTREE_INSERT_USE_RESERVE|flags,
 				    BTREE_INSERT_ENTRY(iter, &k));
@@ -505,7 +811,6 @@ int bch2_btree_delete_at(struct btree_iter *iter, unsigned flags)
 int bch2_btree_insert_list_at(struct btree_iter *iter,
 			     struct keylist *keys,
 			     struct disk_reservation *disk_res,
-			     struct extent_insert_hook *hook,
 			     u64 *journal_seq, unsigned flags)
 {
 	BUG_ON(flags & BTREE_INSERT_ATOMIC);
@@ -513,7 +818,7 @@ int bch2_btree_insert_list_at(struct btree_iter *iter,
 	bch2_verify_keylist_sorted(keys);
 
 	while (!bch2_keylist_empty(keys)) {
-		int ret = bch2_btree_insert_at(iter->c, disk_res, hook,
+		int ret = bch2_btree_insert_at(iter->c, disk_res,
 				journal_seq, flags,
 				BTREE_INSERT_ENTRY(iter, bch2_keylist_front(keys)));
 		if (ret)
@@ -535,7 +840,6 @@ int bch2_btree_insert_list_at(struct btree_iter *iter,
 int bch2_btree_insert(struct bch_fs *c, enum btree_id id,
 		     struct bkey_i *k,
 		     struct disk_reservation *disk_res,
-		     struct extent_insert_hook *hook,
 		     u64 *journal_seq, int flags)
 {
 	struct btree_iter iter;
@@ -543,8 +847,8 @@ int bch2_btree_insert(struct bch_fs *c, enum btree_id id,
 
 	bch2_btree_iter_init(&iter, c, id, bkey_start_pos(&k->k),
 			     BTREE_ITER_INTENT);
-	ret = bch2_btree_insert_at(c, disk_res, hook, journal_seq, flags,
-				  BTREE_INSERT_ENTRY(&iter, k));
+	ret = bch2_btree_insert_at(c, disk_res, journal_seq, flags,
+				   BTREE_INSERT_ENTRY(&iter, k));
 	bch2_btree_iter_unlock(&iter);
 
 	return ret;
@@ -556,12 +860,8 @@ int bch2_btree_insert(struct bch_fs *c, enum btree_id id,
  * Range is a half open interval - [start, end)
  */
 int bch2_btree_delete_range(struct bch_fs *c, enum btree_id id,
-			   struct bpos start,
-			   struct bpos end,
-			   struct bversion version,
-			   struct disk_reservation *disk_res,
-			   struct extent_insert_hook *hook,
-			   u64 *journal_seq)
+			    struct bpos start, struct bpos end,
+			    u64 *journal_seq)
 {
 	struct btree_iter iter;
 	struct bkey_s_c k;
@@ -571,14 +871,12 @@ int bch2_btree_delete_range(struct bch_fs *c, enum btree_id id,
 			     BTREE_ITER_INTENT);
 
 	while ((k = bch2_btree_iter_peek(&iter)).k &&
-	       !(ret = btree_iter_err(k))) {
+	       !(ret = btree_iter_err(k)) &&
+	       bkey_cmp(iter.pos, end) < 0) {
 		unsigned max_sectors = KEY_SIZE_MAX & (~0 << c->block_bits);
 		/* really shouldn't be using a bare, unpadded bkey_i */
 		struct bkey_i delete;
 
-		if (bkey_cmp(iter.pos, end) >= 0)
-			break;
-
 		bkey_init(&delete.k);
 
 		/*
@@ -592,26 +890,16 @@ int bch2_btree_delete_range(struct bch_fs *c, enum btree_id id,
 		 * bkey_start_pos(k.k)).
 		 */
 		delete.k.p = iter.pos;
-		delete.k.version = version;
 
 		if (iter.flags & BTREE_ITER_IS_EXTENTS) {
-			/*
-			 * The extents btree is special - KEY_TYPE_DISCARD is
-			 * used for deletions, not KEY_TYPE_DELETED. This is an
-			 * internal implementation detail that probably
-			 * shouldn't be exposed (internally, KEY_TYPE_DELETED is
-			 * used as a proxy for k->size == 0):
-			 */
-			delete.k.type = KEY_TYPE_DISCARD;
-
 			/* create the biggest key we can */
 			bch2_key_resize(&delete.k, max_sectors);
 			bch2_cut_back(end, &delete.k);
 		}
 
-		ret = bch2_btree_insert_at(c, disk_res, hook, journal_seq,
-					  BTREE_INSERT_NOFAIL,
-					  BTREE_INSERT_ENTRY(&iter, &delete));
+		ret = bch2_btree_insert_at(c, NULL, journal_seq,
+					   BTREE_INSERT_NOFAIL,
+					   BTREE_INSERT_ENTRY(&iter, &delete));
 		if (ret)
 			break;