From: Kent Overstreet <kent.overstreet@gmail.com>
Date: Wed, 16 Oct 2019 17:48:12 +0000 (-0400)
Subject: bcachefs: Fix an iterator counting bug
X-Git-Tag: io_uring-6.7-2023-11-10~119^2~2290
X-Git-Url: https://git.kernel.dk/?a=commitdiff_plain;h=71603f1ffe03d2409d591e4647377a415376a6b9;p=linux-block.git

bcachefs: Fix an iterator counting bug

The iterator counting assumed we're doing an obvious optimization when
only updating the refcount on indirect extents - but we're not doing it
yet.

Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
---

diff --git a/fs/bcachefs/extents.c b/fs/bcachefs/extents.c
index 8b7d2b8759b0..16a328a20fb5 100644
--- a/fs/bcachefs/extents.c
+++ b/fs/bcachefs/extents.c
@@ -971,20 +971,7 @@ static int count_iters_for_insert(struct btree_trans *trans,
 				     POS(0, idx + sectors)) >= 0)
 				break;
 
-			*nr_iters += 1;
-
-			if (overwrite &&
-			    r_k.k->type == KEY_TYPE_reflink_v) {
-				struct bkey_s_c_reflink_v r = bkey_s_c_to_reflink_v(r_k);
-
-				if (le64_to_cpu(r.v->refcount) == 1)
-					*nr_iters += bch2_bkey_nr_alloc_ptrs(r_k);
-			}
-
-			/*
-			 * if we're going to be deleting an entry from
-			 * the reflink btree, need more iters...
-			 */
+			*nr_iters += 1 + bch2_bkey_nr_alloc_ptrs(r_k);
 
 			if (*nr_iters >= max_iters) {
 				struct bpos pos = bkey_start_pos(k.k);