From: Kent Overstreet <kent.overstreet@gmail.com>
Date: Sat, 9 Apr 2022 03:54:14 +0000 (-0400)
Subject: bcachefs: Don't skip triggers in fcollapse()
X-Git-Tag: v6.7-rc1~201^2~994
X-Git-Url: http://git.ipfire.org/?a=commitdiff_plain;h=7c4ca54ae68c4ae24dbfb8b209657a5249a5f0b7;p=thirdparty%2Fkernel%2Flinux.git

bcachefs: Don't skip triggers in fcollapse()

With backpointers this doesn't work anymore - backpointers always need
to be updated to point to the new extent position.

Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com>
---

diff --git a/fs/bcachefs/fs-io.c b/fs/bcachefs/fs-io.c
index 14550ac610c65..85a4484bec658 100644
--- a/fs/bcachefs/fs-io.c
+++ b/fs/bcachefs/fs-io.c
@@ -2894,13 +2894,7 @@ reassemble:
 
 		next_pos = insert ? bkey_start_pos(&delete.k) : delete.k.p;
 
-		if (copy.k->k.size == k.k->size) {
-			/*
-			 * If we're moving the entire extent, we can skip
-			 * running triggers:
-			 */
-			trigger_flags |= BTREE_TRIGGER_NORUN;
-		} else {
+		if (copy.k->k.size != k.k->size) {
 			/* We might end up splitting compressed extents: */
 			unsigned nr_ptrs =
 				bch2_bkey_nr_ptrs_allocated(bkey_i_to_s_c(copy.k));