From 776eaddb2cef366b16dac8857899dbb4cc5dfdb1 Mon Sep 17 00:00:00 2001 From: Kent Overstreet Date: Tue, 12 Oct 2021 14:25:13 -0400 Subject: [PATCH] bcachefs: Fix deletion in __bch2_dev_usrdata_drop() With snapshots, __bch2_dev_usr_data_drop() now uses an ALL_SNAPSHOTS iterator, which isn't an extent iterator - meaning we shouldn't be inserting whiteouts with nonzero size to delete. This fixes a bug where we go RO because we tried to insert an invalid key in the device remove path. Signed-off-by: Kent Overstreet --- fs/bcachefs/migrate.c | 9 +++++++++ 1 file changed, 9 insertions(+) diff --git a/fs/bcachefs/migrate.c b/fs/bcachefs/migrate.c index 9f9eb799337e2..94d5d99ffd2a4 100644 --- a/fs/bcachefs/migrate.c +++ b/fs/bcachefs/migrate.c @@ -73,6 +73,15 @@ static int __bch2_dev_usrdata_drop(struct bch_fs *c, unsigned dev_idx, int flags */ bch2_extent_normalize(c, bkey_i_to_s(sk.k)); + /* + * Since we're not inserting through an extent iterator + * (BTREE_ITER_ALL_SNAPSHOTS iterators aren't extent iterators), + * we aren't using the extent overwrite path to delete, we're + * just using the normal key deletion path: + */ + if (bkey_deleted(&sk.k->k)) + sk.k->k.size = 0; + ret = bch2_btree_iter_traverse(&iter) ?: bch2_trans_update(&trans, &iter, sk.k, BTREE_UPDATE_INTERNAL_SNAPSHOT_NODE) ?: -- 2.30.2