After we have packed all refs, we prune any loose refs that
correspond to what we packed. We do so by first taking a
lock with lock_ref_sha1, and then deleting the loose ref

However, lock_ref_sha1 will refuse to take a lock on any
refs that exist at the top-level of the "refs/" directory,
and we skip pruning the ref.  This is almost certainly not
what we want to happen here. The criteria to be pruned
should not differ from that to be packed; if a ref makes it
to prune_ref, it's because we want it both packed and
pruned (if there are refs you do not want to be packed, they
should be omitted much earlier by pack_ref_is_possible,
which we do in this case if --all is not given).

We can fix this by switching to lock_any_ref_for_update.
This behaves exactly the same with the exception of this
top-level check.

Signed-off-by: Jeff King <>
 refs.c               | 3 ++-
 t/ | 7 +++++++
 2 files changed, 9 insertions(+), 1 deletion(-)

diff --git a/refs.c b/refs.c
index 27927f2..82e5b1b 100644
--- a/refs.c
+++ b/refs.c
@@ -2387,7 +2387,8 @@ static void try_remove_empty_parents(char *name)
 /* make sure nobody touched the ref, and unlink */
 static void prune_ref(struct ref_to_prune *r)
-       struct ref_lock *lock = lock_ref_sha1(r->name + 5, r->sha1);
+       struct ref_lock *lock = lock_any_ref_for_update(r->name, r->sha1,
+                                                       0, NULL);
        if (lock) {
                unlink_or_warn(git_path("%s", r->name));
diff --git a/t/ b/t/
index 1a2080e..3a017bf 100755
--- a/t/
+++ b/t/
@@ -151,4 +151,11 @@ test_expect_success 'delete ref while another dangling 
packed ref' '
        test_cmp /dev/null result
+test_expect_success 'pack ref directly below refs/' '
+       git update-ref refs/top HEAD &&
+       git pack-refs --all --prune &&
+       grep refs/top .git/packed-refs &&
+       test_path_is_missing .git/refs/top

To unsubscribe from this list: send the line "unsubscribe git" in
the body of a message to
More majordomo info at

Reply via email to