btrfs: fix iteration bug in __qgroup_excl_accounting()

__qgroup_excl_accounting() uses the qgroup iterator machinery to
update the account of one qgroups usage for all its parent hierarchy,
when we either add or remove a relation and have only exclusive usage.

However, there is a small bug there: we loop with an extra iteration
temporary qgroup called `cur` but never actually refer to that in the
body of the loop. As a result, we redundantly account the same usage to
the first qgroup in the list.

This can be reproduced in the following way:

  mkfs.btrfs -f -O squota <dev>
  mount <dev> <mnt>
  btrfs subvol create <mnt>/sv
  dd if=/dev/zero of=<mnt>/sv/f bs=1M count=1
  sync
  btrfs qgroup create 1/100 <mnt>
  btrfs qgroup create 2/200 <mnt>
  btrfs qgroup assign 1/100 2/200 <mnt>
  btrfs qgroup assign 0/256 1/100 <mnt>
  btrfs qgroup show <mnt>

and the broken result is (note the 2MiB on 1/100 and 0Mib on 2/100):

  Qgroupid    Referenced    Exclusive   Path
  --------    ----------    ---------   ----
  0/5           16.00KiB     16.00KiB   <toplevel>
  0/256          1.02MiB      1.02MiB   sv

  Qgroupid    Referenced    Exclusive   Path
  --------    ----------    ---------   ----
  0/5           16.00KiB     16.00KiB   <toplevel>
  0/256          1.02MiB      1.02MiB   sv
  1/100          2.03MiB      2.03MiB   2/100<1 member qgroup>
  2/100            0.00B        0.00B   <0 member qgroups>

With this fix, which simply re-uses `qgroup` as the iteration variable,
we see the expected result:

  Qgroupid    Referenced    Exclusive   Path
  --------    ----------    ---------   ----
  0/5           16.00KiB     16.00KiB   <toplevel>
  0/256          1.02MiB      1.02MiB   sv

  Qgroupid    Referenced    Exclusive   Path
  --------    ----------    ---------   ----
  0/5           16.00KiB     16.00KiB   <toplevel>
  0/256          1.02MiB      1.02MiB   sv
  1/100          1.02MiB      1.02MiB   2/100<1 member qgroup>
  2/100          1.02MiB      1.02MiB   <0 member qgroups>

The existing fstests did not exercise two layer inheritance so this bug
was missed. I intend to add that testing there, as well.

Fixes: a0bdc04b0732 ("btrfs: qgroup: use qgroup_iterator in __qgroup_excl_accounting()")
CC: stable@vger.kernel.org # 6.12+
Reviewed-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: Boris Burkov <boris@bur.io>
Signed-off-by: David Sterba <dsterba@suse.com>

diff --git a/fs/btrfs/qgroup.c b/fs/btrfs/qgroup.c
index 1a5972178b3a3c4e45d8fffb026db0192931edd3..ccaa9a3cf1ce377fbc0ddc65d44d29858aec69ee 100644 (file)
--- a/fs/btrfs/qgroup.c
+++ b/fs/btrfs/qgroup.c
@@ -1453,7 +1453,6 @@ static int __qgroup_excl_accounting(struct btrfs_fs_info *fs_info, u64 ref_root,
                                    struct btrfs_qgroup *src, int sign)
 {
        struct btrfs_qgroup *qgroup;
-       struct btrfs_qgroup *cur;
        LIST_HEAD(qgroup_list);
        u64 num_bytes = src->excl;
        int ret = 0;
@@ -1463,7 +1462,7 @@ static int __qgroup_excl_accounting(struct btrfs_fs_info *fs_info, u64 ref_root,
                goto out;
 
        qgroup_iterator_add(&qgroup_list, qgroup);
-       list_for_each_entry(cur, &qgroup_list, iterator) {
+       list_for_each_entry(qgroup, &qgroup_list, iterator) {
                struct btrfs_qgroup_list *glist;
 
                qgroup->rfer += sign * num_bytes;