releases/4.19.31/btrfs-ensure-that-a-dup-or-raid1-block-group-has-exactly-two-stripes.patch

   1 From 349ae63f40638a28c6fce52e8447c2d14b84cc0c Mon Sep 17 00:00:00 2001
   2 From: Johannes Thumshirn <jthumshirn@suse.de>
   3 Date: Mon, 18 Feb 2019 11:28:37 +0100
   4 Subject: btrfs: ensure that a DUP or RAID1 block group has exactly two stripes
   5
   6 From: Johannes Thumshirn <jthumshirn@suse.de>
   7
   8 commit 349ae63f40638a28c6fce52e8447c2d14b84cc0c upstream.
   9
  10 We recently had a customer issue with a corrupted filesystem. When
  11 trying to mount this image btrfs panicked with a division by zero in
  12 calc_stripe_length().
  13
  14 The corrupt chunk had a 'num_stripes' value of 1. calc_stripe_length()
  15 takes this value and divides it by the number of copies the RAID profile
  16 is expected to have to calculate the amount of data stripes. As a DUP
  17 profile is expected to have 2 copies this division resulted in 1/2 = 0.
  18 Later then the 'data_stripes' variable is used as a divisor in the
  19 stripe length calculation which results in a division by 0 and thus a
  20 kernel panic.
  21
  22 When encountering a filesystem with a DUP block group and a
  23 'num_stripes' value unequal to 2, refuse mounting as the image is
  24 corrupted and will lead to unexpected behaviour.
  25
  26 Code inspection showed a RAID1 block group has the same issues.
  27
  28 Fixes: e06cd3dd7cea ("Btrfs: add validadtion checks for chunk loading")
  29 CC: stable@vger.kernel.org # 4.4+
  30 Reviewed-by: Qu Wenruo <wqu@suse.com>
  31 Reviewed-by: Nikolay Borisov <nborisov@suse.com>
  32 Signed-off-by: Johannes Thumshirn <jthumshirn@suse.de>
  33 Reviewed-by: David Sterba <dsterba@suse.com>
  34 Signed-off-by: David Sterba <dsterba@suse.com>
  35 Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
  36
  37 ---
  38  fs/btrfs/volumes.c |    4 ++--
  39  1 file changed, 2 insertions(+), 2 deletions(-)
  40
  41 --- a/fs/btrfs/volumes.c
  42 +++ b/fs/btrfs/volumes.c
  43 @@ -6425,10 +6425,10 @@ static int btrfs_check_chunk_valid(struc
  44         }
  45
  46         if ((type & BTRFS_BLOCK_GROUP_RAID10 && sub_stripes != 2) ||
  47 -           (type & BTRFS_BLOCK_GROUP_RAID1 && num_stripes < 1) ||
  48 +           (type & BTRFS_BLOCK_GROUP_RAID1 && num_stripes != 2) ||
  49             (type & BTRFS_BLOCK_GROUP_RAID5 && num_stripes < 2) ||
  50             (type & BTRFS_BLOCK_GROUP_RAID6 && num_stripes < 3) ||
  51 -           (type & BTRFS_BLOCK_GROUP_DUP && num_stripes > 2) ||
  52 +           (type & BTRFS_BLOCK_GROUP_DUP && num_stripes != 2) ||
  53             ((type & BTRFS_BLOCK_GROUP_PROFILE_MASK) == 0 &&
  54              num_stripes != 1)) {
  55                 btrfs_err(fs_info,