[PATCH 0/2] btrfs: fix mount/remount failure due to past device flush errors

[fdmanana@kernel.org]

From: Filipe Manana <fdmanana@suse.com>

We currently fail mounting and remounting a fs from RO to RW if we had a
device flush error from a past mount. This causes mount/remount when
such errors were transient and don't happen anymore. One example is during
tests that explicitly trigger IO failure, such as the new generic/648 or,
much less frequently, generic/019 and generic/475. The first patch has
more details in its changelog.

V2: Rework to use instead an additional argument to btrfs_check_rw_degradable()
    and solve the same problem for remounts from RO to RW mode. Also added the
    second patch.

Filipe Manana (2):
  btrfs: fix mount failure due to past and transient device flush error
  btrfs: remove the failing device argument from btrfs_check_rw_degradable()

 fs/btrfs/disk-io.c |  4 ++--
 fs/btrfs/super.c   |  2 +-
 fs/btrfs/volumes.c | 32 +++++++++++++++++++++-----------
 fs/btrfs/volumes.h |  2 +-
 4 files changed, 25 insertions(+), 15 deletions(-)

-- 
2.33.0