[PATCH RFC 0/2] btrfs: make read repair work in synchronous mode

[Qu Wenruo <wqu@suse.com>]

The first patch is just a preparation for the 2nd patch, which is also
the core.

It will make repair_one_sector() to wait for the read from other copies
finish, before returning.

This will make the code easier to read, but huge drop in concurrency and
performance for read-repair.

My only justification is read-repair should be a cold path, and we may
be able to afford the change.

Qu Wenruo (2):
  btrfs: introduce a pure data checksum checking helper
  btrfs: do data read repair in synchronous mode

 fs/btrfs/ctree.h     |   2 +
 fs/btrfs/extent_io.c | 111 ++++++++++++++++++++++++++++++-------------
 fs/btrfs/extent_io.h |   3 +-
 fs/btrfs/inode.c     |  73 ++++++++++++++--------------
 4 files changed, 120 insertions(+), 69 deletions(-)

-- 
2.35.1