Btrfs: wake up extent state waiters on unlock through clear_extent_bits
authorFilipe Manana <fdmanana@suse.com>
Thu, 14 May 2015 19:41:07 +0000 (20:41 +0100)
committerChris Mason <clm@fb.com>
Wed, 3 Jun 2015 11:02:56 +0000 (04:02 -0700)
commit0f31871f4411b5c0d42fb4403dec83a21a96100b
treeed012f7057b95cd5b6dcea79ad99b49b14aad6b4
parentc152b63efc94d5fd486a39e9df5e61ae77e08e44
Btrfs: wake up extent state waiters on unlock through clear_extent_bits

When we clear an extent state's EXTENT_LOCKED bit with clear_extent_bits()
through free_io_failure(), we weren't waking up any tasks waiting for the
extent's state EXTENT_LOCKED bit, leading to an hang.

So make sure clear_extent_bits() ends up waking up any waiters if the
bit EXTENT_LOCKED is supplied by its callers.

Zygo Blaxell was experiencing such hangs at inode eviction time after
file unlinks. Thanks to him for a set of scripts to reproduce the issue.

Reported-by: Zygo Blaxell <ce3g8jdj@umail.furryterror.org>
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Reviewed-by: Liu Bo <bo.li.liu@oracle.com>
Signed-off-by: Chris Mason <clm@fb.com>
fs/btrfs/extent_io.c