Btrfs: fix unsafe usage of merge_state
authorXiao Guangrong <xiaoguangrong@cn.fujitsu.com>
Wed, 20 Apr 2011 06:45:49 +0000 (06:45 +0000)
committerChris Mason <chris.mason@oracle.com>
Mon, 23 May 2011 17:24:41 +0000 (13:24 -0400)
commitc7f895a2b2d1a002810d52e7b6653c9dc2fd0b0b
tree1b2a05c2efe1f169983bff7a497d812aa86c3000
parent8233767a227ac5843f1023b88c7272a7b5058f5f
Btrfs: fix unsafe usage of merge_state

merge_state can free the current state if it can be merged with the next node,
but in set_extent_bit(), after merge_state, we still use the current extent to
get the next node and cache it into cached_state

Signed-off-by: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
fs/btrfs/extent_io.c