Btrfs: reduce extent_state lock contention for metadata
authorChris Mason <chris.mason@oracle.com>
Mon, 25 Jul 2011 10:50:50 +0000 (06:50 -0400)
committerChris Mason <chris.mason@oracle.com>
Wed, 27 Jul 2011 16:46:47 +0000 (12:46 -0400)
commit19b6caf4acbf065dc96b47741d99f1b87243c468
tree4bce889648c2459e1dbd9fd835a196db26c06f9b
parent31533fb263928c93a34cda41b66a6e83ade5c766
Btrfs: reduce extent_state lock contention for metadata

For metadata buffers that don't straddle pages (all of them), btrfs
can safely use the page uptodate bits and extent_buffer uptodate bit
instead of needing to use the extent_state tree.

This greatly reduces contention on the state tree lock.

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