Btrfs: speed up snapshot dropping
commit1c4850e21df8b441164d910bc611ef46a01d5d75
authorYan, Zheng <zheng.yan@oracle.com>
Mon, 21 Sep 2009 19:55:59 +0000 (21 15:55 -0400)
committerChris Mason <chris.mason@oracle.com>
Mon, 21 Sep 2009 19:55:59 +0000 (21 15:55 -0400)
treeaeccbf3495421d1343bbe08cb824ac1ae6764e43
parentb917b7c3be50435fa8257591b964934e917f2d45
Btrfs: speed up snapshot dropping

This patch contains two changes to avoid unnecessary tree block reads during
snapshot dropping.

First, check tree block's reference count and flags before reading the tree
block. if reference count > 1 and there is no need to update backrefs, we can
avoid reading the tree block.

Second, save when snapshot was created in root_key.offset. we can compare block
pointer's generation with snapshot's creation generation during updating
backrefs. If a given block was created before snapshot was created, the
snapshot can't be the tree block's owner. So we can avoid reading the block.

Signed-off-by: Yan Zheng <zheng.yan@oracle.com>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
fs/btrfs/extent-tree.c
fs/btrfs/transaction.c