f2fs: optimize fs_lock for better performance
authorYu Chao <chao2.yu@samsung.com>
Thu, 12 Sep 2013 03:17:51 +0000 (11:17 +0800)
committerJaegeuk Kim <jaegeuk.kim@samsung.com>
Tue, 24 Sep 2013 08:45:48 +0000 (17:45 +0900)
commite76eebee70948060d80b5f9edd2d97062049df53
tree0e553ac012ec66bbd50cc1ab0f3165c307b1b0e0
parent4a10c2ac2f368583138b774ca41fac4207911983
f2fs: optimize fs_lock for better performance

There is a performance problem: when all sbi->fs_lock are holded, then
all the following threads may get the same next_lock value from sbi->next_lock_num
in function mutex_lock_op, and wait for the same lock(fs_lock[next_lock]),
it may cause performance reduce.
So we move the sbi->next_lock_num++ before getting lock, this will average the
following threads if all sbi->fs_lock are holded.

v1-->v2:
Drop the needless spin_lock as Jaegeuk suggested.

Suggested-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
Signed-off-by: Yu Chao <chao2.yu@samsung.com>
Signed-off-by: Gu Zheng <guz.fnst@cn.fujitsu.com>
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
fs/f2fs/f2fs.h