f2fs: let sync node IO interrupt async one
authorChao Yu <yuchao0@huawei.com>
Mon, 4 Jun 2018 15:20:36 +0000 (23:20 +0800)
committerJaegeuk Kim <jaegeuk@kernel.org>
Mon, 9 Jul 2018 00:40:58 +0000 (17:40 -0700)
commit57344fb5931e6651021d1e80e1117dd0f5a68734
tree783ff3e514b00641b4d8deb41e2f8551de1b7841
parent6cf9490912e15a3aeb8827f177c54b64d36c82ee
f2fs: let sync node IO interrupt async one

Although mixed sync/async IOs can have continuous LBA, as they have
different IO priority, block IO scheduler will add them into different
queues and commit them separately, result in splited IOs which causes
wrose performance.

This patch gives high priority to synchronous IO of nodes, means that
once synchronous flow starts, it can interrupt asynchronous writeback
flow of system flusher, so more big IOs can be expected.

Signed-off-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
fs/f2fs/checkpoint.c
fs/f2fs/data.c
fs/f2fs/f2fs.h
fs/f2fs/file.c
fs/f2fs/gc.c
fs/f2fs/node.c
fs/f2fs/super.c