f2fs: avoid race in between atomic_read & atomic_inc
Previously, we will miss merging flush command during fsync due to below
race condition:
Thread A Thread B Thread C
- f2fs_issue_flush
- atomic_read(&issing_flush)
- f2fs_issue_flush
- atomic_read(&issing_flush)
- f2fs_issue_flush
- atomic_read(&issing_flush)
- atomic_inc(&issing_flush)
- atomic_inc(&issing_flush)
- atomic_inc(&issing_flush)
- submit_flush_wait
- submit_flush_wait
- submit_flush_wait
It needs to use atomic_inc_return instead to avoid such race.
Signed-off-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c
index 1b688d09..7ac4e6b 100644
--- a/fs/f2fs/segment.c
+++ b/fs/f2fs/segment.c
@@ -536,8 +536,7 @@
return ret;
}
- if (!atomic_read(&fcc->issing_flush)) {
- atomic_inc(&fcc->issing_flush);
+ if (atomic_inc_return(&fcc->issing_flush) == 1) {
ret = submit_flush_wait(sbi);
atomic_dec(&fcc->issing_flush);
@@ -547,7 +546,6 @@
init_completion(&cmd.wait);
- atomic_inc(&fcc->issing_flush);
llist_add(&cmd.llnode, &fcc->issue_list);
/* update issue_list before we wake up issue_flush thread */