f2fs: fix to wait all atomic written pages writeback

This patch fixes the incorrect range (0, LONG_MAX) which is used
in ranged fsync. If we use LONG_MAX as the parameter for indicating
the end of file we want to synchronize, in 32-bits architecture
machine, these datas after 4GB offset may not be persisted in
storage after ->fsync returned.

Here, we alter LONG_MAX to LLONG_MAX to fix this issue.
Signed-off-by: NChao Yu <chao2.yu@samsung.com>
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

fs/f2fs/file.c

@@ -1357,7 +1357,7 @@ static int f2fs_ioc_commit_atomic_write(struct file *filp)
 		commit_inmem_pages(inode, false);
 	}
 
-	ret = f2fs_sync_file(filp, 0, LONG_MAX, 0);
+	ret = f2fs_sync_file(filp, 0, LLONG_MAX, 0);
 	mnt_drop_write_file(filp);
 	return ret;
 }