Hi Zhikang, We dropped vfs caches periodically to reproduce the kernel panic using drop cache command. I mean you have to trigger a checkpoint right after f2fs_mark_inode_dirty_sync() for a new inode. We don't have any special test cases for that and we just triggered to drop caches periodically. It was not easy to reproduce that, but it always occurrs within 24 hours.
Thanks, --------- Original Message --------- Sender : zhangzhikang <zhangzhika...@huawei.com> Date : 2018-01-12 16:11 (GMT+9) Title : 答复: [f2fs-dev] [PATCH] f2fs: prevent newly created inode from being dirtied incorrectly Hi Daeho: We had tried to msleep(10000) after f2fs_mark_inode_dirty_sync() in creating a new file, and then write checkpoint in another thread. But it didn't cause a kernel panic. So can you tell me what test case did you use, and provide the call trace? Thank you! Best regards, Zhikang Zhang
