Hi Song, In v4 following changes were made: - raid1_error and raid10_error description reworked, suggested by Guoqing. - Error messages for raid0 and linear reworked, suggested by Guoqing. - check for sync_request replaced by level checks, suggested by Guoqing. I did manual (mainly on IMSM) tests, here my next TODOs in mdadm: - broken state handling for redundant arrays in mdadm (it is exposed in sysfs). Currently it is working same as before, because broken is not a case for redundant arrays in mdadm. It requires to deal with already defined "FAILED" state in mdadm. - Blocking manual removal of devices (#mdadm --set-faulty). I run following native mdadm tests with and without changes, all passed: #./test --disks=/dev/nullb[1-5] --tests=00raid1,00raid4,00raid5,00raid6,01r1fail, 01r5fail,01replace,02r1add,05r1failfast,05r1re-add,05r1re-add-nosuper Mariusz Tkaczyk (3): raid0, linear, md: add error_handlers for raid0 and linear md: Set MD_BROKEN for RAID1 and RAID10 raid5: introduce MD_BROKEN drivers/md/md-linear.c | 14 +++++++- drivers/md/md.c | 30 +++++++++++------- drivers/md/md.h | 72 ++++++++++++++++++++++-------------------- drivers/md/raid0.c | 14 +++++++- drivers/md/raid1.c | 43 +++++++++++++++---------- drivers/md/raid10.c | 40 +++++++++++++---------- drivers/md/raid5.c | 48 ++++++++++++++-------------- 7 files changed, 155 insertions(+), 106 deletions(-) -- 2.26.2