The work flow of revalidation now is scanning expander phy by the sequence of the phy and check if the phy have changed. This will leads to some issues of swapping disks or replacing a disk with a new one. This patchset addresses the issues above by these main changes: 1. Let the revalidation first scan all phys, mark all changed phys, then revalidate in two steps. First check if we need to unregister some devices. if we need to unregister some devices, tell the upper revalidation that we need to revalidate again. Second, if no devices need to be unregistered, discover new devices. 2. For sata disk, checking the ata devices class and id to ensure the same device after flutter. v1->v2: 1. Do not raise a new bcast but use a loop in sas_revalidate_domain() to retry revalidation if we need to revalidate again. So that we can do the bcast event processing synchronised to the original event 2. Drop some patches splitted from this patchset: https://lkml.org/lkml/2018/9/25/153 3. Drop the SATA PHY connection rate matching patch since John had a better solution for the device discovery phase. https://lkml.org/lkml/2019/1/4/340 4. Re-init negotiated_linkrate when PHY is down. 5. Fix an issue when event in queue reached the limit. Jason Yan (7): scsi: libsas: reset the negotiated_linkrate when phy is down scsi: libsas: only clear phy->in_shutdown after shutdown event done scsi: libsas: optimize the debug print of the revalidate process scsi: libsas: split the replacement of sas disks in two steps scsi: libsas: check if the same device when flutter scsi: libsas: reset the phy address if discover failed scsi: libsas: fix issue of swapping two sas disks drivers/ata/libata-core.c | 3 +- drivers/scsi/libsas/sas_ata.c | 18 +++ drivers/scsi/libsas/sas_discover.c | 25 ++-- drivers/scsi/libsas/sas_expander.c | 239 +++++++++++++++++++++++++++---------- drivers/scsi/libsas/sas_phy.c | 3 +- include/linux/libata.h | 2 + include/scsi/libsas.h | 6 +- 7 files changed, 224 insertions(+), 72 deletions(-) -- 2.14.4
