Hi, John
On 2025/3/6 0:15, John Garry wrote:
On 05/03/2025 08:16, yangxingui wrote:
Sure, something like that - you just need to get libsas to trigger
the proper hw port id assignment for the device. As for specific
implementation in the LLDD, that up to you guys.
Thank you for your suggestions. The disk was finally restored to
normal, but the error handling took a long time. Since the error
handling would set the host to the recovery state, other disks on the
same host would be blocked for a long time.
But that because you have IO inflight for the disk which was unplugged?
No, these are all ATA_CMD_ID_ATA commands newly issued by
ata_eh_revalidate_and_attach() during error handling, and will be
retried three times. The reason is that after the hw port id is changed,
we set it to invalid, and these ATA_CMD_ID_ATA cmds will fail to execute.
Thanks,
Xingui
