On Tue, Jun 14, 2022 at 09:15:36PM +0800, Yu Kuai wrote: > > for making so long, so I guess there must be other delay added by the feature > > of BLK_MQ_F_TAG_HCTX_SHARED. > > Please see details in the reasons 2), scsi scan will call > __scsi_remove_device() a lot of times(each host, each channel, each > target). Well, if the SCSI core does that you are falling back to a sequentical scan instad of using REPORT LUNS. Which seems like a bug on its own that is worth investigation.
