On Tue, 2015-04-07 at 19:27 -0400, Martin K. Petersen wrote: > >>>>> "nab" == Nicholas A Bellinger <nab@xxxxxxxxxxxxx> writes: > > nab> This specifically is to allow LIO to perform WRITE_STRIP + > nab> READ_INSERT operations when functioning with non T10-PI enabled > nab> devices, seperate from any available hw offloads the fabric > nab> supports. > > How do you handle RDPROTECT/WRPROTECT values of 3 if the PI is not > persistent? > AFAICT, this would result in cmd->prot_op = TARGET_PROT_*_PASS and cmd->prot_checks = 0 for RDPROTECT/WRPROTECT == 0x3 in sbc_set_prot_op_checks() code. Do DOUT_STRIP + DIN_INSERT need to be called if a protection buffer is present when RDPROTECT/WRPROTECT == 0x3 if fabric_prot was cleared..? Or should the command be rejected when a protection buffer is present + RDPROTECT/WRPROTECT is non-zero if fabric_prot was cleared..? Also wrt to PI persistence across session restart, one way it can work is to store the last se_sess->sess_prot_type in se_node_acl, and reinstate the previous setting across session restart. This would allow new sessions to pickup the latest fabric_prot_type endpoint attribute value, but make existing ones with PI enabled keep their previous sess_prot_type settings. WDYT..? --nab -- To unsubscribe from this list: send the line "unsubscribe linux-scsi" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
