>>>>> "nab" == Nicholas A Bellinger <nab@xxxxxxxxxxxxxxx> writes: >> The alternative is to avoid this patch but then we might miss DIF >> support when working against legacy initiators. nab> Mmmm, not sure about this one.. nab> Considering that it's only the first two READ_10s that this effects nab> before normal SCSI_PROT_*_PASS operation kicks in, I'm not sure if nab> it's worth it to add these to tcm_loop.. Note that virtio-scsi + nab> vhost-scsi would need similar bits as well.. The fact that initiator and target are capable of handling protection information does not in any way guarantee that all reads and writes will have PI included. We need to be able to turn off checking for things like recovery and RAID. So in my book it is crucial that both initiator and target do exactly what they are told by the ULD. This means the initiator should only do DIX if the prot_op tells it to. And on the target end you should only do DIF transfers and checks if *PROTECT is set in the CDB. Some HBAs support a mode in which they snoop (or silently issue) INQUIRY/READ CAPACITY(16). And they will turn on protected transfers behind the OS' back under the assumption that the OS is not aware of DIF. Let's not go there! -- Martin K. Petersen Oracle Linux Engineering -- To unsubscribe from this list: send the line "unsubscribe target-devel" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
