On 12-10-09 03:04 PM, Nicholas A. Bellinger wrote: > > So we hard-code the prefix for the IEEE NAA Registered Extended Device > Identifier in target kernel code to follow what SCSI clients expect for > this specific type of EVPD=0x83 designator. Which is fair enough for real-world, production environments, indeed. > We do not allow target-mode > user-space code to set this value to a format containing a value that > SCSI clients do not expect. That's a pity. > Using a non world-wide unique value for the one value that initiators > depend on as being world-wide queue is a *bad* idea. Indeed, while I will agree with you on this in any sort of production environment, being able to configure and therefore have predictability of the scsi id that a client will see in a test lab situation has advantages. Thanks much for the response, b.
Attachment:
signature.asc
Description: OpenPGP digital signature
