James Smart wrote:
I'll point out, but not delve into - that you've avoided the target id bindings again, thus forcing the lldd's/adapters to maintain tables of port#/SASaddr to target id (if they care, and I believe enterprise storage will care!). I can understand if you're going to eventually insert a generic abstraction layer for scsi's b/c/t/l addressing scheme, but since it's not here today, I expect you to hold SAS to the same "code sharing" level as FC.
b/c/t/l has no place in layers below scsi core (except in SPI low-level where it has actual meaning).
I don't understand exactly what should be accomplished by a table of WWNs to target IDs --- but any kind of mapping between WWN and device file (and thereby between WWN and a faked b/c/t/l) belongs into userspace, not into the kernel.
Furthermore, there cannot be an "abstraction layer for scsi's b/c/t/l addressing scheme". It is impossible to abstract something which is without meaning.
-- Stefan Richter -=====-=-=-= =--= --==- http://arcgraph.de/sr/ - : 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
