Missing records in the Discovery Descriptor List

[tbzatek]

Tested with #33 applied on top of the timberland_upstream-dev-full branch.

Two HFIs:

• 192.168.122.158
• 192.168.123.158

Defined two boot attempts through HII:

• HFI 1, 192.168.122.1:4420, nqn.2014-08.org.nvmexpress.discovery
• HFI 2, 192.168.123.1:4420, nqn.2014-08.org.nvmexpress.discovery

Each port on the target exports four namespaces. EDK2 performs two discovery attempts as expected and generates NBFT with 8 SSNS records (four namespaces * 2 HFIs). So far so good.

However, examining the NBFT table, only one Discovery record is present:

    "discovery":[
      {
        "index":1,
        "hfi":1,
        "uri":"nvme+tcp://192.168.122.1:4420/",
        "nqn":"nqn.2014-08.org.nvmexpress.discovery"
      }
    ]

Futhermore, the first four SSNS records properly reference this Disovery Descriptor, the other four records do not. The SSNS Discovered Namespace Flag just follows the buggy behaviour, e.g.:

subsystem_ns_list[0]->index=1
subsystem_ns_list[0]->discovery->index=1
subsystem_ns_list[0]->num_hfis=1
subsystem_ns_list[0]->hfis[0]->index=1
subsystem_ns_list[0]->transport=tcp
subsystem_ns_list[0]->traddr=192.168.122.1
subsystem_ns_list[0]->trsvcid=4420
subsystem_ns_list[0]->subsys_port_id=0
subsystem_ns_list[0]->nsid=123
subsystem_ns_list[0]->nid_type=3
subsystem_ns_list[0]->nid=d592cdf3ffffffff84123fcf7d99df27
subsystem_ns_list[0]->subsys_nqn=nqn.2014-08.org.nvmexpress.boot.poc:test-target
subsystem_ns_list[0]->pdu_header_digest_required=0
subsystem_ns_list[0]->data_digest_required=0
subsystem_ns_list[0]->controller_id=2
subsystem_ns_list[0]->asqsz=0
subsystem_ns_list[0]->dhcp_root_path_string=(null)
subsystem_ns_list[0]->discovered=1
subsystem_ns_list[0]->unavailable=0

subsystem_ns_list[4]->index=5
subsystem_ns_list[4]->num_hfis=1
subsystem_ns_list[4]->hfis[0]->index=2
subsystem_ns_list[4]->transport=tcp
subsystem_ns_list[4]->traddr=192.168.123.1
subsystem_ns_list[4]->trsvcid=4420
subsystem_ns_list[4]->subsys_port_id=0
subsystem_ns_list[4]->nsid=123
subsystem_ns_list[4]->nid_type=3
subsystem_ns_list[4]->nid=d592cdf3ffffffff84123fcf7d99df27
subsystem_ns_list[4]->subsys_nqn=nqn.2014-08.org.nvmexpress.boot.poc:test-target
subsystem_ns_list[4]->pdu_header_digest_required=0
subsystem_ns_list[4]->data_digest_required=0
subsystem_ns_list[4]->controller_id=3
subsystem_ns_list[4]->asqsz=0
subsystem_ns_list[4]->dhcp_root_path_string=(null)
subsystem_ns_list[4]->discovered=0
subsystem_ns_list[4]->unavailable=0

# nvme nbft show -o json
# libnvme/.build/test/nbft/nbft-dump

[tbzatek]

Some more tests (failure cases):

1. wrong IP address set for the first boot attempt:

• SSNS records from the discovery of the second boot attempt are present as expected
• the Discovery Descriptor List is empty

2. wrong IP address set for the second boot attempt:

• SSNS records from the discovery of the first boot attempt are present as expected
• fifth SSNS record present, for the second HFI, having a wrong subsys_nqn and controller_id (looks like these data were reused from the first HFI):

      {
        "index":5,
        "num_hfis":1,
        "hfis":[
          2
        ],
        "transport":"tcp",
        "traddr":"192.168.123.12",
        "trsvcid":"4420",
        "subsys_port_id":0,
        "nsid":0,
        "nid":"",
        "subsys_nqn":"nqn.2014-08.org.nvmexpress.boot.poc:test-target",
        "controller_id":17,
        "asqsz":0,
        "pdu_header_digest_required":0,
        "data_digest_required":0
      }

nvme_nbft_show_test2.json
libnvme_nbft-dump_test2.txt

• corresponding Discovery record is present for the first boot attempt, the second is missng

[tbzatek]

Having correct addresses but tearing the NIC link down produces exactly the same results as wrong IP addresses with link up.