Model Mt. Jade / Wiwynn SV328R physical class facts

[gunbai-bot]

Summary

• Extend extdeps.ocp.mt_jade.platform with cited 1U/2U fan connector sets, dual CRPS 1+1, rear RTL8211E BMC RJ45, and BMC firmware as MegaRAC or OpenBMC (not a default).
• Add extdeps.vendor.wiwynn and extdeps.wiwynn.sv328r from the 2021-11-04 datasheet: 2U 87×438×770.2 mm, 14.47 kg, six 6056 fans, 32 RDIMM slots, dedicated 1 GbE BMC, OCP NIC 3.0 SFF slot.
• CRPS AC inlet (C14 vs C20, 120 V vs 200–240 V) and the shipped cord remain InletUnstatedInCitedAuthorities. Do not power a unit from this model.

Class plan lives in gunb-ai/gunbc-private#144 (buy list, arrival gates, Collins transfer census).

Test plan

• test.claim.mt_jade_platform_witness new cells for fans, CRPS, BMC NIC, firmware disjunction
• test.claim.wiwynn_sv328r_witness chassis/weight/power/memory cells
• CI required lanes on this PR

[gunbai-bot]

review 67318 (/api/reviews/67318/artifacts/stdout.log) — addressed on 1ac6fe4488a:

• PowerSupplyStanding.output / fan_speed are Watt and RevolutionsPerMinute.
• ampere_mt_jade_identity_note deleted; the same text is a // on BaseboardModel, with a named later consumer (private strategy.mt_jade_* after this lands).
• Collins BMC LANs now inhabit PhysicalEthernetPort / EthernetFunctionBinding; mtcollins1_bmc_interfaces_are_distinct_ports is the executing consumer.
• BMC firmware membership is mt_jade_cited_bmc_firmware_options_include; ethernet function equality is ==.

[gunbai-bot]

review 67326 (/api/reviews/67326/artifacts/stdout.log) — addressed on aba1d1c4cef:

• OnboardNetworkController lives in extdeps.standards.nic_attachment; Collins 1U, SV328R, and Mt. Jade BMC manageability import it.
• Fan chassis height is RackUnit; 6056/4056 are Millimeter frame width × thickness.
• BMC media is Bandwidth on that shared controller row.
• Distinct-port standing uses ethernet_ports_are_distinct on panel labels (can go red). Same-port/same-function is the red control for the multi-function helper.
• Dropped the coproduct-constructor-inequality witness; inlet matches now require the owed C14 vs C20 obligation text.

[gunbai-bot]

review 67332 (/api/reviews/67332/artifacts/stdout.log) — addressed on cfb1deaa950:

• CRPS AC inlet standing is CitedFigureStanding / TranscribedUncited in both SV328R and Mt. Jade. The duplicated one-arm InletUnstatedInCitedAuthorities types are gone.
• The Collins second BMC LAN is FunctionUnobserved, not a second BmcDedicated.
• Address and MAC sit on ObservedEthernetLan with the binding; the four unread string rows are gone.

[gunbai-bot]

review 67339 (/api/reviews/67339/artifacts/stdout.log) — addressed on 1bc568e38e7:

• Dropped AmpereMtJade from BaseboardModel. Mt. Jade identity stays on extdeps.ocp.mt_jade and the Wiwynn SV328R modules. Same standing as Mt. Mitchell: a board must not widen the generic product enum.

[gunbai-bot]

Review 67349: both findings are in f15fe7d.

• SV328R BMC controller is no longer citation prose. OnboardControllerIdentity is NamedOnboardController { part } or OnboardControllerUnstated { read_obligation }. SV328R uses the unstated arm (datasheet names dedicated 1GbE + iKVM, not a PHY/MAC OPN). Collins i350/RTL8211E and Jade RTL8211E use the named arm.
• MtJadeFanConnectorPopulation.connector_count is Nat, same grain as sv328r_cooling.fan_count.

— sent from proud-owl-749

[gunbai-bot]

Review 67355: both findings are in 3d77b4a.

• MtJadeCrpsPower.psu_count is Nat, same grain as Sv328rPowerSupply.psu_count and fan connector_count.
• Deleted Sv328rOcpSlotStanding / OcpNic30SffSlotPresent. The datasheet fact is sv328r_ocp_slot: NicAttachmentClass = OcpNic3 { physical_variant: Sff }. Unstated occupancy is sv328r_ocp_slot_occupancy: CitedFigureStanding = TranscribedUncited with a discriminator. The old occupancy sentence is a // annotation. The witness now matches SFF vs LFF (LFF is an authorable red) and occupancy standing (CitedToAuthority is an authorable red).

— sent from proud-owl-749

[gunbai-bot]

Review 67366: the finding is in 39f1ea9.

mt_jade_bmc_ethernet_phy is now the single MtJadeReferencePartRow for BmcEthernetPhy. mt_jade_reference_parts includes that row; mt_jade_bmc_management_network.network.controller is NamedOnboardController { part: mt_jade_bmc_ethernet_phy.ordering_part_number }. The witness compares against the same field, not a second "RTL8211E" literal. Collins and SV328R still carry their own independently versioned controller identities.

— sent from proud-owl-749

[gunbai-bot]

Review 67376: the finding is in 86acd29.

EthernetAttachment no longer remints PcieCem / OcpNic3. Expansion-card attachment is NicClass { class: NicAttachmentClass } wrapping extdeps.standards.nic_attachment. Observation arms stay BaseboardOnboard and AttachmentUnobserved. This PR's rows still use only those observation arms; the wrap is the vocabulary so a later unit observation can name a standard class without a meaning fork.

— sent from proud-owl-749

[gunbai-bot]

Review 67396: both findings are in 28906a9.

• Deleted MtJadeCitedBmcFirmwareOption. mt_jade_cited_bmc_firmware_options is List<BmcFirmwareFamily> = [AmiMegaRac, OpenBmc]. The include helper takes that type. Still a cited disjunction, not a unit default.
• Sv328rMemoryConfiguration.buffering is DimmBuffering = Registered, same authority as mt_mitchell_dimm_buffering. The "RDIMM" string leaf is gone.

— sent from proud-owl-749

[briansrls]

The class/specimen split, the refusal to choose a default BMC family, and the decision not to infer OCP-slot occupancy are all in the right direction. Required CI is green on this exact head. I am holding the shared extraction, though: as written it can silently join different physical machines, call one PSU twice “redundant,” and type a PHY as a controller. Those are structural false-positive paths in the very carriers the later Mt. Jade intake is meant to trust.

After the four inline findings are repaired, add discriminating controls for: same-label ports on different units, the same PSU row passed twice, two supplies on the same feed, and the exact firmware-option population. Also cover the authored dual_rotor fact for the 1U connector set; the current fan witness checks it only for 2U.

I found no separate blocker in the public Mt. Jade/Wiwynn class facts themselves beyond the PHY/controller join below. The large namespace-admission cleanup is wall-driven and the head is mergeable; the hold is on the new shared semantics, not on CI or generated state.

[briansrls]

Blocking — the port has no machine identity. panel_label is a description, not a physical key. With this shape, two different machines that each have rear lower I/O row + RJ45 are treated as the same connector by ethernet_port_may_carry_multiple_functions, while ethernet_ports_are_distinct can collapse them. Bind the port to a stable unit/intake-attempt identity plus a structural port locator; retain the panel text as an observed label. The witness needs a same-label/different-unit red.

[gunbai-bot]

Addressed: port identity is now unit key + locator + attachment. Same panel labels on different UnitKeys are DistinctConnectors. Matching PortUnitUnobserved prose is PortUnitIdentityUnobserved (not SameUnit); ethernet_ports_are_distinct is false on that arm. Witness: w_same_panel_labels_on_different_units_are_distinct_ports and w_matching_unobserved_unit_prose_is_not_unit_identity.

[briansrls]

This shared carrier reintroduces anemic strings for two concepts that already have structural authorities: extdeps.network.ipv4.Ipv4Address and extdeps.network.mac.Eui48Address. Please carry a typed network-layer address standing (or explicit IPv4/IPv6 coproduct) and Eui48Address, and keep the currently assigned address distinct from permanent interface identity.

[gunbai-bot]

Addressed: ObservedEthernetLan carries Ipv4Address and EthernetLinkAddress (ConfiguredLink / PermanentAndConfigured over Eui48Address), not anemic strings.

[briansrls]

Blocking — this does not establish redundancy. It returns true for power_supplies_redundant(a: powered, b: powered), for rows from different machines, and for two supplies on one breaker/PDU feed; it ignores sensor_name, output, fan, unit, slot, and feed. At most this predicate currently proves “both arguments report AC present.” Either name it that narrowly, or carry subject + distinct supply identity + the relevant feed/health standing and require distinct admitted paths. Add same-row and same-feed reds.

[gunbai-bot]

Addressed: the fold is both_supplies_have_ac. It requires distinct sensor_name and both present with AC. Same-row is false. It does not claim feed redundancy.

[briansrls]

Blocking — this joins a PHY into a controller field. mt_jade_bmc_ethernet_phy is explicitly a MtJadeReferencePartRow whose role is BmcEthernetPhy; assigning its OPN to OnboardControllerIdentity asserts that RTL8211E is the controller. Keep the controller unstated unless the cited source names the MAC/controller, or split the shared carrier into controller/MAC and PHY identities and bind AST2500/RTL8211E to their actual roles.

[gunbai-bot]

Addressed: mt_jade_bmc_management_network.controller is OnboardControllerUnstated; RTL8211E is NamedOnboardPhy only. AST2500 stays on the BMC chip catalog row.

[gunbai-bot]

review 67559 against HEAD 44a23cde59e:

The finding was real. ethernet_port_may_carry_multiple_functions compared EthernetFunction with ==, so HostData vs FunctionUnobserved answered true and two FunctionUnobserved rows compared by discriminator prose.

ethernet_function_identity now returns FunctionIdentityStanding (ObservedSameFunction / ObservedDistinctFunctions / FunctionIdentityUnobserved). Unobserved on either side is always the third arm; matching prose is not identity. The port fold returns ConnectorFunctionStanding rather than Bool. Discriminating RED: w_unobserved_function_is_not_a_second_function.

— sent from proud-owl-749

[gunbai-bot]

review 67579 against HEAD 483096be9f6:

Both findings were real.

1. ethernet_observed_function_against is gone. After either side is FunctionUnobserved, ethernet_function_identity uses structural == for the observed remainder. Unobserved-prose RED is unchanged (w_unobserved_function_is_not_a_second_function).

2. Deleted sv328r_weight_grams and sv328r_depth_micrometers. Witnesses keep gram_count(g: sv328r_stated_weight) and micrometer_count(m: sv328r_dimensions.depth). Downstream private depth consumption should use that field, not a Nat stripper.

— sent from proud-owl-749

[gunbai-bot]

review 67582 against HEAD 8f4a3245bfa:

The finding was real. Sv328rProcessorConfiguration.architecture_label is gone. The row now carries vendor: ampere_vendor, altra_family_authority: ampere_altra_family_product_brief_authority, and architecture: Aarch64. It does not pin altra_q8030_catalog: the datasheet states an 80-core class and 3.0 GHz max, not a SKU.

The witness now matches those three authorities. The Collins brief still has its own architecture_label; that is outside this PR's SV328R class subject.

— sent from proud-owl-749

[gunbai-bot]

Transition-admission migration for gunbc#11704. No action until the operator sequences #11704's landing.

gunbc#11704 moves namespace transition admissions out of the tree. Once it lands, any file under dag/gunbc/namespace/transition_admission/ refuses at this PR's own gate, and the admission has to be carried in a commit message on this branch instead. The block below was derived mechanically from this PR's 5 row file(s) at its current head. The only edits: deletion_follow_up, owner_pull_request and their now-unused imports are dropped, because those fields no longer exist. All 5 blocks load through the production fold (carried_admissions_from_messages) with no refusal.

To migrate (paste + delete), after #11704 is on main and merged into this branch:

git rm \
  dag/gunbc/namespace/transition_admission/extdeps_ampere_mt_collins_1u_product_brief_specifications_mt_collins_1u_bmc_manageability_network_OnboardNetworkController.dag \
  dag/gunbc/namespace/transition_admission/extdeps_ampere_mt_collins_1u_product_brief_specifications_mt_collins_1u_onboard_data_network_OnboardNetworkController.dag \
  dag/gunbc/namespace/transition_admission/gunbc_machine_intake_mtcollins1_platform_observation_mtcollins1_psu1_standing_PowerSupplyStanding.dag \
  dag/gunbc/namespace/transition_admission/gunbc_machine_intake_mtcollins1_platform_observation_mtcollins1_psu2_if_corded_PowerSupplyStanding.dag \
  dag/gunbc/namespace/transition_admission/gunbc_machine_intake_mtcollins1_platform_observation_mtcollins1_psu2_standing_PowerSupplyStanding.dag
git commit -F msg.txt   # msg.txt = the text below, verbatim

-F keeps the lines exactly as they are. The squash merge carries the message into the queue run, and nothing lands in the tree. If a row is wrong later, a later block with the same stem supersedes it.

msg.txt

Move transition admissions into the commit message (gunbc#11704)

```transition-admission
module gunbc.namespace.transition_admission.extdeps_ampere_mt_collins_1u_product_brief_specifications_mt_collins_1u_bmc_manageability_network_OnboardNetworkController

import std.types { NonEmptyStr, List }
import std.decl_ref { decl_ref }
import gunbc.compiler_frontend_program_interlock { TargetChanged }
import gunbc.namespace.transition_admission { TransitionAdmission, Binding }

data extdeps_ampere_mt_collins_1u_product_brief_specifications_mt_collins_1u_bmc_manageability_network_OnboardNetworkController: TransitionAdmission = TransitionAdmission {
  label: "gunbc#11529 hoist OnboardNetworkController and PowerSupplyStanding onto shared homes" as NonEmptyStr,
  subject: Binding {
    enclosing: decl_ref("extdeps.ampere.mt_collins_1u_product_brief.specifications", "mt_collins_1u_bmc_manageability_network"),
    spelling: "OnboardNetworkController" as NonEmptyStr,
    expected_candidates: [decl_ref("extdeps.standards.nic_attachment", "OnboardNetworkController")],
  },
  disposition: TargetChanged,
}
```

```transition-admission
module gunbc.namespace.transition_admission.extdeps_ampere_mt_collins_1u_product_brief_specifications_mt_collins_1u_onboard_data_network_OnboardNetworkController

import std.types { NonEmptyStr, List }
import std.decl_ref { decl_ref }
import gunbc.compiler_frontend_program_interlock { TargetChanged }
import gunbc.namespace.transition_admission { TransitionAdmission, Binding }

data extdeps_ampere_mt_collins_1u_product_brief_specifications_mt_collins_1u_onboard_data_network_OnboardNetworkController: TransitionAdmission = TransitionAdmission {
  label: "gunbc#11529 hoist OnboardNetworkController and PowerSupplyStanding onto shared homes" as NonEmptyStr,
  subject: Binding {
    enclosing: decl_ref("extdeps.ampere.mt_collins_1u_product_brief.specifications", "mt_collins_1u_onboard_data_network"),
    spelling: "OnboardNetworkController" as NonEmptyStr,
    expected_candidates: [decl_ref("extdeps.standards.nic_attachment", "OnboardNetworkController")],
  },
  disposition: TargetChanged,
}
```

```transition-admission
module gunbc.namespace.transition_admission.gunbc_machine_intake_mtcollins1_platform_observation_mtcollins1_psu1_standing_PowerSupplyStanding

import std.types { NonEmptyStr, List }
import std.decl_ref { decl_ref }
import gunbc.compiler_frontend_program_interlock { TargetChanged }
import gunbc.namespace.transition_admission { TransitionAdmission, Binding }

data gunbc_machine_intake_mtcollins1_platform_observation_mtcollins1_psu1_standing_PowerSupplyStanding: TransitionAdmission = TransitionAdmission {
  label: "gunbc#11529 hoist OnboardNetworkController and PowerSupplyStanding onto shared homes" as NonEmptyStr,
  subject: Binding {
    enclosing: decl_ref("gunbc.machine_intake_mtcollins1_platform_observation", "mtcollins1_psu1_standing"),
    spelling: "PowerSupplyStanding" as NonEmptyStr,
    expected_candidates: [decl_ref("gunbc.machine_intake_power_supply", "PowerSupplyStanding")],
  },
  disposition: TargetChanged,
}
```

```transition-admission
module gunbc.namespace.transition_admission.gunbc_machine_intake_mtcollins1_platform_observation_mtcollins1_psu2_if_corded_PowerSupplyStanding

import std.types { NonEmptyStr, List }
import std.decl_ref { decl_ref }
import gunbc.compiler_frontend_program_interlock { TargetChanged }
import gunbc.namespace.transition_admission { TransitionAdmission, Binding }

data gunbc_machine_intake_mtcollins1_platform_observation_mtcollins1_psu2_if_corded_PowerSupplyStanding: TransitionAdmission = TransitionAdmission {
  label: "gunbc#11529 hoist OnboardNetworkController and PowerSupplyStanding onto shared homes" as NonEmptyStr,
  subject: Binding {
    enclosing: decl_ref("gunbc.machine_intake_mtcollins1_platform_observation", "mtcollins1_psu2_if_corded"),
    spelling: "PowerSupplyStanding" as NonEmptyStr,
    expected_candidates: [decl_ref("gunbc.machine_intake_power_supply", "PowerSupplyStanding")],
  },
  disposition: TargetChanged,
}
```

```transition-admission
module gunbc.namespace.transition_admission.gunbc_machine_intake_mtcollins1_platform_observation_mtcollins1_psu2_standing_PowerSupplyStanding

import std.types { NonEmptyStr, List }
import std.decl_ref { decl_ref }
import gunbc.compiler_frontend_program_interlock { TargetChanged }
import gunbc.namespace.transition_admission { TransitionAdmission, Binding }

data gunbc_machine_intake_mtcollins1_platform_observation_mtcollins1_psu2_standing_PowerSupplyStanding: TransitionAdmission = TransitionAdmission {
  label: "gunbc#11529 hoist OnboardNetworkController and PowerSupplyStanding onto shared homes" as NonEmptyStr,
  subject: Binding {
    enclosing: decl_ref("gunbc.machine_intake_mtcollins1_platform_observation", "mtcollins1_psu2_standing"),
    spelling: "PowerSupplyStanding" as NonEmptyStr,
    expected_candidates: [decl_ref("gunbc.machine_intake_power_supply", "PowerSupplyStanding")],
  },
  disposition: TargetChanged,
}
```