feat: add Sushy emulator vendor implementation

[fabiendupont]

Add Sushy (sushy-tools) as a recognized Redfish vendor for development and
testing without physical BMC hardware. Sushy is an OpenStack Redfish emulator
backed by libvirt — it translates Redfish power/boot operations into libvirt
domain commands.

The Sushy Bmc wraps RedfishStandard and delegates read operations and
power/boot control (which Sushy handles via libvirt). Operations that Sushy
cannot support are stubbed:

• Account management (create_user, delete_user, change_password,
  get_accounts): accept silently — Sushy has no AccountService endpoint
• BIOS configuration (machine_setup, set_bios, reset_bios): no-op —
  VMs have no real BIOS
• Lockdown: no-op
• GPU sensors, secure boot certificates, attestation: NotSupported
• Boot order / infinite boot: always report as set up

Vendor detection matches "sushy", "contoso", or "redvirt" in the
ServiceRoot Vendor field. "Contoso" is the fictional manufacturer name from
DMTF's Redfish mockup data that Sushy ships with in its default Chassis
templates. "RedVirt" is an alternate configuration alias.

Testing

• cargo check
• cargo clippy --all-targets -- -D warnings
• End-to-end: NICo deployed on CRC with 3 Sushy VMs explored successfully

[copy-pr-bot]

This pull request requires additional validation before any workflows can run on NVIDIA's runners.

Pull request vetters can view their responsibilities here.

Contributors can view more details about this message here.

[Copilot]

Pull request overview

This PR adds a new Redfish vendor implementation for the Sushy (sushy-tools) emulator so the library can be used for development/testing against libvirt-backed virtual BMCs without physical hardware.

Changes:

• Introduces a new Sushy vendor (src/sushy.rs) that wraps RedfishStandard and stubs/overrides unsupported behaviors.
• Extends vendor detection to recognize sushy, contoso, and redvirt in ServiceRoot.Vendor.
• Wires the new vendor into the standard vendor-dispatch path.

Reviewed changes

Copilot reviewed 4 out of 4 changed files in this pull request and generated 5 comments.

File	Description
src/sushy.rs	Adds the Sushy-specific Redfish implementation that delegates most calls to RedfishStandard and overrides emulator-incompatible operations.
src/standard.rs	Adds vendor dispatch for RedfishVendor::Sushy to construct the new crate::sushy::Bmc.
src/model/service_root.rs	Adds RedfishVendor::Sushy and maps "sushy", "contoso", and "redvirt" to it during vendor detection.
src/lib.rs	Registers the new sushy module in the crate.

Comments suppressed due to low confidence (1)

src/sushy.rs:412

• Per the PR description, BIOS reset should be a no-op for Sushy. Delegating reset_bios to RedfishStandard can return errors (or perform unsupported operations) on the emulator.

    fn reset_bios<'a>(&'a self) -> crate::RedfishFuture<'a, Result<(), RedfishError>> {
        Box::pin(async move { self.standard.reset_bios().await })
    }

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[s3rj1k]

Amazing! Also check my Sushy fork with additional features - https://github.com/s3rj1k/sushy-tools

All of the changes are proposed as MRs in OpenDev.

My point is that Sushy is missing some of the functionality that NiCO needs to have and it is way better to fix that on Sushy side and keep Rust side clearer.

https://review.opendev.org/c/openstack/sushy-tools/+/991912
https://review.opendev.org/c/openstack/sushy-tools/+/991910
https://review.opendev.org/c/openstack/sushy-tools/+/991909
https://review.opendev.org/c/openstack/sushy-tools/+/991908
https://review.opendev.org/c/openstack/sushy-tools/+/991907
https://review.opendev.org/c/openstack/sushy-tools/+/991906
https://review.opendev.org/c/openstack/sushy-tools/+/991905
https://review.opendev.org/c/openstack/sushy-tools/+/991904

[fabiendupont]

@s3rj1k, absolutely agree that extending Sushy is the proper way, just slower :)
That's why I went with stubbing in Rust, so I would have a short term solution. And it's easy to remove the stubs as the MRs are merged.

[s3rj1k]

@s3rj1k, absolutely agree that extending Sushy is the proper way, just slower :) That's why I went with stubbing in Rust, so I would have a short term solution. And it's easy to remove the stubs as the MRs are merged.

I would be nice to ping OpenStack people and facilitate reviews of MRs TBH.

[dtantsur]

I would be nice to ping OpenStack people and facilitate reviews of MRs TBH.

I'm aware of them but please note that extending sushy-tools is very far from being a priority for any of us. If it's blocking this PR, please go ahead with stubs.

[kfelternv]

@fabiendupont We want to get this merged in, could you resolve conflicts and mark any review comments as resolved?

[fabiendupont]

@kfelternv, I'm on PTO without my computer right now. I should be able to do it next week. Is that okay time wise?

[fabiendupont]

@kfelternv, back to work :)
I marked all comments as resolved and rebased.

[poroh]

/ok to test 4ba6c5c

[poroh]

@fabiendupont cargo build failed:

error[E0050]: method `bmc_reset` has 1 parameter but the declaration in trait `Redfish::bmc_reset` has 2
   --> src/sushy.rs:182:22
    |
182 |       fn bmc_reset<'a>(&'a self) -> crate::RedfishFuture<'a, Result<(), RedfishError>> {
    |                        ^^^^^^^^ expected 2 parameters, found 1
    |
   ::: src/lib.rs:200:9
    |
200 | /         &'a self,
201 | |         reset_type: Option<ManagerResetType>,
    | |____________________________________________- trait requires 2 parameters

error[E0061]: this method takes 1 argument but 0 arguments were supplied
   --> src/sushy.rs:183:45
    |
183 |         Box::pin(async move { self.standard.bmc_reset().await })
    |                                             ^^^^^^^^^-- argument #1 of type `std::option::Option<manager::ManagerResetType>` is missing
    |
note: method defined here

[poroh]

@fabiendupont I wanted to merge you PR first and then #132 but because your PR had build error #132 was merged first. Could you please remove methods that delegates calls to standard implmenetation. Now it happens by default, you don't need to provide this implementation.

Example of function from you PR that can be removed:

    fn update_firmware<'a>(
        &'a self,
        firmware: tokio::fs::File,
    ) -> crate::RedfishFuture<'a, Result<crate::model::task::Task, RedfishError>> {
        Box::pin(async move { self.standard.update_firmware(firmware).await })
    }

[fabiendupont]

@poroh, #132 is really nice and allowed me to remove a good chunck of code.
Locally, the build and tests pass.

[poroh]

/ok to test b19881f

[poroh]

@fabiendupont Next one is formatting:

Diff in /__w/libredfish/libredfish/src/sushy.rs:43:
 use crate::model::task::Task;
 use crate::model::update_service::ComponentType;
 use crate::{
-    standard::RedfishStandard,
-    BiosProfileType, MachineSetupStatus, Redfish, RedfishError, RoleId,
+    standard::RedfishStandard, BiosProfileType, MachineSetupStatus, Redfish, RedfishError, RoleId,
 };
 
 pub struct Bmc {
Diff in /__w/libredfish/libredfish/src/sushy.rs:184:
     ) -> crate::RedfishFuture<'a, Result<crate::Status, RedfishError>> {
         Box::pin(async move {
             info!("Sushy: reporting lockdown as enabled (emulator)");
-            Ok(crate::Status { status: crate::StatusInternal::Enabled, message: "Sushy emulator".to_string() })
+            Ok(crate::Status {
+                status: crate::StatusInternal::Enabled,
+                message: "Sushy emulator".to_string(),
+            })
         })
     }

[fabiendupont]

Sorry @poroh. Went too fast. Updated.

[poroh]

/ok to test 6fa3844

[poroh]

@fabiendupont FYI: https://github.com/NVIDIA/libredfish/releases/tag/v0.47.0

[s3rj1k]

@fabiendupont so I've run latest master of NiCO that already contains this change against vanilla master of Sushy and it does not work.

HTTP 404 Not Found at https://10.42.71.103:8000/redfish/v1/UpdateService/FirmwareInventory

Please share your experience.

[s3rj1k]

So, after fixing above issue directly in Sushy with some of previously proposed changes to Sushy itself, I get stuck in HOSTINITIALIZING/WAITINGFORLOCKDOWN independent of what I suppy from admin-cli (might be NiCO bug)

  With the vendor pinned to Sushy, there is no configuration that gets past handler.rs:7337. It needs the one-line change in libredfish's sushy.rs:181-192:

  Err(RedfishError::NotSupported("lockdown_status".to_string()))

[fabiendupont]

@s3rj1k, it's my fault. I was not using a vanilla Sushy, but a version from Metal3 with a custom vendor.json. I switch to vanilla and created #137 to fix the issues.