vmm_test: MANA VF reset tests

[erfrimod]

Adding to the VMM Petri integration test suite with GDMA infrastructure and Petri coverage for the MANA VF reset.

• Adding a gdma device resolver that allows tests to send HWC EQEs for the MANA emulator.
• Adding Petri control for a paired VTL2 device with subordinate VTL0 device.
• Adding VPCI OFFER handling needed to run the tests.
• OpenHCL Linux-direct MANA NIC tests refactored into a separate file.
  • New OpenHCL linux-direct tests for VF Reset (with and without VTL0 revoke)

[Copilot]

🟡 Changes recommended

The new VF reset test path has an identified race that can make the integration tests flaky without a bounded wait/retry after asynchronous reconfiguration.

Once you've addressed the issues Copilot identified, you can request another Copilot review.

Pull request overview

This PR expands the Petri-based VMM integration test suite to cover MANA NIC behavior in OpenHCL Linux-direct scenarios, including VF reset/reconfiguration, by adding a test-controllable GDMA/MANA device path and the minimal supporting plumbing needed to exercise it end-to-end.

Changes:

• Introduce a test-controllable GDMA device (gdma_test) that can inject HWC EQEs (e.g., VF reset) into the emulator.
• Extend Petri OpenVMM configuration to create a paired VTL2 device with a subordinate VTL0 VF for MANA testing.
• Add/refactor OpenHCL Linux-direct MANA integration tests and add GET-side VPCI device control handling needed by the new tests.

File summaries

File	Description
vmm_tests/vmm_tests/tests/tests/x86_64/openhcl_linux_direct/mana_nic.rs	New OpenHCL Linux-direct MANA NIC tests, including VF reset coverage.
vmm_tests/vmm_tests/tests/tests/x86_64/openhcl_linux_direct.rs	Registers the new mana_nic test module.
vm/devices/net/gdma/src/resolver.rs	Refactors vport resolution into a shared helper for production + test resolvers.
vm/devices/net/gdma/src/queues.rs	Exposes queue ID offset internally for test injection plumbing.
vm/devices/net/gdma/src/lib.rs	Adds test_helpers feature module to support cross-crate test injection.
vm/devices/net/gdma/src/bnic.rs	Implements VTL2 filter move/query support needed by VF reconfiguration flows.
vm/devices/net/gdma/Cargo.toml	Adds test_helpers feature flag.
vm/devices/net/gdma_test/src/resolver.rs	New resolver creating GDMA device + background EQE injection task for tests.
vm/devices/net/gdma_test/src/lib.rs	New gdma_test crate entry point.
vm/devices/net/gdma_test/Cargo.toml	New crate manifest wiring gdma with test_helpers.
vm/devices/net/gdma_resources/src/lib.rs	Adds MeshPayload resource handle + request enum for test-controlled GDMA.
vm/devices/get/guest_emulation_device/src/lib.rs	Adds handling for VPCI_DEVICE_CONTROL messages (OFFER/REVOKE/RESET).
petri/src/vm/openvmm/modify.rs	Adds paired VTL2/VTL0 NIC config and ManaTestControl request API.
petri/src/vm/openvmm/mod.rs	Re-exports ManaTestControl for test use.
openvmm/openvmm_resources/src/lib.rs	Registers the new GdmaTestDeviceResolver.
openvmm/openvmm_resources/Cargo.toml	Adds gdma_test dependency.
Cargo.toml	Adds gdma_test to workspace dependencies.
Cargo.lock	Records the new gdma_test package.

Review details

• Files reviewed: 17/18 changed files
• Comments generated: 1
• Review effort level: Lite

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.

[erfrimod]

I agree that vf reset is racy and validation should wait, but I don't have a good signal to wait for. Ideally, I will add a counter for vf_reset and then the test can poll inspect for the increment.

As-is, the test is still useful as a smoke test. I don't anticipate flakey test failures, but rather that the test may pass before the code is finished reconfiguring.

[github-actions]

3 Petri tests failed (1 unstable)