Fix non-deterministic PCI topology across VM stop/start cycles

[jamepark4]

PciManager counters (pci_slot, pci_chassis, pci_addr) were initialized once in QemuMachine.initialize but never reset between runs. When process_device_commands was called on VM restart, the counters continued from where they left off, producing different slot/chassis/addr assignments — and therefore different bus names (e.g., root5 on first boot, root11 on second boot).

This caused two problems after clean_boot (stop + restart):

• Spare PCIe root port bus names (@spare_pcie_root_port_N@) changed, breaking hotplug tests that reference them
• Network/storage devices landed on different PCI slots, causing Windows to treat them as new hardware and lose network configuration

Fix: Extract counter initialization into PciManager#reset and call it at the start of process_device_commands, matching the existing pattern of resetting @device_commands. This makes PCI allocation deterministic across restarts.

[Copilot]

Pull request overview

Refreshes QEMU replacement mappings after Functest VM reboot and snapshot operations.

Changes:

• Exposes the current VM replacement map.
• Refreshes mappings after VM recreation.

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 1 comment.

File	Description
lib/setupmanagers/qemuhck/qemuhck.rb	Exposes VM mappings.
lib/setupmanagers/functest_client.rb	Refreshes mappings after reboot/snapshot operations.

---

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

[kostyanf14]

In the event that a testcase uses clean_boot: true along with pcie-spare-root-ports, the new VM's allocated PCI devices get new bus name calculations.

This is a bug. We can create wrong network connections too.

In this case, the recalculation mask issue.
Stop/Start should not change the topology of any device, not only the PCI bus

[jamepark4]

In the event that a testcase uses clean_boot: true along with pcie-spare-root-ports, the new VM's allocated PCI devices get new bus name calculations.

This is a bug. We can create wrong network connections too.

In this case, the recalculation mask issue. Stop/Start should not change the topology of any device, not only the PCI bus

I'll try and retool this to address the underlying issue.