|WARNING| net/idpf: propagate PCI NUMA node to vport ethdev

[dpdklab at iol.unh.edu]

Test-Label: ai-code-review
Test-Status: WARNING
https://dpdk.org/patch/167443

_AI Code Review_

=== Patch Review: series.patch (via Claude) ===
Review date: 2026-08-11

# DPDK Patch Review

## Patch Summary
The patch fixes missing NUMA node propagation in the idpf driver when creating vport ethdevs. The issue causes `rte_eth_dev_socket_id()` to always return 0, leading to incorrect NUMA-unaware memory allocations.

---

## Review Findings

### **Errors**: None

### **Warnings**: None

### **Info**: None

---

## Analysis

### Correctness
The fix is correct. The idpf driver bypasses the normal PCI probe path (which calls `rte_eth_copy_pci_info()`) when creating vport ethdevs via `rte_eth_dev_create()`. Setting `dev->data->numa_node` from the parent PCI device's `numa_node` is the appropriate way to restore NUMA affinity.

The assignment occurs early in initialization (before any allocations that depend on NUMA node), and the source (`dev->device->numa_node`) is guaranteed to be valid at this point since it comes from the parent PCI device.

### Code Style
- The added line follows DPDK code style
- Placement is logical (immediately after setting `dev_ops`, before other vport initialization)
- No forbidden tokens or patterns introduced

### API and Documentation
- No API changes - this is an internal fix
- The `Fixes:` tag correctly identifies the commit that introduced the regression
- `Cc: stable at dpdk.org` is appropriate for a bug fix affecting NUMA-aware systems

### Process Compliance
- The patch is minimal and focused
- It fixes the specific issue without introducing unrelated changes
- Backportable to stable branches (simple one-line fix)

---

## Conclusion

**This patch is acceptable as-is.**

The fix correctly restores NUMA node information that was lost when the driver switched to `rte_eth_dev_create()`. The approach matches the standard pattern used in other DPDK drivers, and the placement in the initialization sequence is appropriate.