|SUCCESS| net/gve: check the Rx scatter MTU at device start

[dpdklab at iol.unh.edu]

Test-Label: ai-code-review
Test-Status: SUCCESS
https://dpdk.org/patch/171827

_AI Code Review_

=== Patch Review: series.patch (via Claude) ===
Review date: 2026-10-22

# DPDK Patch Review

## Patch Summary
This patch fixes an MTU validation issue in the GVE driver where reconfiguring a stopped port with a larger MTU could skip validation checks, potentially causing the driver to chain large frames when scatter Rx was not enabled by the application.

---

## ERRORS

None identified.

---

## WARNINGS

None identified.

---

## INFO

**1. Function extraction improves maintainability**

The patch extracts the MTU vs scatter check into a separate `gve_rx_check_scatter()` function and calls it from both `gve_dev_start()` and `gve_dev_mtu_set()`. This is good design -- the validation logic is now in one place and runs at both configuration points (start and MTU change).

**2. Comment accurately describes the issue**

The comment added in `gve_dev_start()`:
```c
/* rte_eth_dev_configure() sets the MTU and the Rx offloads without
 * calling mtu_set, and keeps Rx queues already set up, so check again.
 */
```

This correctly explains why the check is needed at start time in addition to `mtu_set()`.

**3. Consistent error handling pattern**

The patch follows the existing driver pattern:
- Returns `-EINVAL` for configuration errors
- Uses `PMD_DRV_LOG(ERR, ...)` for error messages
- Propagates error codes from helper functions

---

## VERIFICATION NOTES

I verified the following aspects of the patch:

**Correctness:**
- No resource allocations, so no leak risk
- Error paths correctly propagate return values
- The new function is called at both necessary points (start and mtu_set)
- The check is read-only (does not modify state), so safe to call multiple times

**Style:**
- Code follows DPDK C style (tabs for indentation, proper brace placement)
- Comment style is consistent with DPDK guidelines
- Variable names follow lowercase-with-underscores convention
- Explicit NULL comparison in `if (rxq != NULL ...)` follows DPDK style
- Explicit zero comparison in return value checks follows DPDK style

**Logic:**
- The function correctly skips the check for DQA (when `!gve_is_gqi(priv)`)
- The function correctly allows scatter-capable configurations (when `RTE_ETH_RX_OFFLOAD_SCATTER` is set)
- The loop correctly checks all configured Rx queues
- The `gve_rx_gqi_needs_scatter()` helper is already present and used correctly

**Process:**
- Patch includes proper `Fixes:` tag referencing the original commit
- Includes `Cc: stable at dpdk.org` for backport
- Includes `Reported-by:` credit
- Author states the patch is built with `-Dwerror=true` for both gcc and clang

---

Review-Result: CLEAN