|WARNING| net/bnxt: fix forcing of mbuf fast free

[dpdklab at iol.unh.edu]

Test-Label: ai-code-review
Test-Status: WARNING
https://dpdk.org/patch/169868

_AI Code Review_

=== Patch Review: series.patch (via Claude) ===
Review date: 2026-09-21

# DPDK Patch Review

## Summary
This patch fixes a bug in the bnxt driver where `RTE_ETH_TX_OFFLOAD_MBUF_FAST_FREE` was unconditionally enabled at device start, overriding application configuration. The fix tracks whether the offload was requested at queue setup and only enables it when appropriate.

---

## Errors

None identified.

---

## Warnings

None identified.

---

## Info

### 1. Comparison style (Info)

**File:** `drivers/net/bnxt/bnxt_txq.c:181`

**Issue:**
```c
txq->fast_free_requested =
    (txq->offloads & RTE_ETH_TX_OFFLOAD_MBUF_FAST_FREE) != 0;
```

**Current vs Preferred:**
While the explicit `!= 0` comparison is acceptable for bitfield tests, assigning the result to a `bool` makes the comparison redundant. The bitwise AND result can be directly assigned to the boolean variable, as C will convert any non-zero value to `true`.

**Suggested alternative (optional):**
```c
txq->fast_free_requested =
    (txq->offloads & RTE_ETH_TX_OFFLOAD_MBUF_FAST_FREE);
```

This is a minor style preference. Both forms are valid; the current form is more explicit about the boolean conversion.

---

## Overall Assessment

**PASS** - The patch correctly fixes the reported bug without introducing correctness issues.

### Strengths:
1. **Correctness**: Properly saves the user's intent at queue setup time and respects it at device start
2. **Logic flow**: The flag is set after combining both `dev_conf.txmode.offloads` and `tx_conf->offloads`, capturing the complete configuration
3. **Existing safety**: The PTP path already clears the offload bit after the new flag is saved, so the flag correctly reflects the original request even in the PTP case
4. **No resource issues**: The new boolean field doesn't require cleanup and is properly initialized in the allocation path

### Technical correctness:
- The patch addresses a real bug where driver policy overrode application configuration
- The fix correctly distinguishes "user requested fast-free" from "driver forced fast-free"
- No error paths are affected since this is purely configuration tracking
- No ABI impact (internal struct change only)

The single Info-level style comment is purely optional and does not affect correctness.