|SUCCESS| net/mlx5: fix NOP insertion for modify field read-write

[dpdklab at iol.unh.edu]

Test-Label: ai-code-review
Test-Status: SUCCESS
https://dpdk.org/patch/172741

_AI Code Review_

=== Patch Review: series.patch (via Claude) ===
Review date: 2026-11-09

# DPDK Patch Review

## Patch Summary
This patch fixes NOP insertion logic in the mlx5 PMD's hardware steering modify field action handling. The issue occurs when a COPY or ADD_FIELD reads a field that is subsequently written by SET or ADD, causing firmware rejection (syndrome 0xa0a3a8).

---

## ERRORS

None identified.

---

## WARNINGS

None identified.

---

## INFO

**Code clarity - logic could be documented**

The fix adds three new conditions checking `last_cmd.field` (the source field of COPY/ADD_FIELD operations). While the code is correct, a brief comment explaining why both source and destination must be checked would help future maintainers understand the firmware constraint being addressed:

```c
/* When a COPY or ADD_FIELD reads a field (last_cmd.field),
 * firmware requires a NOP before that field is written.
 */
```

This would make the relationship between the checks and the firmware syndrome mentioned in the commit message more explicit.

---

**Verification suggestion (not a code issue)**

The patch mentions firmware syndrome 0xa0a3a8 occurs "at template table creation" when the pattern is rejected. If there are existing tests in `app/test-pmd` or the functional test suite that exercise modify field actions, verifying they cover COPY-SET and ADD_FIELD-SET sequences on the same field would confirm this fix is tested. (Not requesting new tests if coverage already exists, just noting this for thoroughness.)

---

Review-Result: CLEAN