|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

## Summary

This patch fixes NOP insertion logic in the mlx5 PMD's modify header compilation for HWS (hardware steering) non-root groups. The issue is that when a COPY or ADD_FIELD command reads a field, and a subsequent SET or ADD writes to that same field, no NOP (no-operation) was inserted between them, causing firmware to reject the pattern.

## Review

### Correctness

**No issues found.**

The logic changes are straightforward:
- Before: only checked if the new command's destination matches the previous command's destination
- After: also checks if the new command touches (reads or writes) any field that was read by the previous COPY/ADD_FIELD

The additional conditions are:
1. For SET/ADD following COPY/ADD_FIELD: check if new destination matches previous source (`new_cmd.field == last_cmd.field`)
2. For COPY/ADD_FIELD following COPY/ADD_FIELD: check if new destination matches previous source (`new_cmd.dst_field == last_cmd.field`)

These are correct read-after-write hazard checks.

### Code Style

**No issues found.**

- Indentation and formatting follow DPDK style
- Boolean expressions use explicit comparisons
- Line continuations are properly aligned
- Comment explains the fix clearly

### API and Documentation

**Info:**

The commit message provides good technical detail about the failure mode (firmware syndrome, which commands trigger it). However, it could benefit from a brief mention in the release notes since this fixes a template table creation failure that users may encounter.

Consider adding a brief entry to `doc/guides/rel_notes/release_26_11.rst` (or the appropriate current release file) under "Fixed Issues" for the mlx5 PMD, something like:

```rst
* **net/mlx5: Fixed modify header pattern rejection in HWS.**

  Fixed an issue where firmware would reject modify header patterns
  (syndrome 0xa0a3a8) when a modify field action read and then wrote
  the same field without an intervening NOP in HWS non-root groups.
```

Note: This is Info-level because the patch itself is correct. Release notes are recommended for user-facing fixes but their absence is not blocking.

### Process

- Patch includes `Fixes:` tag referencing the original commit
- Includes `Cc: stable at dpdk.org` for backporting
- Has `Acked-by:` from maintainer
- Targets a driver-specific fix (no ABI impact)

**No issues found.**

Review-Result: CLEAN