|WARNING| event/cnxk: fix new-event burst enqueue with LMTST

[dpdklab at iol.unh.edu]

Test-Label: ai-code-review
Test-Status: WARNING
https://dpdk.org/patch/172718

_AI Code Review_

=== Patch Review: series.patch (via Claude) ===
Review date: 2026-11-07

# Review of DPDK Patch: event/cnxk: fix new-event burst enqueue with LMTST

## ERRORS

None identified.

## WARNINGS

**1. Missing release notes update**

The patch fixes two identified bugs (resource handling in LMTST submissions) but does not update the release notes. Bug fixes of this nature should be documented in the current release notes file under a "Fixed Issues" or "Bug Fixes" section.

**Suggested fix:** Add an entry to `doc/guides/rel_notes/release_<version>.rst` documenting these fixes.

---

**2. Missing `Cc: stable at dpdk.org` tag**

The patch includes `Fixes:` tags identifying commits that introduced the bugs, but lacks the `Cc: stable at dpdk.org` tag. Bug fixes that reference specific commits with `Fixes:` tags should typically be backported to stable branches.

**Suggested fix:** Add `Cc: stable at dpdk.org` to the commit message.

---

## INFO

**1. I/O barrier placement**

The addition of `rte_io_wmb()` after LMTST submissions is correct for ordering stores before the next iteration. This ensures memory-mapped I/O writes complete before proceeding, which is the appropriate barrier for device register access.

The placement after both `roc_lmt_submit_steorl()` calls (when `sz1` is non-zero) is correct, as it orders all LMTST operations in the current iteration before moving to the next.

---

**2. Small burst optimization**

The change to handle bursts smaller than `BIT(ROC_SSO_AW_PER_LMT_LINE_LOG2)` event-by-event instead of using the LMTST path is a reasonable optimization. It avoids LMTST overhead for small bursts where the per-event path may be more efficient.

The early return pattern (breaking on first failure and returning the number of successfully enqueued events) is consistent with the existing `cn10k_sso_hws_new_event()` / `cn20k_sso_hws_new_event()` semantics.

---

**3. Code duplication**

The patch applies identical changes to both `cn10k_worker.c` and `cn20k_worker.c`. This is consistent with the existing code structure where CN10K and CN20K have parallel implementations. While code duplication is generally discouraged, this appears to be the established pattern for this driver due to hardware differences between the variants.

---

Review-Result: WARNING