|SUCCESS| common/cnxk: disable flow control on pool destroy

[dpdklab at iol.unh.edu]

Test-Label: ai-code-review
Test-Status: SUCCESS
https://dpdk.org/patch/172274

_AI Code Review_

=== Patch Review: series.patch (via Claude) ===
Review date: 2026-10-30

# DPDK Patch Review

## Analysis

This patch modifies the `npa_aura_pool_fini()` function in `drivers/common/cnxk/roc_npa.c` to disable flow control when destroying NPA (Network Pool Allocator) pools and auras by clearing the `fc_ena` field in both pool and aura contexts.

### Evidence Review

The patch adds four lines:
- Line 369-370: `pool_req->pool.fc_ena = 0;` and `pool_req->pool_mask.fc_ena = ~pool_req->pool_mask.fc_ena;`
- Line 386-387: `aura_req->aura.fc_ena = 0;` and `aura_req->aura_mask.fc_ena = ~aura_req->aura_mask.fc_ena;`

These mirror the existing pattern for the `ena` field on lines 367-368 and 384-385.

### Correctness Check

The code follows the same pattern as the existing `ena` field handling:
1. Set the field to 0 (disable)
2. Set the corresponding mask field to all-1s (bitwise NOT of 0)

This is consistent with the NPA AQ (Administrative Queue) instruction mechanism for partial updates, where the mask indicates which fields should be written.

### Style Check

- No forbidden tokens introduced
- Follows existing code style and indentation
- Uses consistent formatting with surrounding code
- No trailing whitespace
- Proper alignment with existing statements

### API and Documentation

- This is an internal driver change, not a public API modification
- No new API functions added
- No release notes appear necessary for this internal driver fix
- This appears to be a bug fix that ensures flow control is properly disabled during cleanup

### Testing

The patch does not include tests, but this is acceptable for a driver-internal fix. Testing would occur through normal pool/aura lifecycle testing in the hardware.

## Findings

No issues identified. The patch correctly implements flow control disable during pool/aura finalization, following the established code pattern and maintaining consistency with the existing `ena` field handling.

Review-Result: CLEAN