[Intel-wired-lan] [PATCH bpf v2 3/9] ice: fix rxq info registering in mbuf packets

[bot+bpf-ci at kernel.org]

> diff --git a/drivers/net/ethernet/intel/ice/ice_base.c b/drivers/net/ethernet/intel/ice/ice_base.c
> index eadb1e3d12b3..511d803cf0a4 100644
> --- a/drivers/net/ethernet/intel/ice/ice_base.c
> +++ b/drivers/net/ethernet/intel/ice/ice_base.c
> @@ -663,23 +663,12 @@ static int ice_vsi_cfg_rxq(struct ice_rx_ring *ring)
>  	int err;
>
>  	if (ring->vsi->type == ICE_VSI_PF || ring->vsi->type == ICE_VSI_SF) {

[ ... ]

> @@ -702,14 +691,13 @@ static int ice_vsi_cfg_rxq(struct ice_rx_ring *ring)
>  			if (err)
>  				return err;
>
> -			if (!xdp_rxq_info_is_reg(&ring->xdp_rxq)) {
> -				err = __xdp_rxq_info_reg(&ring->xdp_rxq, ring->netdev,
> -							 ring->q_index,
> -							 ring->q_vector->napi.napi_id,
> -							 ring->rx_buf_len);
> -				if (err)
> -					goto err_destroy_fq;
> -			}
> +			err = __xdp_rxq_info_reg(&ring->xdp_rxq, ring->netdev,
> +						 ring->q_index,
> +						 ring->q_vector->napi.napi_id,
> +						 ring->rx_buf_len);
> +			if (err)
> +				goto err_destroy_fq;

The registration condition check was removed, so __xdp_rxq_info_reg() is
now called unconditionally. This works correctly when cleanup properly
unregisters the xdp_rxq, but there appears to be a gap for ICE_VSI_SF
in non-XSK mode.

Looking at ice_clean_rx_ring() in ice_txrx.c:

    if (rx_ring->vsi->type == ICE_VSI_PF &&
        xdp_rxq_info_is_reg(&rx_ring->xdp_rxq)) {
        xdp_rxq_info_detach_mem_model(&rx_ring->xdp_rxq);
        xdp_rxq_info_unreg(&rx_ring->xdp_rxq);
    }

This only unregisters for ICE_VSI_PF, not ICE_VSI_SF. With the old code,
the conditional registration skipped re-registration for ICE_VSI_SF. With
this change, for ICE_VSI_SF in non-XSK mode during reconfiguration:

1. ice_clean_rx_ring() - doesn't unregister (only checks ICE_VSI_PF)
2. ice_vsi_cfg_rxq() - unconditionally calls __xdp_rxq_info_reg()
3. __xdp_rxq_info_reg() sees REG_STATE_REGISTERED and triggers:

    WARN(1, "Missing unregister, handled but fix driver");

While this is handled gracefully by the XDP core (it unregisters internally
before re-registering), will this cause spurious WARN splats for subfunction
devices on every reconfiguration?

Should the ICE_VSI_PF check in ice_clean_rx_ring() be updated to also
include ICE_VSI_SF to match the registration path?


---
AI reviewed your patch. Please fix the bug or email reply why it's not a bug.
See: https://github.com/kernel-patches/vmtest/blob/master/ci/claude/README.md

CI run summary: https://github.com/kernel-patches/bpf/actions/runs/21960574055

AI-authorship-score: low
AI-authorship-explanation: The commit shows domain-specific knowledge of ice driver internals and XDP subsystem, with targeted changes that suggest human authorship by an Intel driver developer.
issues-found: 1
issue-severity-score: low
issue-severity-explanation: The issue causes WARN splats for ICE_VSI_SF devices during reconfiguration, but is handled gracefully by XDP core without functional impact.