Re: [PATCH net-next] net: bcmgenet: convert RX path to page_pool

[Simon Horman]

On Thu, Jun 04, 2026 at 06:32:06PM +0100, Simon Horman wrote:
> From: 'Simon Horman' <horms@xxxxxxxxxx>
> 
> This is an AI-generated review of your patch. The human sending this
> email has considered the AI review valid, or at least plausible.
> Full review at: https://netdev-ai.bots.linux.dev/sashiko/
> ---
> Now I have the exact diff. Let me produce the report.
> 
> commit ff8ce4b2334ded9490897cdd2cff5a25a4a0a438
> Author: Nicolai Buchwitz <nb@xxxxxxxxxxx>
> 
> net: bcmgenet: convert RX path to page_pool
> 
> This patch converts the bcmgenet RX path from per-packet
> __netdev_alloc_skb() + dma_map_single() to page_pool, builds SKBs from
> pool pages via napi_build_skb() with skb_mark_for_recycle(), and removes
> the now-unused priv->rx_buf_len field and rx_dma_failed soft MIB
> counter.
> 
> > diff --git a/drivers/net/ethernet/broadcom/genet/bcmgenet.c b/drivers/net/ethernet/broadcom/genet/bcmgenet.c
> > index 7c11cf916762..80dbfba9fa88 100644
> > --- a/drivers/net/ethernet/broadcom/genet/bcmgenet.c
> > +++ b/drivers/net/ethernet/broadcom/genet/bcmgenet.c
> 
> [ ... ]
> 
> > @@ -1153,7 +1159,6 @@ static const struct bcmgenet_stats bcmgenet_gstrings_stats[] = {
> >  			UMAC_RBUF_ERR_CNT_V1),
> >  	STAT_GENET_MISC("mdf_err_cnt", mib.mdf_err_cnt, UMAC_MDF_ERR_CNT),
> >  	STAT_GENET_SOFT_MIB("alloc_rx_buff_failed", mib.alloc_rx_buff_failed),
> > -	STAT_GENET_SOFT_MIB("rx_dma_failed", mib.rx_dma_failed),
> >  	STAT_GENET_SOFT_MIB("tx_dma_failed", mib.tx_dma_failed),
> 
> [Medium]
> The string "rx_dma_failed" exposed via ethtool -S is user-visible: monitoring
> agents (collectd, telegraf, Prometheus exporters, ad-hoc scripts) may key off
> that name, and the strings list is ordered, so removing one shifts the index
> of every counter that follows for tools that index by offset rather than name.
> 
> Should the changelog mention this user-visible change explicitly, or could
> the entry be retained reporting 0 to preserve the existing string set?
> 
> Worth noting that the sibling driver drivers/net/ethernet/broadcom/bcmsysport.c
> still exposes rx_dma_failed, so the two drivers in the same Broadcom directory
> will diverge after this patch.

FWIIW, I feel that mentioning the change in the patch description
would be sufficient. And I don't buy the index argument, in my view these
are key value pairs.

...