Re: [PATCH net-next 3/8] net: bcmgenet: use CHECKSUM_COMPLETE for NETIF_F_RXCSUM

[Jakub Kicinski]

On Tue, 17 Dec 2019 13:02:24 -0800, Doug Berger wrote:
> @@ -1793,6 +1792,10 @@ static unsigned int bcmgenet_desc_rx(struct bcmgenet_rx_ring *ring,
>  
>  			status = (struct status_64 *)skb->data;
>  			dma_length_status = status->length_status;
> +			if (priv->desc_rxchk_en) {
> +				skb->csum = ntohs(status->rx_csum & 0xffff);

This adds a new warning for a W=1 C=1 build:

../drivers/net/ethernet/broadcom/genet/bcmgenet.c:1796:45: warning: cast to res\
tricted __be16                                                                  
../drivers/net/ethernet/broadcom/genet/bcmgenet.c:1796:45: warning: cast to res\
tricted __be16                                                                  
../drivers/net/ethernet/broadcom/genet/bcmgenet.c:1796:45: warning: cast to res\
tricted __be16                                                                  
../drivers/net/ethernet/broadcom/genet/bcmgenet.c:1796:45: warning: cast to res\
tricted __be16                                                                  
../drivers/net/ethernet/broadcom/genet/bcmgenet.c:1796:43: warning: incorrect t\
ype in assignment (different base types)                                        
../drivers/net/ethernet/broadcom/genet/bcmgenet.c:1796:43:    expected restrict\
ed __wsum [usertype] csum                                                       
../drivers/net/ethernet/broadcom/genet/bcmgenet.c:1796:43:    got int           
   
Could you perhaps consider a __force cast or such?

> +				skb->ip_summed = CHECKSUM_COMPLETE;
> +			}
>  		}
>  
>  		/* DMA flags and length are still valid no matter how