Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Message-ID: <8b2589bd6303133fd27cab1af27b096a5f848074.camel@redhat.com>
Subject: Re: [PATCH net-next 2/5] net: hns3: support ndo_select_queue()
From:   Paolo Abeni <pabeni@redhat.com>
To:     Guangbin Huang <huangguangbin2@huawei.com>, davem@davemloft.net,
        kuba@kernel.org
Cc:     edumazet@google.com, netdev@vger.kernel.org,
        linux-kernel@vger.kernel.org, lipeng321@huawei.com,
        lanhao@huawei.com
Date:   Tue, 06 Sep 2022 15:15:31 +0200
In-Reply-To: <20220905081539.62131-3-huangguangbin2@huawei.com>
References: <20220905081539.62131-1-huangguangbin2@huawei.com>
         <20220905081539.62131-3-huangguangbin2@huawei.com>
Content-Type: text/plain; charset="UTF-8"
User-Agent: Evolution 3.42.4 (3.42.4-2.fc35) 
MIME-Version: 1.0
Content-Transfer-Encoding: 7bit
Precedence: bulk

On Mon, 2022-09-05 at 16:15 +0800, Guangbin Huang wrote:
> To support tx packets to select queue according to its dscp field after
> setting dscp and tc map relationship, this patch implements
> ndo_select_queue() to set skb->priority according to the user's setting
> dscp and priority map relationship.
> 
> Signed-off-by: Guangbin Huang <huangguangbin2@huawei.com>
> ---
>  .../net/ethernet/hisilicon/hns3/hns3_enet.c   | 46 +++++++++++++++++++
>  1 file changed, 46 insertions(+)
> 
> diff --git a/drivers/net/ethernet/hisilicon/hns3/hns3_enet.c b/drivers/net/ethernet/hisilicon/hns3/hns3_enet.c
> index 481a300819ad..82f83e3f8162 100644
> --- a/drivers/net/ethernet/hisilicon/hns3/hns3_enet.c
> +++ b/drivers/net/ethernet/hisilicon/hns3/hns3_enet.c
> @@ -2963,6 +2963,51 @@ static int hns3_nic_set_vf_mac(struct net_device *netdev, int vf_id, u8 *mac)
>  	return h->ae_algo->ops->set_vf_mac(h, vf_id, mac);
>  }
>  
> +#define HNS3_INVALID_DSCP		0xff
> +#define HNS3_DSCP_SHIFT			2
> +
> +static u8 hns3_get_skb_dscp(struct sk_buff *skb)
> +{
> +	__be16 protocol = skb->protocol;
> +	u8 dscp = HNS3_INVALID_DSCP;
> +
> +	if (protocol == htons(ETH_P_8021Q))
> +		protocol = vlan_get_protocol(skb);
> +
> +	if (protocol == htons(ETH_P_IP))
> +		dscp = ipv4_get_dsfield(ip_hdr(skb)) >> HNS3_DSCP_SHIFT;
> +	else if (protocol == htons(ETH_P_IPV6))
> +		dscp = ipv6_get_dsfield(ipv6_hdr(skb)) >> HNS3_DSCP_SHIFT;
> +
> +	return dscp;
> +}
> +
> +static u16 hns3_nic_select_queue(struct net_device *netdev,
> +				 struct sk_buff *skb,
> +				 struct net_device *sb_dev)
> +{
> +	struct hnae3_handle *h = hns3_get_handle(netdev);
> +	u8 dscp, priority;
> +	int ret;
> +
> +	if (h->kinfo.tc_map_mode != HNAE3_TC_MAP_MODE_DSCP ||
> +	    !h->ae_algo->ops->get_dscp_prio)
> +		goto out;
> +
> +	dscp = hns3_get_skb_dscp(skb);
> +	if (unlikely(dscp == HNS3_INVALID_DSCP))
> +		goto out;
> +
> +	ret = h->ae_algo->ops->get_dscp_prio(h, dscp, NULL, &priority);

This introduces an additional, unneeded indirect call in the fast path,
you could consider replacing the above with a direct call to
hclge_get_dscp_prio() - again taking care of the CONFIG_HNS3_DCB
dependency.

Cheers,

Paolo