Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
Subject: Re: [PATCH 2/3] net: openvswitch: use core API for updating TX stats
To:     Lev Stipakov <lstipakov@gmail.com>,
        Pravin B Shelar <pshelar@ovn.org>,
        "David S. Miller" <davem@davemloft.net>,
        Jakub Kicinski <kuba@kernel.org>, netdev@vger.kernel.org,
        dev@openvswitch.org, linux-kernel@vger.kernel.org
Cc:     Lev Stipakov <lev@openvpn.net>
References: <20201112111150.34361-1-lev@openvpn.net>
From:   Heiner Kallweit <hkallweit1@gmail.com>
Message-ID: <598c779c-fb0b-a9a6-43be-3a7cd5b38946@gmail.com>
Date:   Fri, 13 Nov 2020 08:40:42 +0100
User-Agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:78.0) Gecko/20100101
 Thunderbird/78.4.3
MIME-Version: 1.0
In-Reply-To: <20201112111150.34361-1-lev@openvpn.net>
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: 7bit
Precedence: bulk

Am 12.11.2020 um 12:11 schrieb Lev Stipakov:
> Commit d3fd65484c781 ("net: core: add dev_sw_netstats_tx_add") has added
> function "dev_sw_netstats_tx_add()" to update net device per-cpu TX
> stats.
> 
> Use this function instead of own code. While on it, replace
> "len" variable with "skb->len".
> 
Using dev_sw_netstats_tx_add() is fine, however you have to keep
variable len, see remark in the code.

In addition you can replace internal_get_stats() with dev_get_tstats64().

> Signed-off-by: Lev Stipakov <lev@openvpn.net>
> ---
>  net/openvswitch/vport-internal_dev.c | 16 +++++-----------
>  1 file changed, 5 insertions(+), 11 deletions(-)
> 
> diff --git a/net/openvswitch/vport-internal_dev.c b/net/openvswitch/vport-internal_dev.c
> index 1e30d8df3ba5..116738d36e02 100644
> --- a/net/openvswitch/vport-internal_dev.c
> +++ b/net/openvswitch/vport-internal_dev.c
> @@ -33,23 +33,17 @@ static struct internal_dev *internal_dev_priv(struct net_device *netdev)
>  static netdev_tx_t
>  internal_dev_xmit(struct sk_buff *skb, struct net_device *netdev)
>  {
> -	int len, err;
> +	int err;
>  
> -	len = skb->len;
>  	rcu_read_lock();
>  	err = ovs_vport_receive(internal_dev_priv(netdev)->vport, skb, NULL);

We would have a well-hidden problem here. ovs_vport_receive() calls function
ovs_dp_process_packet() that frees the skb under certain circumstances.
Note that the skb can be freed even if ovs_vport_receive() returns OK.
Using skb->len after calling ovs_vport_receive() could result in a
use-after-free therefore. So you have to keep variable len.

>  	rcu_read_unlock();
>  
> -	if (likely(!err)) {
> -		struct pcpu_sw_netstats *tstats = this_cpu_ptr(netdev->tstats);
> -
> -		u64_stats_update_begin(&tstats->syncp);
> -		tstats->tx_bytes += len;
> -		tstats->tx_packets++;
> -		u64_stats_update_end(&tstats->syncp);
> -	} else {
> +	if (likely(!err))
> +		dev_sw_netstats_tx_add(netdev, 1, skb->len);
> +	else
>  		netdev->stats.tx_errors++;
> -	}
> +
>  	return NETDEV_TX_OK;
>  }
>  
>