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;
Date:   Sun, 16 Apr 2023 16:38:11 -0400
From:   "Michael S. Tsirkin" <mst@redhat.com>
To:     Alvaro Karsz <alvaro.karsz@solid-run.com>
Cc:     jasowang@redhat.com, davem@davemloft.net, edumazet@google.com,
        kuba@kernel.org, pabeni@redhat.com,
        virtualization@lists.linux-foundation.org, netdev@vger.kernel.org,
        linux-kernel@vger.kernel.org
Subject: Re: [PATCH net] virtio-net: reject small vring sizes
Message-ID: <20230416163751-mutt-send-email-mst@kernel.org>
References: <20230416074607.292616-1-alvaro.karsz@solid-run.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20230416074607.292616-1-alvaro.karsz@solid-run.com>
Precedence: bulk

On Sun, Apr 16, 2023 at 10:46:07AM +0300, Alvaro Karsz wrote:
> Check vring size and fail probe if a transmit/receive vring size is
> smaller than MAX_SKB_FRAGS + 2.
> 
> At the moment, any vring size is accepted. This is problematic because
> it may result in attempting to transmit a packet with more fragments
> than there are descriptors in the ring.
> 
> Furthermore, it leads to an immediate bug:
> 
> The condition: (sq->vq->num_free >= 2 + MAX_SKB_FRAGS) in
> virtnet_poll_cleantx and virtnet_poll_tx always evaluates to false,
> so netif_tx_wake_queue is not called, leading to TX timeouts.
> 
> Signed-off-by: Alvaro Karsz <alvaro.karsz@solid-run.com>
> ---
>  drivers/net/virtio_net.c | 24 ++++++++++++++++++++++++
>  1 file changed, 24 insertions(+)
> 
> diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
> index 2396c28c012..59676252c5c 100644
> --- a/drivers/net/virtio_net.c
> +++ b/drivers/net/virtio_net.c
> @@ -3745,6 +3745,26 @@ static int init_vqs(struct virtnet_info *vi)
>  	return ret;
>  }
>  
> +static int virtnet_validate_vqs(struct virtnet_info *vi)
> +{
> +	u32 i, min_size = roundup_pow_of_two(MAX_SKB_FRAGS + 2);

why power of two?

> +
> +	/* Transmit/Receive vring size must be at least MAX_SKB_FRAGS + 2
> +	 * (fragments + linear part + virtio header)
> +	 */
> +	for (i = 0; i < vi->max_queue_pairs; i++) {
> +		if (virtqueue_get_vring_size(vi->sq[i].vq) < min_size ||
> +		    virtqueue_get_vring_size(vi->rq[i].vq) < min_size) {
> +			dev_warn(&vi->vdev->dev,
> +				 "Transmit/Receive virtqueue vring size must be at least %u\n",
> +				 min_size);
> +			return -EINVAL;
> +		}
> +	}
> +
> +	return 0;
> +}
> +
>  #ifdef CONFIG_SYSFS
>  static ssize_t mergeable_rx_buffer_size_show(struct netdev_rx_queue *queue,
>  		char *buf)
> @@ -4056,6 +4076,10 @@ static int virtnet_probe(struct virtio_device *vdev)
>  	if (err)
>  		goto free;
>  
> +	err = virtnet_validate_vqs(vi);
> +	if (err)
> +		goto free_vqs;
> +
>  #ifdef CONFIG_SYSFS
>  	if (vi->mergeable_rx_bufs)
>  		dev->sysfs_rx_queue_group = &virtio_net_mrg_rx_group;
> -- 
> 2.34.1