Date:   Mon, 6 Mar 2023 13:07:42 +0100
From:   Stefano Garzarella <sgarzare@redhat.com>
To:     Arseniy Krasnov <avkrasnov@sberdevices.ru>
Cc:     Stefan Hajnoczi <stefanha@redhat.com>,
        "David S. Miller" <davem@davemloft.net>,
        Eric Dumazet <edumazet@google.com>,
        Jakub Kicinski <kuba@kernel.org>,
        Paolo Abeni <pabeni@redhat.com>,
        Bobby Eshleman <bobby.eshleman@bytedance.com>,
        kvm@vger.kernel.org, virtualization@lists.linux-foundation.org,
        netdev@vger.kernel.org, linux-kernel@vger.kernel.org,
        kernel@sberdevices.ru, oxffffaa@gmail.com
Subject: Re: [RFC PATCH v2 3/4] virtio/vsock: free skb on data copy failure
Message-ID: <20230306120742.v6ss4w22ku7pe45a@sgarzare-redhat>
References: <a7ab414b-5e41-c7b6-250b-e8401f335859@sberdevices.ru>
 <ef98aad4-f86d-fe60-9a35-792363a78a68@sberdevices.ru>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Disposition: inline
In-Reply-To: <ef98aad4-f86d-fe60-9a35-792363a78a68@sberdevices.ru>
Precedence: bulk

On Sun, Mar 05, 2023 at 11:08:38PM +0300, Arseniy Krasnov wrote:
>This fixes two things in case when 'memcpy_to_msg()' fails:
>1) Update credit parameters of the socket, like this skbuff was
>   copied to user successfully. This is needed because when skbuff was
>   received it's length was used to update 'rx_bytes', thus when we drop
>   skbuff here, we must account rest of it's data in 'rx_bytes'.
>2) Free skbuff which was removed from socket's queue.
>
>Fixes: 71dc9ec9ac7d ("virtio/vsock: replace virtio_vsock_pkt with sk_buff")
>Signed-off-by: Arseniy Krasnov <AVKrasnov@sberdevices.ru>
>---
> net/vmw_vsock/virtio_transport_common.c | 6 +++++-
> 1 file changed, 5 insertions(+), 1 deletion(-)
>
>diff --git a/net/vmw_vsock/virtio_transport_common.c b/net/vmw_vsock/virtio_transport_common.c
>index 30b0539990ba..ffb1af4f2b52 100644
>--- a/net/vmw_vsock/virtio_transport_common.c
>+++ b/net/vmw_vsock/virtio_transport_common.c
>@@ -379,8 +379,12 @@ virtio_transport_stream_do_dequeue(struct vsock_sock *vsk,
> 		spin_unlock_bh(&vvs->rx_lock);
>
> 		err = memcpy_to_msg(msg, skb->data, bytes);
>-		if (err)
>+		if (err) {
>+			skb_pull(skb, skb->len);
>+			virtio_transport_dec_rx_pkt(vvs, skb);
>+			consume_skb(skb);

I'm not sure it's the right thing to do, if we fail to copy the content
into the user's buffer, I think we should queue it again.

In fact, before commit 71dc9ec9ac7d ("virtio/vsock: replace
virtio_vsock_pkt with sk_buff"), we used to remove the packet from the
rx_queue, only if memcpy_to_msg() was successful.

Maybe it is better to do as we did before and use skb_peek() at the
beginning of the loop and __skb_unlink() when skb->len == 0.

Thanks,
Stefano

> 			goto out;
>+		}
>
> 		spin_lock_bh(&vvs->rx_lock);
>
>-- 
>2.25.1
>