Stefano Garzarella
2023-Mar-09 16:27 UTC
[RFC PATCH v3 1/4] virtio/vsock: don't use skbuff state to account credit
On Thu, Mar 09, 2023 at 01:11:46PM +0300, Arseniy Krasnov wrote:>This replaces use of skbuff state to calculate new 'rx_bytes'/'fwd_cnt' >values with explicit value as input argument. This makes code more >simple, because it is not needed to change skbuff state before each >call to update 'rx_bytes'/'fwd_cnt'.I think we should also describe the issues you found that we are fixinig now, for example the wrong calculation in virtio_transport_dec_rx_pkt(). Something like this: `skb->len` can vary when we partially read the data, this complicates the calculation of credit to be updated in virtio_transport_inc_rx_pkt()/virtio_transport_dec_rx_pkt(). Also in virtio_transport_dec_rx_pkt() we were miscalculating the credit since `skb->len` was redundant. For these reasons, let's replace the use ... (continue with what is written in this commit message) And we should add the Fixes tag: Fixes: 71dc9ec9ac7d ("virtio/vsock: replace virtio_vsock_pkt with sk_buff")> >Signed-off-by: Arseniy Krasnov <AVKrasnov at sberdevices.ru> >--- > net/vmw_vsock/virtio_transport_common.c | 23 +++++++++++------------ > 1 file changed, 11 insertions(+), 12 deletions(-) > >diff --git a/net/vmw_vsock/virtio_transport_common.c b/net/vmw_vsock/virtio_transport_common.c >index a1581c77cf84..618680fd9906 100644 >--- a/net/vmw_vsock/virtio_transport_common.c >+++ b/net/vmw_vsock/virtio_transport_common.c >@@ -241,21 +241,18 @@ static int virtio_transport_send_pkt_info(struct vsock_sock *vsk, > } > > static bool virtio_transport_inc_rx_pkt(struct virtio_vsock_sock *vvs, >- struct sk_buff *skb) >+ u32 len) > { >- if (vvs->rx_bytes + skb->len > vvs->buf_alloc) >+ if (vvs->rx_bytes + len > vvs->buf_alloc) > return false; > >- vvs->rx_bytes += skb->len; >+ vvs->rx_bytes += len; > return true; > } > > static void virtio_transport_dec_rx_pkt(struct virtio_vsock_sock *vvs, >- struct sk_buff *skb) >+ u32 len) > { >- int len; >- >- len = skb_headroom(skb) - sizeof(struct virtio_vsock_hdr) - skb->len; > vvs->rx_bytes -= len; > vvs->fwd_cnt += len; > } >@@ -388,7 +385,9 @@ virtio_transport_stream_do_dequeue(struct vsock_sock *vsk, > skb_pull(skb, bytes); > > if (skb->len == 0) { >- virtio_transport_dec_rx_pkt(vvs, skb); >+ u32 pkt_len = le32_to_cpu(virtio_vsock_hdr(skb)->len);Good catch! In my proposal I used `bytes` wrongly! The rest LGTM! Stefano>+ >+ virtio_transport_dec_rx_pkt(vvs, pkt_len); > consume_skb(skb); > } else { > __skb_queue_head(&vvs->rx_queue, skb); >@@ -437,17 +436,17 @@ static int virtio_transport_seqpacket_do_dequeue(struct vsock_sock *vsk, > > while (!msg_ready) { > struct virtio_vsock_hdr *hdr; >+ size_t pkt_len; > > skb = __skb_dequeue(&vvs->rx_queue); > if (!skb) > break; > hdr = virtio_vsock_hdr(skb); >+ pkt_len = (size_t)le32_to_cpu(hdr->len); > > if (dequeued_len >= 0) { >- size_t pkt_len; > size_t bytes_to_copy; > >- pkt_len = (size_t)le32_to_cpu(hdr->len); > bytes_to_copy = min(user_buf_len, pkt_len); > > if (bytes_to_copy) { >@@ -484,7 +483,7 @@ static int virtio_transport_seqpacket_do_dequeue(struct vsock_sock *vsk, > msg->msg_flags |= MSG_EOR; > } > >- virtio_transport_dec_rx_pkt(vvs, skb); >+ virtio_transport_dec_rx_pkt(vvs, pkt_len); > kfree_skb(skb); > } > >@@ -1040,7 +1039,7 @@ virtio_transport_recv_enqueue(struct vsock_sock *vsk, > > spin_lock_bh(&vvs->rx_lock); > >- can_enqueue = virtio_transport_inc_rx_pkt(vvs, skb); >+ can_enqueue = virtio_transport_inc_rx_pkt(vvs, len); > if (!can_enqueue) { > free_pkt = true; > goto out; >-- >2.25.1 >