Message-ID: <CAMB2axMXzcxrFr+zWV6CFJxDrKwH+U85F7dkeXfJjAO10EmSAg@mail.gmail.com>
Date: Mon, 29 Jul 2024 15:51:52 -0700
From: Amery Hung <ameryhung@...il.com>
To: Arseniy Krasnov <avkrasnov@...utedevices.com>
Cc: stefanha@...hat.com, sgarzare@...hat.com, mst@...hat.com, 
	jasowang@...hat.com, xuanzhuo@...ux.alibaba.com, davem@...emloft.net, 
	edumazet@...gle.com, kuba@...nel.org, pabeni@...hat.com, kys@...rosoft.com, 
	haiyangz@...rosoft.com, wei.liu@...nel.org, decui@...rosoft.com, 
	bryantan@...are.com, vdasa@...are.com, pv-drivers@...are.com, 
	dan.carpenter@...aro.org, simon.horman@...igine.com, oxffffaa@...il.com, 
	kvm@...r.kernel.org, virtualization@...ts.linux-foundation.org, 
	netdev@...r.kernel.org, linux-kernel@...r.kernel.org, 
	linux-hyperv@...r.kernel.org, bpf@...r.kernel.org, 
	bobby.eshleman@...edance.com, jiang.wang@...edance.com, 
	amery.hung@...edance.com, xiyou.wangcong@...il.com, kernel@...rdevices.ru
Subject: Re: [RFC PATCH net-next v6 07/14] virtio/vsock: add common datagram
 send path

On Mon, Jul 29, 2024 at 1:12 PM Arseniy Krasnov
<avkrasnov@...utedevices.com> wrote:
>
> Hi,
>
> > diff --git a/net/vmw_vsock/virtio_transport_common.c b/net/vmw_vsock/virtio_transport_common.c
> > index a1c76836d798..46cd1807f8e3 100644
> > --- a/net/vmw_vsock/virtio_transport_common.c
> > +++ b/net/vmw_vsock/virtio_transport_common.c
> > @@ -1040,13 +1040,98 @@ int virtio_transport_shutdown(struct vsock_sock *vsk, int mode)
> >  }
> >  EXPORT_SYMBOL_GPL(virtio_transport_shutdown);
> >
> > +static int virtio_transport_dgram_send_pkt_info(struct vsock_sock *vsk,
> > +                                             struct virtio_vsock_pkt_info *info)
> > +{
> > +     u32 src_cid, src_port, dst_cid, dst_port;
> > +     const struct vsock_transport *transport;
> > +     const struct virtio_transport *t_ops;
> > +     struct sock *sk = sk_vsock(vsk);
> > +     struct virtio_vsock_hdr *hdr;
> > +     struct sk_buff *skb;
> > +     void *payload;
> > +     int noblock = 0;
> > +     int err;
> > +
> > +     info->type = virtio_transport_get_type(sk_vsock(vsk));
> > +
> > +     if (info->pkt_len > VIRTIO_VSOCK_MAX_PKT_BUF_SIZE)
> > +             return -EMSGSIZE;
>
> Small suggestion, i think we can check for packet length earlier ? Before
> info->type = ...

Certainly.

>
> > +
> > +     transport = vsock_dgram_lookup_transport(info->remote_cid, info->remote_flags);
> > +     t_ops = container_of(transport, struct virtio_transport, transport);
> > +     if (unlikely(!t_ops))
> > +             return -EFAULT;
> > +
> > +     if (info->msg)
> > +             noblock = info->msg->msg_flags & MSG_DONTWAIT;
> > +
> > +     /* Use sock_alloc_send_skb to throttle by sk_sndbuf. This helps avoid
> > +      * triggering the OOM.
> > +      */
> > +     skb = sock_alloc_send_skb(sk, info->pkt_len + VIRTIO_VSOCK_SKB_HEADROOM,
> > +                               noblock, &err);
> > +     if (!skb)
> > +             return err;
> > +
> > +     skb_reserve(skb, VIRTIO_VSOCK_SKB_HEADROOM);
> > +
> > +     src_cid = t_ops->transport.get_local_cid();
> > +     src_port = vsk->local_addr.svm_port;
> > +     dst_cid = info->remote_cid;
> > +     dst_port = info->remote_port;
> > +
> > +     hdr = virtio_vsock_hdr(skb);
> > +     hdr->type       = cpu_to_le16(info->type);
> > +     hdr->op         = cpu_to_le16(info->op);
> > +     hdr->src_cid    = cpu_to_le64(src_cid);
> > +     hdr->dst_cid    = cpu_to_le64(dst_cid);
> > +     hdr->src_port   = cpu_to_le32(src_port);
> > +     hdr->dst_port   = cpu_to_le32(dst_port);
> > +     hdr->flags      = cpu_to_le32(info->flags);
> > +     hdr->len        = cpu_to_le32(info->pkt_len);
>
> There is function 'virtio_transport_init_hdr()' in this file, may be reuse it ?

Will do.

>
> > +
> > +     if (info->msg && info->pkt_len > 0) {
>
> If pkt_len is 0, do we really need to send such packets ? Because for connectible
> sockets, we ignore empty OP_RW packets.

Thanks for pointing this out. I think virtio dgram should also follow that.

>
> > +             payload = skb_put(skb, info->pkt_len);
> > +             err = memcpy_from_msg(payload, info->msg, info->pkt_len);
> > +             if (err)
> > +                     goto out;
> > +     }
> > +
> > +     trace_virtio_transport_alloc_pkt(src_cid, src_port,
> > +                                      dst_cid, dst_port,
> > +                                      info->pkt_len,
> > +                                      info->type,
> > +                                      info->op,
> > +                                      info->flags,
> > +                                      false);
>
> ^^^ For SOCK_DGRAM, include/trace/events/vsock_virtio_transport_common.h also should
> be updated?

Can you elaborate what needs to be changed?

Thank you,
Amery

>
> > +
> > +     return t_ops->send_pkt(skb);
> > +out:
> > +     kfree_skb(skb);
> > +     return err;
> > +}
> > +
> >  int
> >  virtio_transport_dgram_enqueue(struct vsock_sock *vsk,
> >                              struct sockaddr_vm *remote_addr,
> >                              struct msghdr *msg,
> >                              size_t dgram_len)
> >  {
> > -     return -EOPNOTSUPP;
> > +     /* Here we are only using the info struct to retain style uniformity
> > +      * and to ease future refactoring and merging.
> > +      */
> > +     struct virtio_vsock_pkt_info info = {
> > +             .op = VIRTIO_VSOCK_OP_RW,
> > +             .remote_cid = remote_addr->svm_cid,
> > +             .remote_port = remote_addr->svm_port,
> > +             .remote_flags = remote_addr->svm_flags,
> > +             .msg = msg,
> > +             .vsk = vsk,
> > +             .pkt_len = dgram_len,
> > +     };
> > +
> > +     return virtio_transport_dgram_send_pkt_info(vsk, &info);
> >  }
> >  EXPORT_SYMBOL_GPL(virtio_transport_dgram_enqueue);
> >
> > --
> > 2.20.1
>
> Thanks, Arseniy

Powered by blists - more mailing lists