Messages in this thread | | | Subject | Re: [RFC PATCH net-next 10/12] vhost_net: build xdp buff | From | Jason Wang <> | Date | Tue, 22 May 2018 20:41:09 +0800 |
| |
On 2018年05月22日 00:56, Jesse Brandeburg wrote: > On Mon, 21 May 2018 17:04:31 +0800 Jason wrote: >> This patch implement build XDP buffers in vhost_net. The idea is do >> userspace copy in vhost_net and build XDP buff based on the >> page. Vhost_net can then submit one or an array of XDP buffs to >> underlayer socket (e.g TUN). TUN can choose to do XDP or call >> build_skb() to build skb. To support build skb, vnet header were also >> stored into the header of the XDP buff. >> >> This userspace copy and XDP buffs building is key to achieve XDP >> batching in TUN, since TUN does not need to care about userspace copy >> and then can disable premmption for several XDP buffs to achieve >> batching from XDP. >> >> TODO: reserve headroom based on the TUN XDP. >> >> Signed-off-by: Jason Wang <jasowang@redhat.com> >> --- >> drivers/vhost/net.c | 74 +++++++++++++++++++++++++++++++++++++++++++++++++++++ >> 1 file changed, 74 insertions(+) >> >> diff --git a/drivers/vhost/net.c b/drivers/vhost/net.c >> index f0639d7..1209e84 100644 >> --- a/drivers/vhost/net.c >> +++ b/drivers/vhost/net.c >> @@ -492,6 +492,80 @@ static bool vhost_has_more_pkts(struct vhost_net *net, >> likely(!vhost_exceeds_maxpend(net)); >> } >> >> +#define VHOST_NET_HEADROOM 256 >> +#define VHOST_NET_RX_PAD (NET_IP_ALIGN + NET_SKB_PAD) >> + >> +static int vhost_net_build_xdp(struct vhost_net_virtqueue *nvq, >> + struct iov_iter *from, >> + struct xdp_buff *xdp) >> +{ >> + struct vhost_virtqueue *vq = &nvq->vq; >> + struct page_frag *alloc_frag = ¤t->task_frag; >> + struct virtio_net_hdr *gso; >> + size_t len = iov_iter_count(from); >> + int buflen = SKB_DATA_ALIGN(sizeof(struct skb_shared_info)); >> + int pad = SKB_DATA_ALIGN(VHOST_NET_RX_PAD + VHOST_NET_HEADROOM >> + + nvq->sock_hlen); >> + int sock_hlen = nvq->sock_hlen; >> + void *buf; >> + int copied; >> + >> + if (len < nvq->sock_hlen) >> + return -EFAULT; >> + >> + if (SKB_DATA_ALIGN(len + pad) + >> + SKB_DATA_ALIGN(sizeof(struct skb_shared_info)) > PAGE_SIZE) >> + return -ENOSPC; >> + >> + buflen += SKB_DATA_ALIGN(len + pad); > maybe store the result of SKB_DATA_ALIGN in a local instead of doing > the work twice?
Ok.
> >> + alloc_frag->offset = ALIGN((u64)alloc_frag->offset, SMP_CACHE_BYTES); >> + if (unlikely(!skb_page_frag_refill(buflen, alloc_frag, GFP_KERNEL))) >> + return -ENOMEM; >> + >> + buf = (char *)page_address(alloc_frag->page) + alloc_frag->offset; >> + >> + /* We store two kinds of metadata in the header which will be >> + * used for XDP_PASS to do build_skb(): >> + * offset 0: buflen >> + * offset sizeof(int): vnet header >> + */ >> + copied = copy_page_from_iter(alloc_frag->page, >> + alloc_frag->offset + sizeof(int), sock_hlen, from); >> + if (copied != sock_hlen) >> + return -EFAULT; >> + >> + gso = (struct virtio_net_hdr *)(buf + sizeof(int)); >> + >> + if ((gso->flags & VIRTIO_NET_HDR_F_NEEDS_CSUM) && >> + vhost16_to_cpu(vq, gso->csum_start) + >> + vhost16_to_cpu(vq, gso->csum_offset) + 2 > >> + vhost16_to_cpu(vq, gso->hdr_len)) { >> + gso->hdr_len = cpu_to_vhost16(vq, >> + vhost16_to_cpu(vq, gso->csum_start) + >> + vhost16_to_cpu(vq, gso->csum_offset) + 2); >> + >> + if (vhost16_to_cpu(vq, gso->hdr_len) > len) >> + return -EINVAL; >> + } >> + >> + len -= sock_hlen; >> + copied = copy_page_from_iter(alloc_frag->page, >> + alloc_frag->offset + pad, >> + len, from); >> + if (copied != len) >> + return -EFAULT; >> + >> + xdp->data_hard_start = buf; >> + xdp->data = buf + pad; >> + xdp->data_end = xdp->data + len; >> + *(int *)(xdp->data_hard_start)= buflen; > space before =
Yes.
Thanks
> >> + >> + get_page(alloc_frag->page); >> + alloc_frag->offset += buflen; >> + >> + return 0; >> +} >> + >> static void handle_tx_copy(struct vhost_net *net) >> { >> struct vhost_net_virtqueue *nvq = &net->vqs[VHOST_NET_VQ_TX];
| |