[RFC,0/9] virtio_net: support multi buffer xdp

Message

Heng Qi Nov. 22, 2022, 7:43 a.m. UTC

Currently, virtio net only supports xdp for single-buffer packets
or linearized multi-buffer packets. This patchset supports xdp for
multi-buffer packets, then GRO_HW related features can be
negotiated, and do not affect the processing of single-buffer xdp.

In order to build multi-buffer xdp neatly, we integrated the code
into virtnet_build_xdp_buff() for xdp. The first buffer is used
for prepared xdp buff, and the rest of the buffers are added to
its skb_shared_info structure. This structure can also be
conveniently converted during XDP_PASS to get the corresponding skb.

Since virtio net uses comp pages, and bpf_xdp_frags_increase_tail()
is based on the assumption of the page pool,
(rxq->frag_size - skb_frag_size(frag) - skb_frag_off(frag))
is negative in most cases. So we didn't set xdp_rxq->frag_size in
virtnet_open() to disable the tail increase.

Heng Qi (9):
  virtio_net: disable the hole mechanism for xdp
  virtio_net: set up xdp for multi buffer packets
  virtio_net: update bytes calculation for xdp_frame
  virtio_net: remove xdp related info from page_to_skb()
  virtio_net: build xdp_buff with multi buffers
  virtio_net: construct multi-buffer xdp in mergeable
  virtio_net: build skb from multi-buffer xdp
  virtio_net: transmit the multi-buffer xdp
  virtio_net: support multi-buffer xdp

 drivers/net/virtio_net.c | 356 ++++++++++++++++++++++++---------------
 1 file changed, 219 insertions(+), 137 deletions(-)

Comments

Heng Qi Dec. 2, 2022, 4:50 a.m. UTC | #1

Hi, Jason.

Do you have any comments on this series?

Thanks.

在 2022/11/22 下午3:43, Heng Qi 写道:
> Currently, virtio net only supports xdp for single-buffer packets
> or linearized multi-buffer packets. This patchset supports xdp for
> multi-buffer packets, then GRO_HW related features can be
> negotiated, and do not affect the processing of single-buffer xdp.
>
> In order to build multi-buffer xdp neatly, we integrated the code
> into virtnet_build_xdp_buff() for xdp. The first buffer is used
> for prepared xdp buff, and the rest of the buffers are added to
> its skb_shared_info structure. This structure can also be
> conveniently converted during XDP_PASS to get the corresponding skb.
>
> Since virtio net uses comp pages, and bpf_xdp_frags_increase_tail()
> is based on the assumption of the page pool,
> (rxq->frag_size - skb_frag_size(frag) - skb_frag_off(frag))
> is negative in most cases. So we didn't set xdp_rxq->frag_size in
> virtnet_open() to disable the tail increase.
>
> Heng Qi (9):
>    virtio_net: disable the hole mechanism for xdp
>    virtio_net: set up xdp for multi buffer packets
>    virtio_net: update bytes calculation for xdp_frame
>    virtio_net: remove xdp related info from page_to_skb()
>    virtio_net: build xdp_buff with multi buffers
>    virtio_net: construct multi-buffer xdp in mergeable
>    virtio_net: build skb from multi-buffer xdp
>    virtio_net: transmit the multi-buffer xdp
>    virtio_net: support multi-buffer xdp
>
>   drivers/net/virtio_net.c | 356 ++++++++++++++++++++++++---------------
>   1 file changed, 219 insertions(+), 137 deletions(-)
>

Jason Wang Dec. 2, 2022, 5:41 a.m. UTC | #2

On Fri, Dec 2, 2022 at 12:50 PM Heng Qi <[e-mail]> wrote:
>
> Hi, Jason.
>
> Do you have any comments on this series?

-EBUSY this week, but I will review it next week for sure.

Thanks

>
> Thanks.
>
> 在 2022/11/22 下午3:43, Heng Qi 写道:
> > Currently, virtio net only supports xdp for single-buffer packets
> > or linearized multi-buffer packets. This patchset supports xdp for
> > multi-buffer packets, then GRO_HW related features can be
> > negotiated, and do not affect the processing of single-buffer xdp.
> >
> > In order to build multi-buffer xdp neatly, we integrated the code
> > into virtnet_build_xdp_buff() for xdp. The first buffer is used
> > for prepared xdp buff, and the rest of the buffers are added to
> > its skb_shared_info structure. This structure can also be
> > conveniently converted during XDP_PASS to get the corresponding skb.
> >
> > Since virtio net uses comp pages, and bpf_xdp_frags_increase_tail()
> > is based on the assumption of the page pool,
> > (rxq->frag_size - skb_frag_size(frag) - skb_frag_off(frag))
> > is negative in most cases. So we didn't set xdp_rxq->frag_size in
> > virtnet_open() to disable the tail increase.
> >
> > Heng Qi (9):
> >    virtio_net: disable the hole mechanism for xdp
> >    virtio_net: set up xdp for multi buffer packets
> >    virtio_net: update bytes calculation for xdp_frame
> >    virtio_net: remove xdp related info from page_to_skb()
> >    virtio_net: build xdp_buff with multi buffers
> >    virtio_net: construct multi-buffer xdp in mergeable
> >    virtio_net: build skb from multi-buffer xdp
> >    virtio_net: transmit the multi-buffer xdp
> >    virtio_net: support multi-buffer xdp
> >
> >   drivers/net/virtio_net.c | 356 ++++++++++++++++++++++++---------------
> >   1 file changed, 219 insertions(+), 137 deletions(-)
> >
>