Re: [PATCH net-next v6 07/10] virtio_net: xsk: rx: support fill with xsk buffer

[Xuan Zhuo <xuanzhuo@linux.alibaba.com>]

On Mon, 1 Jul 2024 11:05:33 +0800, Jason Wang <jasowang@redhat.com> wrote:
> On Fri, Jun 28, 2024 at 1:44 PM Xuan Zhuo <xuanzhuo@linux.alibaba.com> wrote:
> >
> > On Fri, 28 Jun 2024 10:19:37 +0800, Jason Wang <jasowang@redhat.com> wrote:
> > > On Tue, Jun 18, 2024 at 3:57 PM Xuan Zhuo <xuanzhuo@linux.alibaba.com> wrote:
> > > >
> > > > Implement the logic of filling rq with XSK buffers.
> > > >
> > > > Signed-off-by: Xuan Zhuo <xuanzhuo@linux.alibaba.com>
> > > > ---
> > > >  drivers/net/virtio_net.c | 68 ++++++++++++++++++++++++++++++++++++++--
> > > >  1 file changed, 66 insertions(+), 2 deletions(-)
> > > >
> > > > diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
> > > > index 2bbc715f22c6..2ac5668a94ce 100644
> > > > --- a/drivers/net/virtio_net.c
> > > > +++ b/drivers/net/virtio_net.c
> > > > @@ -355,6 +355,8 @@ struct receive_queue {
> > > >
> > > >                 /* xdp rxq used by xsk */
> > > >                 struct xdp_rxq_info xdp_rxq;
> > > > +
> > > > +               struct xdp_buff **xsk_buffs;
> > > >         } xsk;
> > > >  };
> > > >
> > > > @@ -1032,6 +1034,53 @@ static void check_sq_full_and_disable(struct virtnet_info *vi,
> > > >         }
> > > >  }
> > > >
> > > > +static void sg_fill_dma(struct scatterlist *sg, dma_addr_t addr, u32 len)
> > > > +{
> > > > +       sg->dma_address = addr;
> > > > +       sg->length = len;
> > > > +}
> > > > +
> > > > +static int virtnet_add_recvbuf_xsk(struct virtnet_info *vi, struct receive_queue *rq,
> > > > +                                  struct xsk_buff_pool *pool, gfp_t gfp)
> > > > +{
> > > > +       struct xdp_buff **xsk_buffs;
> > > > +       dma_addr_t addr;
> > > > +       u32 len, i;
> > > > +       int err = 0;
> > > > +       int num;
> > > > +
> > > > +       xsk_buffs = rq->xsk.xsk_buffs;
> > > > +
> > > > +       num = xsk_buff_alloc_batch(pool, xsk_buffs, rq->vq->num_free);
> > > > +       if (!num)
> > > > +               return -ENOMEM;
> > > > +
> > > > +       len = xsk_pool_get_rx_frame_size(pool) + vi->hdr_len;
> > > > +
> > > > +       for (i = 0; i < num; ++i) {
> > > > +               /* use the part of XDP_PACKET_HEADROOM as the virtnet hdr space */
> > > > +               addr = xsk_buff_xdp_get_dma(xsk_buffs[i]) - vi->hdr_len;
> > >
> > > We had VIRTIO_XDP_HEADROOM, can we reuse it? Or if it's redundant
> > > let's send a patch to switch to XDP_PACKET_HEADROOM.
> >
> > Do you mean replace it inside the comment?
>
> I meant a patch to s/VIRTIO_XDP_HEADROOM/XDP_PACKET_HEADROOM/g.

I see.


>
> >
> > I want to describe use the headroom of xsk, the size of the headroom is
> > XDP_PACKET_HEADROOM.
> >
> > >
> > > Btw, the code assumes vi->hdr_len < xsk_pool_get_headroom(). It's
> > > better to fail if it's not true when enabling xsk.
> >
> > It is ok.
>
> I mean do we need a check to fail xsk binding if vi->hdr_len >
> xsk_pool_get_headroom() or it has been guaranteed by the code already.

YES.

Thanks.


>
> Thanks
>
> >
> > Thanks.
> >
> >
> > >
> > > Thanks
> > >
> >
>