Re: [PATCH net-next v5 08/15] virtio_net: sq support premapped mode

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

On Tue, 18 Jun 2024 09:00:56 +0800, Jason Wang <jasowang@redhat.com> wrote:
> On Mon, Jun 17, 2024 at 3:41 PM Xuan Zhuo <xuanzhuo@linux.alibaba.com> wrote:
> >
> > On Mon, 17 Jun 2024 14:28:05 +0800, Jason Wang <jasowang@redhat.com> wrote:
> > > On Mon, Jun 17, 2024 at 1:00 PM Jason Wang <jasowang@redhat.com> wrote:
> > > >
> > > > On Fri, Jun 14, 2024 at 2:39 PM Xuan Zhuo <xuanzhuo@linux.alibaba.com> wrote:
> > > > >
> > > > > If the xsk is enabling, the xsk tx will share the send queue.
> > > > > But the xsk requires that the send queue use the premapped mode.
> > > > > So the send queue must support premapped mode when it is bound to
> > > > > af-xdp.
> > > > >
> > > > > * virtnet_sq_set_premapped(sq, true) is used to enable premapped mode.
> > > > >
> > > > >     In this mode, the driver will record the dma info when skb or xdp
> > > > >     frame is sent.
> > > > >
> > > > >     Currently, the SQ premapped mode is operational only with af-xdp. In
> > > > >     this mode, af-xdp, the kernel stack, and xdp tx/redirect will share
> > > > >     the same SQ. Af-xdp independently manages its DMA. The kernel stack
> > > > >     and xdp tx/redirect utilize this DMA metadata to manage the DMA
> > > > >     info.
> > > > >
> > >
> > > Note that there's indeed a mode when we have exclusive XDP TX queue:
> > >
> > >         /* XDP requires extra queues for XDP_TX */
> > >         if (curr_qp + xdp_qp > vi->max_queue_pairs) {
> > >                 netdev_warn_once(dev, "XDP request %i queues but max
> > > is %i. XDP_TX and XDP_REDIRECT will operate in a slower locked tx
> > > mode.\n",
> > >                                  curr_qp + xdp_qp, vi->max_queue_pairs);
> > >                 xdp_qp = 0;
> > >         }
> > >
> > > So we need to mention how the code works in this patch.
> >
> > Sorry, I do not get it.
> >
> > Could you say more?
>
> I meant in the commit log, you said:
>
> """
> In this mode, af-xdp, the kernel stack, and xdp tx/redirect will share
> the same SQ.
> """
>
> is not correct if we have sufficient queue pairs.
>
> We need to tweak it and explain if the code can still work if we have
> exclusive XDP TX queues.


YES, it can work.

I will explain in next version.

Thanks.


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