Message-ID: <CACGkMEsyP7bxOchyaKPb=y+td=1F34NwxxP3atyNBwFAtNOsxw@mail.gmail.com>
Date: Tue, 27 Jun 2023 16:03:35 +0800
From: Jason Wang <jasowang@...hat.com>
To: Xuan Zhuo <xuanzhuo@...ux.alibaba.com>
Cc: virtualization@...ts.linux-foundation.org, 
	"Michael S. Tsirkin" <mst@...hat.com>, "David S. Miller" <davem@...emloft.net>, Eric Dumazet <edumazet@...gle.com>, 
	Jakub Kicinski <kuba@...nel.org>, Paolo Abeni <pabeni@...hat.com>, Alexei Starovoitov <ast@...nel.org>, 
	Daniel Borkmann <daniel@...earbox.net>, Jesper Dangaard Brouer <hawk@...nel.org>, 
	John Fastabend <john.fastabend@...il.com>, netdev@...r.kernel.org, bpf@...r.kernel.org
Subject: Re: [PATCH vhost v10 10/10] virtio_net: support dma premapped

On Fri, Jun 2, 2023 at 5:22 PM Xuan Zhuo <xuanzhuo@...ux.alibaba.com> wrote:
>
> Introduce the module param "experiment_premapped" to enable the function
> that the virtio-net do dma mapping.
>
> If that is true, the vq of virtio-net is under the premapped mode.
> It just handle the sg with dma_address. And the driver must get the dma
> address of the buffer to unmap after get the buffer from virtio core.
>
> That will be useful when AF_XDP is enable, AF_XDP tx and the kernel packet
> xmit will share the tx queue, so the skb xmit must support the premapped
> mode.
>
> Signed-off-by: Xuan Zhuo <xuanzhuo@...ux.alibaba.com>
> ---
>  drivers/net/virtio_net.c | 163 +++++++++++++++++++++++++++++++++------
>  1 file changed, 141 insertions(+), 22 deletions(-)
>
> diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
> index 2396c28c0122..5898212fcb3c 100644
> --- a/drivers/net/virtio_net.c
> +++ b/drivers/net/virtio_net.c
> @@ -26,10 +26,11 @@
>  static int napi_weight = NAPI_POLL_WEIGHT;
>  module_param(napi_weight, int, 0444);
>
> -static bool csum = true, gso = true, napi_tx = true;
> +static bool csum = true, gso = true, napi_tx = true, experiment_premapped;
>  module_param(csum, bool, 0444);
>  module_param(gso, bool, 0444);
>  module_param(napi_tx, bool, 0644);
> +module_param(experiment_premapped, bool, 0644);

Having a module parameter is sub-optimal. I think we can demonstrate
real benefit:

In the case of a merge rx buffer, if the mapping is done by the
virtio-core, it needs to be done per buffer (< PAGE_SIZE).

But if it is done by the virtio-net, we have a chance to map the
buffer per page. Which can save a lot of mappings and unmapping. A lot
of other optimizations could be done on top as well.

If we manage to prove this, we don't need any experimental module
parameters at all.

Thanks


>
>  /* FIXME: MTU in config. */
>  #define GOOD_PACKET_LEN (ETH_HLEN + VLAN_HLEN + ETH_DATA_LEN)
> @@ -142,6 +143,9 @@ struct send_queue {
>
>         /* Record whether sq is in reset state. */
>         bool reset;
> +
> +       /* The vq is premapped mode. */
> +       bool premapped;
>  };
>
>  /* Internal representation of a receive virtqueue */
> @@ -174,6 +178,9 @@ struct receive_queue {
>         char name[16];
>
>         struct xdp_rxq_info xdp_rxq;
> +
> +       /* The vq is premapped mode. */
> +       bool premapped;
>  };
>
>  /* This structure can contain rss message with maximum settings for indirection table and keysize
> @@ -546,6 +553,105 @@ static struct sk_buff *page_to_skb(struct virtnet_info *vi,
>         return skb;
>  }
>
> +static int virtnet_generic_unmap(struct virtqueue *vq, struct virtqueue_detach_cursor *cursor)
> +{
> +       enum dma_data_direction dir;
> +       dma_addr_t addr;
> +       u32 len;
> +       int err;
> +
> +       do {
> +               err = virtqueue_detach(vq, cursor, &addr, &len, &dir);
> +               if (!err || err == -EAGAIN)
> +                       dma_unmap_page_attrs(virtqueue_dma_dev(vq), addr, len, dir, 0);
> +
> +       } while (err == -EAGAIN);
> +
> +       return err;
> +}
> +
> +static void *virtnet_detach_unused_buf(struct virtqueue *vq, bool premapped)
> +{
> +       struct virtqueue_detach_cursor cursor;
> +       void *buf;
> +
> +       if (!premapped)
> +               return virtqueue_detach_unused_buf(vq);
> +
> +       buf = virtqueue_detach_unused_buf_premapped(vq, &cursor);
> +       if (buf)
> +               virtnet_generic_unmap(vq, &cursor);
> +
> +       return buf;
> +}
> +
> +static void *virtnet_get_buf_ctx(struct virtqueue *vq, bool premapped, u32 *len, void **ctx)
> +{
> +       struct virtqueue_detach_cursor cursor;
> +       void *buf;
> +
> +       if (!premapped)
> +               return virtqueue_get_buf_ctx(vq, len, ctx);
> +
> +       buf = virtqueue_get_buf_premapped(vq, len, ctx, &cursor);
> +       if (buf)
> +               virtnet_generic_unmap(vq, &cursor);
> +
> +       return buf;
> +}
> +
> +#define virtnet_rq_get_buf(rq, plen, pctx) \
> +({ \
> +       typeof(rq) _rq = (rq); \
> +       virtnet_get_buf_ctx(_rq->vq, _rq->premapped, plen, pctx); \
> +})
> +
> +#define virtnet_sq_get_buf(sq, plen, pctx) \
> +({ \
> +       typeof(sq) _sq = (sq); \
> +       virtnet_get_buf_ctx(_sq->vq, _sq->premapped, plen, pctx); \
> +})
> +
> +static int virtnet_add_sg(struct virtqueue *vq, bool premapped,
> +                         struct scatterlist *sg, unsigned int num, bool out,
> +                         void *data, void *ctx, gfp_t gfp)
> +{
> +       enum dma_data_direction dir;
> +       struct device *dev;
> +       int err, ret;
> +
> +       if (!premapped)
> +               return virtqueue_add_sg(vq, sg, num, out, data, ctx, gfp);
> +
> +       dir = out ? DMA_TO_DEVICE : DMA_FROM_DEVICE;
> +       dev = virtqueue_dma_dev(vq);
> +
> +       ret = dma_map_sg_attrs(dev, sg, num, dir, 0);
> +       if (ret != num)
> +               goto err;
> +
> +       err = virtqueue_add_sg(vq, sg, num, out, data, ctx, gfp);
> +       if (err < 0)
> +               goto err;
> +
> +       return 0;
> +
> +err:
> +       dma_unmap_sg_attrs(dev, sg, num, dir, 0);
> +       return -ENOMEM;
> +}
> +
> +static int virtnet_add_outbuf(struct send_queue *sq, unsigned int num, void *data)
> +{
> +       return virtnet_add_sg(sq->vq, sq->premapped, sq->sg, num, true, data, NULL, GFP_ATOMIC);
> +}
> +
> +static int virtnet_add_inbuf(struct receive_queue *rq, unsigned int num, void *data,
> +                            void *ctx, gfp_t gfp)
> +{
> +       return virtnet_add_sg(rq->vq, rq->premapped, rq->sg, num, false, data, ctx, gfp);
> +}
> +
>  static void free_old_xmit_skbs(struct send_queue *sq, bool in_napi)
>  {
>         unsigned int len;
> @@ -553,7 +659,7 @@ static void free_old_xmit_skbs(struct send_queue *sq, bool in_napi)
>         unsigned int bytes = 0;
>         void *ptr;
>
> -       while ((ptr = virtqueue_get_buf(sq->vq, &len)) != NULL) {
> +       while ((ptr = virtnet_sq_get_buf(sq, &len, NULL)) != NULL) {
>                 if (likely(!is_xdp_frame(ptr))) {
>                         struct sk_buff *skb = ptr;
>
> @@ -667,8 +773,7 @@ static int __virtnet_xdp_xmit_one(struct virtnet_info *vi,
>                             skb_frag_size(frag), skb_frag_off(frag));
>         }
>
> -       err = virtqueue_add_outbuf(sq->vq, sq->sg, nr_frags + 1,
> -                                  xdp_to_ptr(xdpf), GFP_ATOMIC);
> +       err = virtnet_add_outbuf(sq, nr_frags + 1, xdp_to_ptr(xdpf));
>         if (unlikely(err))
>                 return -ENOSPC; /* Caller handle free/refcnt */
>
> @@ -744,7 +849,7 @@ static int virtnet_xdp_xmit(struct net_device *dev,
>         }
>
>         /* Free up any pending old buffers before queueing new ones. */
> -       while ((ptr = virtqueue_get_buf(sq->vq, &len)) != NULL) {
> +       while ((ptr = virtnet_sq_get_buf(sq, &len, NULL)) != NULL) {
>                 if (likely(is_xdp_frame(ptr))) {
>                         struct xdp_frame *frame = ptr_to_xdp(ptr);
>
> @@ -828,7 +933,7 @@ static struct page *xdp_linearize_page(struct receive_queue *rq,
>                 void *buf;
>                 int off;
>
> -               buf = virtqueue_get_buf(rq->vq, &buflen);
> +               buf = virtnet_rq_get_buf(rq, &buflen, NULL);
>                 if (unlikely(!buf))
>                         goto err_buf;
>
> @@ -1119,7 +1224,7 @@ static int virtnet_build_xdp_buff_mrg(struct net_device *dev,
>                 return -EINVAL;
>
>         while (--*num_buf > 0) {
> -               buf = virtqueue_get_buf_ctx(rq->vq, &len, &ctx);
> +               buf = virtnet_rq_get_buf(rq, &len, &ctx);
>                 if (unlikely(!buf)) {
>                         pr_debug("%s: rx error: %d buffers out of %d missing\n",
>                                  dev->name, *num_buf,
> @@ -1344,7 +1449,7 @@ static struct sk_buff *receive_mergeable(struct net_device *dev,
>         while (--num_buf) {
>                 int num_skb_frags;
>
> -               buf = virtqueue_get_buf_ctx(rq->vq, &len, &ctx);
> +               buf = virtnet_rq_get_buf(rq, &len, &ctx);
>                 if (unlikely(!buf)) {
>                         pr_debug("%s: rx error: %d buffers out of %d missing\n",
>                                  dev->name, num_buf,
> @@ -1407,7 +1512,7 @@ static struct sk_buff *receive_mergeable(struct net_device *dev,
>  err_skb:
>         put_page(page);
>         while (num_buf-- > 1) {
> -               buf = virtqueue_get_buf(rq->vq, &len);
> +               buf = virtnet_rq_get_buf(rq, &len, NULL);
>                 if (unlikely(!buf)) {
>                         pr_debug("%s: rx error: %d buffers missing\n",
>                                  dev->name, num_buf);
> @@ -1534,7 +1639,7 @@ static int add_recvbuf_small(struct virtnet_info *vi, struct receive_queue *rq,
>         alloc_frag->offset += len;
>         sg_init_one(rq->sg, buf + VIRTNET_RX_PAD + xdp_headroom,
>                     vi->hdr_len + GOOD_PACKET_LEN);
> -       err = virtqueue_add_inbuf_ctx(rq->vq, rq->sg, 1, buf, ctx, gfp);
> +       err = virtnet_add_inbuf(rq, 1, buf, ctx, gfp);
>         if (err < 0)
>                 put_page(virt_to_head_page(buf));
>         return err;
> @@ -1581,8 +1686,8 @@ static int add_recvbuf_big(struct virtnet_info *vi, struct receive_queue *rq,
>
>         /* chain first in list head */
>         first->private = (unsigned long)list;
> -       err = virtqueue_add_inbuf(rq->vq, rq->sg, vi->big_packets_num_skbfrags + 2,
> -                                 first, gfp);
> +       err = virtnet_add_inbuf(rq, vi->big_packets_num_skbfrags + 2,
> +                               first, NULL, gfp);
>         if (err < 0)
>                 give_pages(rq, first);
>
> @@ -1645,7 +1750,7 @@ static int add_recvbuf_mergeable(struct virtnet_info *vi,
>
>         sg_init_one(rq->sg, buf, len);
>         ctx = mergeable_len_to_ctx(len + room, headroom);
> -       err = virtqueue_add_inbuf_ctx(rq->vq, rq->sg, 1, buf, ctx, gfp);
> +       err = virtnet_add_inbuf(rq, 1, buf, ctx, gfp);
>         if (err < 0)
>                 put_page(virt_to_head_page(buf));
>
> @@ -1768,13 +1873,13 @@ static int virtnet_receive(struct receive_queue *rq, int budget,
>                 void *ctx;
>
>                 while (stats.packets < budget &&
> -                      (buf = virtqueue_get_buf_ctx(rq->vq, &len, &ctx))) {
> +                      (buf = virtnet_rq_get_buf(rq, &len, &ctx))) {
>                         receive_buf(vi, rq, buf, len, ctx, xdp_xmit, &stats);
>                         stats.packets++;
>                 }
>         } else {
>                 while (stats.packets < budget &&
> -                      (buf = virtqueue_get_buf(rq->vq, &len)) != NULL) {
> +                      (buf = virtnet_rq_get_buf(rq, &len, NULL)) != NULL) {
>                         receive_buf(vi, rq, buf, len, NULL, xdp_xmit, &stats);
>                         stats.packets++;
>                 }
> @@ -1984,7 +2089,7 @@ static int xmit_skb(struct send_queue *sq, struct sk_buff *skb)
>                         return num_sg;
>                 num_sg++;
>         }
> -       return virtqueue_add_outbuf(sq->vq, sq->sg, num_sg, skb, GFP_ATOMIC);
> +       return virtnet_add_outbuf(sq, num_sg, skb);
>  }
>
>  static netdev_tx_t start_xmit(struct sk_buff *skb, struct net_device *dev)
> @@ -3552,15 +3657,17 @@ static void free_unused_bufs(struct virtnet_info *vi)
>         int i;
>
>         for (i = 0; i < vi->max_queue_pairs; i++) {
> -               struct virtqueue *vq = vi->sq[i].vq;
> -               while ((buf = virtqueue_detach_unused_buf(vq)) != NULL)
> -                       virtnet_sq_free_unused_buf(vq, buf);
> +               struct send_queue *sq = &vi->sq[i];
> +
> +               while ((buf = virtnet_detach_unused_buf(sq->vq, sq->premapped)) != NULL)
> +                       virtnet_sq_free_unused_buf(sq->vq, buf);
>         }
>
>         for (i = 0; i < vi->max_queue_pairs; i++) {
> -               struct virtqueue *vq = vi->rq[i].vq;
> -               while ((buf = virtqueue_detach_unused_buf(vq)) != NULL)
> -                       virtnet_rq_free_unused_buf(vq, buf);
> +               struct receive_queue *rq = &vi->rq[i];
> +
> +               while ((buf = virtnet_detach_unused_buf(rq->vq, rq->premapped)) != NULL)
> +                       virtnet_rq_free_unused_buf(rq->vq, buf);
>         }
>  }
>
> @@ -3658,6 +3765,18 @@ static int virtnet_find_vqs(struct virtnet_info *vi)
>                 vi->rq[i].vq = vqs[rxq2vq(i)];
>                 vi->rq[i].min_buf_len = mergeable_min_buf_len(vi, vi->rq[i].vq);
>                 vi->sq[i].vq = vqs[txq2vq(i)];
> +
> +               if (experiment_premapped) {
> +                       if (!virtqueue_set_premapped(vi->rq[i].vq))
> +                               vi->rq[i].premapped = true;
> +                       else
> +                               netdev_warn(vi->dev, "RXQ (%d) enable premapped failure.\n", i);
> +
> +                       if (!virtqueue_set_premapped(vi->sq[i].vq))
> +                               vi->sq[i].premapped = true;
> +                       else
> +                               netdev_warn(vi->dev, "TXQ (%d) enable premapped failure.\n", i);
> +               }
>         }
>
>         /* run here: ret == 0. */
> --
> 2.32.0.3.g01195cf9f
>

Powered by blists - more mailing lists