[PATCH net-next v2 00/14] virtio_net: refactor xdp codes

[PATCH net-next v2 00/14] virtio_net: refactor xdp codes

[Xuan Zhuo]

Due to historical reasons, the implementation of XDP in virtio-net is relatively
chaotic. For example, the processing of XDP actions has two copies of similar
code. Such as page, xdp_page processing, etc.

The purpose of this patch set is to refactor these code. Reduce the difficulty
of subsequent maintenance. Subsequent developers will not introduce new bugs
because of some complex logical relationships.

In addition, the supporting to AF_XDP that I want to submit later will also need
to reuse the logic of XDP, such as the processing of actions, I don't want to
introduce a new similar code. In this way, I can reuse these codes in the
future.

Please review.

Thanks.

v2:
    1. re-split to make review more convenient

v1:
    1. fix some variables are uninitialized

Xuan Zhuo (14):
  virtio_net: mergeable xdp: put old page immediately
  virtio_net: introduce mergeable_xdp_prepare()
  virtio_net: optimize mergeable_xdp_prepare()
  virtio_net: introduce virtnet_xdp_handler() to seprate the logic of
    run xdp
  virtio_net: introduce xdp res enums
  virtio_net: separate the logic of freeing xdp shinfo
  virtio_net: separate the logic of freeing the rest mergeable buf
  virtio_net: auto release xdp shinfo
  virtio_net: introduce receive_mergeable_xdp()
  virtio_net: merge: remove skip_xdp
  virtio_net: introduce receive_small_xdp()
  virtio_net: small: optimize code
  virtio_net: small: optimize code
  virtio_net: small: remove skip_xdp

 drivers/net/virtio_net.c | 625 +++++++++++++++++++++++----------------
 1 file changed, 362 insertions(+), 263 deletions(-)

--
2.32.0.3.g01195cf9f

_______________________________________________
Virtualization mailing list
Virtualization@lists.linux-foundation.org
https://lists.linuxfoundation.org/mailman/listinfo/virtualization

[PATCH net-next v2 01/14] virtio_net: mergeable xdp: put old page immediately

[Xuan Zhuo]

In the xdp implementation of virtio-net mergeable, it always checks
whether two page is used and a page is selected to release. This is
complicated for the processing of action, and be careful.

In the entire process, we have such principles:
* If xdp_page is used (PASS, TX, Redirect), then we release the old
  page.
* If it is a drop case, we will release two. The old page obtained from
  buf is release inside err_xdp, and xdp_page needs be relased by us.

But in fact, when we allocate a new page, we can release the old page
immediately. Then just one is using, we just need to release the new
page for drop case. On the drop path, err_xdp will release the variable
"page", so we only need to let "page" point to the new xdp_page in
advance.

Signed-off-by: Xuan Zhuo <xuanzhuo@linux.alibaba.com>
---
 drivers/net/virtio_net.c | 19 +++++++------------
 1 file changed, 7 insertions(+), 12 deletions(-)

diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
index e2560b6f7980..42435e762d72 100644
--- a/drivers/net/virtio_net.c
+++ b/drivers/net/virtio_net.c
@@ -1245,6 +1245,9 @@ static struct sk_buff *receive_mergeable(struct net_device *dev,
 			if (!xdp_page)
 				goto err_xdp;
 			offset = VIRTIO_XDP_HEADROOM;
+
+			put_page(page);
+			page = xdp_page;
 		} else if (unlikely(headroom < virtnet_get_headroom(vi))) {
 			xdp_room = SKB_DATA_ALIGN(VIRTIO_XDP_HEADROOM +
 						  sizeof(struct skb_shared_info));
@@ -1259,11 +1262,12 @@ static struct sk_buff *receive_mergeable(struct net_device *dev,
 			       page_address(page) + offset, len);
 			frame_sz = PAGE_SIZE;
 			offset = VIRTIO_XDP_HEADROOM;
-		} else {
-			xdp_page = page;
+
+			put_page(page);
+			page = xdp_page;
 		}
 
-		data = page_address(xdp_page) + offset;
+		data = page_address(page) + offset;
 		err = virtnet_build_xdp_buff_mrg(dev, vi, rq, &xdp, data, len, frame_sz,
 						 &num_buf, &xdp_frags_truesz, stats);
 		if (unlikely(err))
@@ -1278,8 +1282,6 @@ static struct sk_buff *receive_mergeable(struct net_device *dev,
 			if (unlikely(!head_skb))
 				goto err_xdp_frags;
 
-			if (unlikely(xdp_page != page))
-				put_page(page);
 			rcu_read_unlock();
 			return head_skb;
 		case XDP_TX:
@@ -1297,8 +1299,6 @@ static struct sk_buff *receive_mergeable(struct net_device *dev,
 				goto err_xdp_frags;
 			}
 			*xdp_xmit |= VIRTIO_XDP_TX;
-			if (unlikely(xdp_page != page))
-				put_page(page);
 			rcu_read_unlock();
 			goto xdp_xmit;
 		case XDP_REDIRECT:
@@ -1307,8 +1307,6 @@ static struct sk_buff *receive_mergeable(struct net_device *dev,
 			if (err)
 				goto err_xdp_frags;
 			*xdp_xmit |= VIRTIO_XDP_REDIR;
-			if (unlikely(xdp_page != page))
-				put_page(page);
 			rcu_read_unlock();
 			goto xdp_xmit;
 		default:
@@ -1321,9 +1319,6 @@ static struct sk_buff *receive_mergeable(struct net_device *dev,
 			goto err_xdp_frags;
 		}
 err_xdp_frags:
-		if (unlikely(xdp_page != page))
-			__free_pages(xdp_page, 0);
-
 		if (xdp_buff_has_frags(&xdp)) {
 			shinfo = xdp_get_shared_info_from_buff(&xdp);
 			for (i = 0; i < shinfo->nr_frags; i++) {
-- 
2.32.0.3.g01195cf9f

_______________________________________________
Virtualization mailing list
Virtualization@lists.linux-foundation.org
https://lists.linuxfoundation.org/mailman/listinfo/virtualization

[PATCH net-next v2 02/14] virtio_net: introduce mergeable_xdp_prepare()

[Xuan Zhuo]

Separating the logic of preparation for xdp from receive_mergeable.

The purpose of this is to simplify the logic of execution of XDP.

The main logic here is that when headroom is insufficient, we need to
allocate a new page and calculate offset. It should be noted that if
there is new page, the variable page will refer to the new page.

Signed-off-by: Xuan Zhuo <xuanzhuo@linux.alibaba.com>
---
 drivers/net/virtio_net.c | 135 +++++++++++++++++++++++----------------
 1 file changed, 79 insertions(+), 56 deletions(-)

diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
index 42435e762d72..12559062ffb6 100644
--- a/drivers/net/virtio_net.c
+++ b/drivers/net/virtio_net.c
@@ -1162,6 +1162,81 @@ static int virtnet_build_xdp_buff_mrg(struct net_device *dev,
 	return 0;
 }
 
+static void *mergeable_xdp_prepare(struct virtnet_info *vi,
+				   struct receive_queue *rq,
+				   struct bpf_prog *xdp_prog,
+				   void *ctx,
+				   unsigned int *frame_sz,
+				   int *num_buf,
+				   struct page **page,
+				   int offset,
+				   unsigned int *len,
+				   struct virtio_net_hdr_mrg_rxbuf *hdr)
+{
+	unsigned int truesize = mergeable_ctx_to_truesize(ctx);
+	unsigned int headroom = mergeable_ctx_to_headroom(ctx);
+	struct page *xdp_page;
+	unsigned int xdp_room;
+
+	/* Transient failure which in theory could occur if
+	 * in-flight packets from before XDP was enabled reach
+	 * the receive path after XDP is loaded.
+	 */
+	if (unlikely(hdr->hdr.gso_type))
+		return NULL;
+
+	/* Now XDP core assumes frag size is PAGE_SIZE, but buffers
+	 * with headroom may add hole in truesize, which
+	 * make their length exceed PAGE_SIZE. So we disabled the
+	 * hole mechanism for xdp. See add_recvbuf_mergeable().
+	 */
+	*frame_sz = truesize;
+
+	/* This happens when headroom is not enough because
+	 * of the buffer was prefilled before XDP is set.
+	 * This should only happen for the first several packets.
+	 * In fact, vq reset can be used here to help us clean up
+	 * the prefilled buffers, but many existing devices do not
+	 * support it, and we don't want to bother users who are
+	 * using xdp normally.
+	 */
+	if (!xdp_prog->aux->xdp_has_frags &&
+	    (*num_buf > 1 || headroom < virtnet_get_headroom(vi))) {
+		/* linearize data for XDP */
+		xdp_page = xdp_linearize_page(rq, num_buf,
+					      *page, offset,
+					      VIRTIO_XDP_HEADROOM,
+					      len);
+		*frame_sz = PAGE_SIZE;
+
+		if (!xdp_page)
+			return NULL;
+		offset = VIRTIO_XDP_HEADROOM;
+
+		put_page(*page);
+		*page = xdp_page;
+	} else if (unlikely(headroom < virtnet_get_headroom(vi))) {
+		xdp_room = SKB_DATA_ALIGN(VIRTIO_XDP_HEADROOM +
+					  sizeof(struct skb_shared_info));
+		if (*len + xdp_room > PAGE_SIZE)
+			return NULL;
+
+		xdp_page = alloc_page(GFP_ATOMIC);
+		if (!xdp_page)
+			return NULL;
+
+		memcpy(page_address(xdp_page) + VIRTIO_XDP_HEADROOM,
+		       page_address(*page) + offset, *len);
+		*frame_sz = PAGE_SIZE;
+		offset = VIRTIO_XDP_HEADROOM;
+
+		put_page(*page);
+		*page = xdp_page;
+	}
+
+	return page_address(*page) + offset;
+}
+
 static struct sk_buff *receive_mergeable(struct net_device *dev,
 					 struct virtnet_info *vi,
 					 struct receive_queue *rq,
@@ -1181,7 +1256,7 @@ static struct sk_buff *receive_mergeable(struct net_device *dev,
 	unsigned int headroom = mergeable_ctx_to_headroom(ctx);
 	unsigned int tailroom = headroom ? sizeof(struct skb_shared_info) : 0;
 	unsigned int room = SKB_DATA_ALIGN(headroom + tailroom);
-	unsigned int frame_sz, xdp_room;
+	unsigned int frame_sz;
 	int err;
 
 	head_skb = NULL;
@@ -1211,63 +1286,11 @@ static struct sk_buff *receive_mergeable(struct net_device *dev,
 		u32 act;
 		int i;
 
-		/* Transient failure which in theory could occur if
-		 * in-flight packets from before XDP was enabled reach
-		 * the receive path after XDP is loaded.
-		 */
-		if (unlikely(hdr->hdr.gso_type))
+		data = mergeable_xdp_prepare(vi, rq, xdp_prog, ctx, &frame_sz,
+					     &num_buf, &page, offset, &len, hdr);
+		if (unlikely(!data))
 			goto err_xdp;
 
-		/* Now XDP core assumes frag size is PAGE_SIZE, but buffers
-		 * with headroom may add hole in truesize, which
-		 * make their length exceed PAGE_SIZE. So we disabled the
-		 * hole mechanism for xdp. See add_recvbuf_mergeable().
-		 */
-		frame_sz = truesize;
-
-		/* This happens when headroom is not enough because
-		 * of the buffer was prefilled before XDP is set.
-		 * This should only happen for the first several packets.
-		 * In fact, vq reset can be used here to help us clean up
-		 * the prefilled buffers, but many existing devices do not
-		 * support it, and we don't want to bother users who are
-		 * using xdp normally.
-		 */
-		if (!xdp_prog->aux->xdp_has_frags &&
-		    (num_buf > 1 || headroom < virtnet_get_headroom(vi))) {
-			/* linearize data for XDP */
-			xdp_page = xdp_linearize_page(rq, &num_buf,
-						      page, offset,
-						      VIRTIO_XDP_HEADROOM,
-						      &len);
-			frame_sz = PAGE_SIZE;
-
-			if (!xdp_page)
-				goto err_xdp;
-			offset = VIRTIO_XDP_HEADROOM;
-
-			put_page(page);
-			page = xdp_page;
-		} else if (unlikely(headroom < virtnet_get_headroom(vi))) {
-			xdp_room = SKB_DATA_ALIGN(VIRTIO_XDP_HEADROOM +
-						  sizeof(struct skb_shared_info));
-			if (len + xdp_room > PAGE_SIZE)
-				goto err_xdp;
-
-			xdp_page = alloc_page(GFP_ATOMIC);
-			if (!xdp_page)
-				goto err_xdp;
-
-			memcpy(page_address(xdp_page) + VIRTIO_XDP_HEADROOM,
-			       page_address(page) + offset, len);
-			frame_sz = PAGE_SIZE;
-			offset = VIRTIO_XDP_HEADROOM;
-
-			put_page(page);
-			page = xdp_page;
-		}
-
-		data = page_address(page) + offset;
 		err = virtnet_build_xdp_buff_mrg(dev, vi, rq, &xdp, data, len, frame_sz,
 						 &num_buf, &xdp_frags_truesz, stats);
 		if (unlikely(err))
-- 
2.32.0.3.g01195cf9f

_______________________________________________
Virtualization mailing list
Virtualization@lists.linux-foundation.org
https://lists.linuxfoundation.org/mailman/listinfo/virtualization

[PATCH net-next v2 03/14] virtio_net: optimize mergeable_xdp_prepare()

[Xuan Zhuo]

The previous patch, in order to facilitate review, I do not do any
modification. This patch has made some optimization on the top.

* remove some repeated logics in this function.
* add fast check for passing without any alloc.

Signed-off-by: Xuan Zhuo <xuanzhuo@linux.alibaba.com>
---
 drivers/net/virtio_net.c | 29 ++++++++++++++---------------
 1 file changed, 14 insertions(+), 15 deletions(-)

diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
index 12559062ffb6..50dc64d80d3b 100644
--- a/drivers/net/virtio_net.c
+++ b/drivers/net/virtio_net.c
@@ -1192,6 +1192,11 @@ static void *mergeable_xdp_prepare(struct virtnet_info *vi,
 	 */
 	*frame_sz = truesize;
 
+	if (likely(headroom >= virtnet_get_headroom(vi) &&
+		   (*num_buf == 1 || xdp_prog->aux->xdp_has_frags))) {
+		return page_address(*page) + offset;
+	}
+
 	/* This happens when headroom is not enough because
 	 * of the buffer was prefilled before XDP is set.
 	 * This should only happen for the first several packets.
@@ -1200,22 +1205,15 @@ static void *mergeable_xdp_prepare(struct virtnet_info *vi,
 	 * support it, and we don't want to bother users who are
 	 * using xdp normally.
 	 */
-	if (!xdp_prog->aux->xdp_has_frags &&
-	    (*num_buf > 1 || headroom < virtnet_get_headroom(vi))) {
+	if (!xdp_prog->aux->xdp_has_frags) {
 		/* linearize data for XDP */
 		xdp_page = xdp_linearize_page(rq, num_buf,
 					      *page, offset,
 					      VIRTIO_XDP_HEADROOM,
 					      len);
-		*frame_sz = PAGE_SIZE;
-
 		if (!xdp_page)
 			return NULL;
-		offset = VIRTIO_XDP_HEADROOM;
-
-		put_page(*page);
-		*page = xdp_page;
-	} else if (unlikely(headroom < virtnet_get_headroom(vi))) {
+	} else {
 		xdp_room = SKB_DATA_ALIGN(VIRTIO_XDP_HEADROOM +
 					  sizeof(struct skb_shared_info));
 		if (*len + xdp_room > PAGE_SIZE)
@@ -1227,14 +1225,15 @@ static void *mergeable_xdp_prepare(struct virtnet_info *vi,
 
 		memcpy(page_address(xdp_page) + VIRTIO_XDP_HEADROOM,
 		       page_address(*page) + offset, *len);
-		*frame_sz = PAGE_SIZE;
-		offset = VIRTIO_XDP_HEADROOM;
-
-		put_page(*page);
-		*page = xdp_page;
 	}
 
-	return page_address(*page) + offset;
+	*frame_sz = PAGE_SIZE;
+
+	put_page(*page);
+
+	*page = xdp_page;
+
+	return page_address(*page) + VIRTIO_XDP_HEADROOM;
 }
 
 static struct sk_buff *receive_mergeable(struct net_device *dev,
-- 
2.32.0.3.g01195cf9f

_______________________________________________
Virtualization mailing list
Virtualization@lists.linux-foundation.org
https://lists.linuxfoundation.org/mailman/listinfo/virtualization

[PATCH net-next v2 04/14] virtio_net: introduce virtnet_xdp_handler() to seprate the logic of run xdp

[Xuan Zhuo]

At present, we have two similar logic to perform the XDP prog.

Therefore, this PATCH separates the code of executing XDP, which is
conducive to later maintenance.

Signed-off-by: Xuan Zhuo <xuanzhuo@linux.alibaba.com>
---
 drivers/net/virtio_net.c | 118 +++++++++++++++++++--------------------
 1 file changed, 58 insertions(+), 60 deletions(-)

diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
index 50dc64d80d3b..0fa64c314ea7 100644
--- a/drivers/net/virtio_net.c
+++ b/drivers/net/virtio_net.c
@@ -789,6 +789,60 @@ static int virtnet_xdp_xmit(struct net_device *dev,
 	return ret;
 }
 
+static int virtnet_xdp_handler(struct bpf_prog *xdp_prog, struct xdp_buff *xdp,
+			       struct net_device *dev,
+			       unsigned int *xdp_xmit,
+			       struct virtnet_rq_stats *stats)
+{
+	struct xdp_frame *xdpf;
+	int err;
+	u32 act;
+
+	act = bpf_prog_run_xdp(xdp_prog, xdp);
+	stats->xdp_packets++;
+
+	switch (act) {
+	case XDP_PASS:
+		return act;
+
+	case XDP_TX:
+		stats->xdp_tx++;
+		xdpf = xdp_convert_buff_to_frame(xdp);
+		if (unlikely(!xdpf)) {
+			netdev_dbg(dev, "convert buff to frame failed for xdp\n");
+			return XDP_DROP;
+		}
+
+		err = virtnet_xdp_xmit(dev, 1, &xdpf, 0);
+		if (unlikely(!err)) {
+			xdp_return_frame_rx_napi(xdpf);
+		} else if (unlikely(err < 0)) {
+			trace_xdp_exception(dev, xdp_prog, act);
+			return XDP_DROP;
+		}
+		*xdp_xmit |= VIRTIO_XDP_TX;
+		return act;
+
+	case XDP_REDIRECT:
+		stats->xdp_redirects++;
+		err = xdp_do_redirect(dev, xdp, xdp_prog);
+		if (err)
+			return XDP_DROP;
+
+		*xdp_xmit |= VIRTIO_XDP_REDIR;
+		return act;
+
+	default:
+		bpf_warn_invalid_xdp_action(dev, xdp_prog, act);
+		fallthrough;
+	case XDP_ABORTED:
+		trace_xdp_exception(dev, xdp_prog, act);
+		fallthrough;
+	case XDP_DROP:
+		return XDP_DROP;
+	}
+}
+
 static unsigned int virtnet_get_headroom(struct virtnet_info *vi)
 {
 	return vi->xdp_enabled ? VIRTIO_XDP_HEADROOM : 0;
@@ -876,7 +930,6 @@ static struct sk_buff *receive_small(struct net_device *dev,
 	struct page *page = virt_to_head_page(buf);
 	unsigned int delta = 0;
 	struct page *xdp_page;
-	int err;
 	unsigned int metasize = 0;
 
 	len -= vi->hdr_len;
@@ -898,7 +951,6 @@ static struct sk_buff *receive_small(struct net_device *dev,
 	xdp_prog = rcu_dereference(rq->xdp_prog);
 	if (xdp_prog) {
 		struct virtio_net_hdr_mrg_rxbuf *hdr = buf + header_offset;
-		struct xdp_frame *xdpf;
 		struct xdp_buff xdp;
 		void *orig_data;
 		u32 act;
@@ -931,8 +983,8 @@ static struct sk_buff *receive_small(struct net_device *dev,
 		xdp_prepare_buff(&xdp, buf + VIRTNET_RX_PAD + vi->hdr_len,
 				 xdp_headroom, len, true);
 		orig_data = xdp.data;
-		act = bpf_prog_run_xdp(xdp_prog, &xdp);
-		stats->xdp_packets++;
+
+		act = virtnet_xdp_handler(xdp_prog, &xdp, dev, xdp_xmit, stats);
 
 		switch (act) {
 		case XDP_PASS:
@@ -942,35 +994,10 @@ static struct sk_buff *receive_small(struct net_device *dev,
 			metasize = xdp.data - xdp.data_meta;
 			break;
 		case XDP_TX:
-			stats->xdp_tx++;
-			xdpf = xdp_convert_buff_to_frame(&xdp);
-			if (unlikely(!xdpf))
-				goto err_xdp;
-			err = virtnet_xdp_xmit(dev, 1, &xdpf, 0);
-			if (unlikely(!err)) {
-				xdp_return_frame_rx_napi(xdpf);
-			} else if (unlikely(err < 0)) {
-				trace_xdp_exception(vi->dev, xdp_prog, act);
-				goto err_xdp;
-			}
-			*xdp_xmit |= VIRTIO_XDP_TX;
-			rcu_read_unlock();
-			goto xdp_xmit;
 		case XDP_REDIRECT:
-			stats->xdp_redirects++;
-			err = xdp_do_redirect(dev, &xdp, xdp_prog);
-			if (err)
-				goto err_xdp;
-			*xdp_xmit |= VIRTIO_XDP_REDIR;
 			rcu_read_unlock();
 			goto xdp_xmit;
 		default:
-			bpf_warn_invalid_xdp_action(vi->dev, xdp_prog, act);
-			fallthrough;
-		case XDP_ABORTED:
-			trace_xdp_exception(vi->dev, xdp_prog, act);
-			goto err_xdp;
-		case XDP_DROP:
 			goto err_xdp;
 		}
 	}
@@ -1278,7 +1305,6 @@ static struct sk_buff *receive_mergeable(struct net_device *dev,
 	if (xdp_prog) {
 		unsigned int xdp_frags_truesz = 0;
 		struct skb_shared_info *shinfo;
-		struct xdp_frame *xdpf;
 		struct page *xdp_page;
 		struct xdp_buff xdp;
 		void *data;
@@ -1295,8 +1321,7 @@ static struct sk_buff *receive_mergeable(struct net_device *dev,
 		if (unlikely(err))
 			goto err_xdp_frags;
 
-		act = bpf_prog_run_xdp(xdp_prog, &xdp);
-		stats->xdp_packets++;
+		act = virtnet_xdp_handler(xdp_prog, &xdp, dev, xdp_xmit, stats);
 
 		switch (act) {
 		case XDP_PASS:
@@ -1307,38 +1332,11 @@ static struct sk_buff *receive_mergeable(struct net_device *dev,
 			rcu_read_unlock();
 			return head_skb;
 		case XDP_TX:
-			stats->xdp_tx++;
-			xdpf = xdp_convert_buff_to_frame(&xdp);
-			if (unlikely(!xdpf)) {
-				netdev_dbg(dev, "convert buff to frame failed for xdp\n");
-				goto err_xdp_frags;
-			}
-			err = virtnet_xdp_xmit(dev, 1, &xdpf, 0);
-			if (unlikely(!err)) {
-				xdp_return_frame_rx_napi(xdpf);
-			} else if (unlikely(err < 0)) {
-				trace_xdp_exception(vi->dev, xdp_prog, act);
-				goto err_xdp_frags;
-			}
-			*xdp_xmit |= VIRTIO_XDP_TX;
-			rcu_read_unlock();
-			goto xdp_xmit;
 		case XDP_REDIRECT:
-			stats->xdp_redirects++;
-			err = xdp_do_redirect(dev, &xdp, xdp_prog);
-			if (err)
-				goto err_xdp_frags;
-			*xdp_xmit |= VIRTIO_XDP_REDIR;
 			rcu_read_unlock();
 			goto xdp_xmit;
 		default:
-			bpf_warn_invalid_xdp_action(vi->dev, xdp_prog, act);
-			fallthrough;
-		case XDP_ABORTED:
-			trace_xdp_exception(vi->dev, xdp_prog, act);
-			fallthrough;
-		case XDP_DROP:
-			goto err_xdp_frags;
+			break;
 		}
 err_xdp_frags:
 		if (xdp_buff_has_frags(&xdp)) {
-- 
2.32.0.3.g01195cf9f

_______________________________________________
Virtualization mailing list
Virtualization@lists.linux-foundation.org
https://lists.linuxfoundation.org/mailman/listinfo/virtualization

[PATCH net-next v2 05/14] virtio_net: introduce xdp res enums

[Xuan Zhuo]

virtnet_xdp_handler() is to process all the logic related to XDP. The
caller only needs to care about how to deal with the buf. So this commit
introduces new enums:

1. VIRTNET_XDP_RES_PASS: make skb by the buf
2. VIRTNET_XDP_RES_DROP: xdp return drop action or some error, caller
   should release the buf
3. VIRTNET_XDP_RES_CONSUMED: xdp consumed the buf, the caller doesnot to
   do anything

Signed-off-by: Xuan Zhuo <xuanzhuo@linux.alibaba.com>
---
 drivers/net/virtio_net.c | 42 ++++++++++++++++++++++++++--------------
 1 file changed, 27 insertions(+), 15 deletions(-)

diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
index 0fa64c314ea7..4dfdc211d355 100644
--- a/drivers/net/virtio_net.c
+++ b/drivers/net/virtio_net.c
@@ -301,6 +301,15 @@ struct padded_vnet_hdr {
 	char padding[12];
 };
 
+enum {
+	/* xdp pass */
+	VIRTNET_XDP_RES_PASS,
+	/* drop packet. the caller needs to release the page. */
+	VIRTNET_XDP_RES_DROP,
+	/* packet is consumed by xdp. the caller needs to do nothing. */
+	VIRTNET_XDP_RES_CONSUMED,
+};
+
 static void virtnet_rq_free_unused_buf(struct virtqueue *vq, void *buf);
 static void virtnet_sq_free_unused_buf(struct virtqueue *vq, void *buf);
 
@@ -803,14 +812,14 @@ static int virtnet_xdp_handler(struct bpf_prog *xdp_prog, struct xdp_buff *xdp,
 
 	switch (act) {
 	case XDP_PASS:
-		return act;
+		return VIRTNET_XDP_RES_PASS;
 
 	case XDP_TX:
 		stats->xdp_tx++;
 		xdpf = xdp_convert_buff_to_frame(xdp);
 		if (unlikely(!xdpf)) {
 			netdev_dbg(dev, "convert buff to frame failed for xdp\n");
-			return XDP_DROP;
+			return VIRTNET_XDP_RES_DROP;
 		}
 
 		err = virtnet_xdp_xmit(dev, 1, &xdpf, 0);
@@ -818,19 +827,20 @@ static int virtnet_xdp_handler(struct bpf_prog *xdp_prog, struct xdp_buff *xdp,
 			xdp_return_frame_rx_napi(xdpf);
 		} else if (unlikely(err < 0)) {
 			trace_xdp_exception(dev, xdp_prog, act);
-			return XDP_DROP;
+			return VIRTNET_XDP_RES_DROP;
 		}
+
 		*xdp_xmit |= VIRTIO_XDP_TX;
-		return act;
+		return VIRTNET_XDP_RES_CONSUMED;
 
 	case XDP_REDIRECT:
 		stats->xdp_redirects++;
 		err = xdp_do_redirect(dev, xdp, xdp_prog);
 		if (err)
-			return XDP_DROP;
+			return VIRTNET_XDP_RES_DROP;
 
 		*xdp_xmit |= VIRTIO_XDP_REDIR;
-		return act;
+		return VIRTNET_XDP_RES_CONSUMED;
 
 	default:
 		bpf_warn_invalid_xdp_action(dev, xdp_prog, act);
@@ -839,7 +849,7 @@ static int virtnet_xdp_handler(struct bpf_prog *xdp_prog, struct xdp_buff *xdp,
 		trace_xdp_exception(dev, xdp_prog, act);
 		fallthrough;
 	case XDP_DROP:
-		return XDP_DROP;
+		return VIRTNET_XDP_RES_DROP;
 	}
 }
 
@@ -987,17 +997,18 @@ static struct sk_buff *receive_small(struct net_device *dev,
 		act = virtnet_xdp_handler(xdp_prog, &xdp, dev, xdp_xmit, stats);
 
 		switch (act) {
-		case XDP_PASS:
+		case VIRTNET_XDP_RES_PASS:
 			/* Recalculate length in case bpf program changed it */
 			delta = orig_data - xdp.data;
 			len = xdp.data_end - xdp.data;
 			metasize = xdp.data - xdp.data_meta;
 			break;
-		case XDP_TX:
-		case XDP_REDIRECT:
+
+		case VIRTNET_XDP_RES_CONSUMED:
 			rcu_read_unlock();
 			goto xdp_xmit;
-		default:
+
+		case VIRTNET_XDP_RES_DROP:
 			goto err_xdp;
 		}
 	}
@@ -1324,18 +1335,19 @@ static struct sk_buff *receive_mergeable(struct net_device *dev,
 		act = virtnet_xdp_handler(xdp_prog, &xdp, dev, xdp_xmit, stats);
 
 		switch (act) {
-		case XDP_PASS:
+		case VIRTNET_XDP_RES_PASS:
 			head_skb = build_skb_from_xdp_buff(dev, vi, &xdp, xdp_frags_truesz);
 			if (unlikely(!head_skb))
 				goto err_xdp_frags;
 
 			rcu_read_unlock();
 			return head_skb;
-		case XDP_TX:
-		case XDP_REDIRECT:
+
+		case VIRTNET_XDP_RES_CONSUMED:
 			rcu_read_unlock();
 			goto xdp_xmit;
-		default:
+
+		case VIRTNET_XDP_RES_DROP:
 			break;
 		}
 err_xdp_frags:
-- 
2.32.0.3.g01195cf9f

_______________________________________________
Virtualization mailing list
Virtualization@lists.linux-foundation.org
https://lists.linuxfoundation.org/mailman/listinfo/virtualization

[PATCH net-next v2 06/14] virtio_net: separate the logic of freeing xdp shinfo

[Xuan Zhuo]

This patch introduce a new function that releases the
xdp shinfo. The subsequent patch will reuse this function.

Signed-off-by: Xuan Zhuo <xuanzhuo@linux.alibaba.com>
Acked-by: Jason Wang <jasowang@redhat.com>
---
 drivers/net/virtio_net.c | 27 ++++++++++++++++-----------
 1 file changed, 16 insertions(+), 11 deletions(-)

diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
index 4dfdc211d355..5cec4b418110 100644
--- a/drivers/net/virtio_net.c
+++ b/drivers/net/virtio_net.c
@@ -798,6 +798,21 @@ static int virtnet_xdp_xmit(struct net_device *dev,
 	return ret;
 }
 
+static void put_xdp_frags(struct xdp_buff *xdp)
+{
+	struct skb_shared_info *shinfo;
+	struct page *xdp_page;
+	int i;
+
+	if (xdp_buff_has_frags(xdp)) {
+		shinfo = xdp_get_shared_info_from_buff(xdp);
+		for (i = 0; i < shinfo->nr_frags; i++) {
+			xdp_page = skb_frag_page(&shinfo->frags[i]);
+			put_page(xdp_page);
+		}
+	}
+}
+
 static int virtnet_xdp_handler(struct bpf_prog *xdp_prog, struct xdp_buff *xdp,
 			       struct net_device *dev,
 			       unsigned int *xdp_xmit,
@@ -1315,12 +1330,9 @@ static struct sk_buff *receive_mergeable(struct net_device *dev,
 	xdp_prog = rcu_dereference(rq->xdp_prog);
 	if (xdp_prog) {
 		unsigned int xdp_frags_truesz = 0;
-		struct skb_shared_info *shinfo;
-		struct page *xdp_page;
 		struct xdp_buff xdp;
 		void *data;
 		u32 act;
-		int i;
 
 		data = mergeable_xdp_prepare(vi, rq, xdp_prog, ctx, &frame_sz,
 					     &num_buf, &page, offset, &len, hdr);
@@ -1351,14 +1363,7 @@ static struct sk_buff *receive_mergeable(struct net_device *dev,
 			break;
 		}
 err_xdp_frags:
-		if (xdp_buff_has_frags(&xdp)) {
-			shinfo = xdp_get_shared_info_from_buff(&xdp);
-			for (i = 0; i < shinfo->nr_frags; i++) {
-				xdp_page = skb_frag_page(&shinfo->frags[i]);
-				put_page(xdp_page);
-			}
-		}
-
+		put_xdp_frags(&xdp);
 		goto err_xdp;
 	}
 	rcu_read_unlock();
-- 
2.32.0.3.g01195cf9f

_______________________________________________
Virtualization mailing list
Virtualization@lists.linux-foundation.org
https://lists.linuxfoundation.org/mailman/listinfo/virtualization

[PATCH net-next v2 07/14] virtio_net: separate the logic of freeing the rest mergeable buf

[Xuan Zhuo]

This patch introduce a new function that frees the rest mergeable buf.
The subsequent patch will reuse this function.

Signed-off-by: Xuan Zhuo <xuanzhuo@linux.alibaba.com>
Acked-by: Jason Wang <jasowang@redhat.com>
---
 drivers/net/virtio_net.c | 36 ++++++++++++++++++++++++------------
 1 file changed, 24 insertions(+), 12 deletions(-)

diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
index 5cec4b418110..e2eade87d2d4 100644
--- a/drivers/net/virtio_net.c
+++ b/drivers/net/virtio_net.c
@@ -1078,6 +1078,28 @@ static struct sk_buff *receive_big(struct net_device *dev,
 	return NULL;
 }
 
+static void mergeable_buf_free(struct receive_queue *rq, int num_buf,
+			       struct net_device *dev,
+			       struct virtnet_rq_stats *stats)
+{
+	struct page *page;
+	void *buf;
+	int len;
+
+	while (num_buf-- > 1) {
+		buf = virtqueue_get_buf(rq->vq, &len);
+		if (unlikely(!buf)) {
+			pr_debug("%s: rx error: %d buffers missing\n",
+				 dev->name, num_buf);
+			dev->stats.rx_length_errors++;
+			break;
+		}
+		stats->bytes += len;
+		page = virt_to_head_page(buf);
+		put_page(page);
+	}
+}
+
 /* Why not use xdp_build_skb_from_frame() ?
  * XDP core assumes that xdp frags are PAGE_SIZE in length, while in
  * virtio-net there are 2 points that do not match its requirements:
@@ -1439,18 +1461,8 @@ static struct sk_buff *receive_mergeable(struct net_device *dev,
 	stats->xdp_drops++;
 err_skb:
 	put_page(page);
-	while (num_buf-- > 1) {
-		buf = virtqueue_get_buf(rq->vq, &len);
-		if (unlikely(!buf)) {
-			pr_debug("%s: rx error: %d buffers missing\n",
-				 dev->name, num_buf);
-			dev->stats.rx_length_errors++;
-			break;
-		}
-		stats->bytes += len;
-		page = virt_to_head_page(buf);
-		put_page(page);
-	}
+	mergeable_buf_free(rq, num_buf, dev, stats);
+
 err_buf:
 	stats->drops++;
 	dev_kfree_skb(head_skb);
-- 
2.32.0.3.g01195cf9f

_______________________________________________
Virtualization mailing list
Virtualization@lists.linux-foundation.org
https://lists.linuxfoundation.org/mailman/listinfo/virtualization

[PATCH net-next v2 08/14] virtio_net: auto release xdp shinfo

[Xuan Zhuo]

virtnet_build_xdp_buff_mrg() and virtnet_xdp_handler() auto
release xdp shinfo then the caller no need to careful the xdp shinfo.

Signed-off-by: Xuan Zhuo <xuanzhuo@linux.alibaba.com>
---
 drivers/net/virtio_net.c | 29 +++++++++++++++++------------
 1 file changed, 17 insertions(+), 12 deletions(-)

diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
index e2eade87d2d4..266c1670beda 100644
--- a/drivers/net/virtio_net.c
+++ b/drivers/net/virtio_net.c
@@ -834,7 +834,7 @@ static int virtnet_xdp_handler(struct bpf_prog *xdp_prog, struct xdp_buff *xdp,
 		xdpf = xdp_convert_buff_to_frame(xdp);
 		if (unlikely(!xdpf)) {
 			netdev_dbg(dev, "convert buff to frame failed for xdp\n");
-			return VIRTNET_XDP_RES_DROP;
+			goto drop;
 		}
 
 		err = virtnet_xdp_xmit(dev, 1, &xdpf, 0);
@@ -842,7 +842,7 @@ static int virtnet_xdp_handler(struct bpf_prog *xdp_prog, struct xdp_buff *xdp,
 			xdp_return_frame_rx_napi(xdpf);
 		} else if (unlikely(err < 0)) {
 			trace_xdp_exception(dev, xdp_prog, act);
-			return VIRTNET_XDP_RES_DROP;
+			goto drop;
 		}
 
 		*xdp_xmit |= VIRTIO_XDP_TX;
@@ -852,7 +852,7 @@ static int virtnet_xdp_handler(struct bpf_prog *xdp_prog, struct xdp_buff *xdp,
 		stats->xdp_redirects++;
 		err = xdp_do_redirect(dev, xdp, xdp_prog);
 		if (err)
-			return VIRTNET_XDP_RES_DROP;
+			goto drop;
 
 		*xdp_xmit |= VIRTIO_XDP_REDIR;
 		return VIRTNET_XDP_RES_CONSUMED;
@@ -864,8 +864,12 @@ static int virtnet_xdp_handler(struct bpf_prog *xdp_prog, struct xdp_buff *xdp,
 		trace_xdp_exception(dev, xdp_prog, act);
 		fallthrough;
 	case XDP_DROP:
-		return VIRTNET_XDP_RES_DROP;
+		break;
 	}
+
+drop:
+	put_xdp_frags(xdp);
+	return VIRTNET_XDP_RES_DROP;
 }
 
 static unsigned int virtnet_get_headroom(struct virtnet_info *vi)
@@ -1201,7 +1205,7 @@ static int virtnet_build_xdp_buff_mrg(struct net_device *dev,
 				 dev->name, *num_buf,
 				 virtio16_to_cpu(vi->vdev, hdr->num_buffers));
 			dev->stats.rx_length_errors++;
-			return -EINVAL;
+			goto err;
 		}
 
 		stats->bytes += len;
@@ -1220,7 +1224,7 @@ static int virtnet_build_xdp_buff_mrg(struct net_device *dev,
 			pr_debug("%s: rx error: len %u exceeds truesize %lu\n",
 				 dev->name, len, (unsigned long)(truesize - room));
 			dev->stats.rx_length_errors++;
-			return -EINVAL;
+			goto err;
 		}
 
 		frag = &shinfo->frags[shinfo->nr_frags++];
@@ -1235,6 +1239,10 @@ static int virtnet_build_xdp_buff_mrg(struct net_device *dev,
 
 	*xdp_frags_truesize = xdp_frags_truesz;
 	return 0;
+
+err:
+	put_xdp_frags(xdp);
+	return -EINVAL;
 }
 
 static void *mergeable_xdp_prepare(struct virtnet_info *vi,
@@ -1364,7 +1372,7 @@ static struct sk_buff *receive_mergeable(struct net_device *dev,
 		err = virtnet_build_xdp_buff_mrg(dev, vi, rq, &xdp, data, len, frame_sz,
 						 &num_buf, &xdp_frags_truesz, stats);
 		if (unlikely(err))
-			goto err_xdp_frags;
+			goto err_xdp;
 
 		act = virtnet_xdp_handler(xdp_prog, &xdp, dev, xdp_xmit, stats);
 
@@ -1372,7 +1380,7 @@ static struct sk_buff *receive_mergeable(struct net_device *dev,
 		case VIRTNET_XDP_RES_PASS:
 			head_skb = build_skb_from_xdp_buff(dev, vi, &xdp, xdp_frags_truesz);
 			if (unlikely(!head_skb))
-				goto err_xdp_frags;
+				goto err_xdp;
 
 			rcu_read_unlock();
 			return head_skb;
@@ -1382,11 +1390,8 @@ static struct sk_buff *receive_mergeable(struct net_device *dev,
 			goto xdp_xmit;
 
 		case VIRTNET_XDP_RES_DROP:
-			break;
+			goto err_xdp;
 		}
-err_xdp_frags:
-		put_xdp_frags(&xdp);
-		goto err_xdp;
 	}
 	rcu_read_unlock();
 
-- 
2.32.0.3.g01195cf9f

_______________________________________________
Virtualization mailing list
Virtualization@lists.linux-foundation.org
https://lists.linuxfoundation.org/mailman/listinfo/virtualization

[PATCH net-next v2 09/14] virtio_net: introduce receive_mergeable_xdp()

[Xuan Zhuo]

The purpose of this patch is to simplify the receive_mergeable().
Separate all the logic of XDP into a function.

Signed-off-by: Xuan Zhuo <xuanzhuo@linux.alibaba.com>
---
 drivers/net/virtio_net.c | 100 ++++++++++++++++++++++++---------------
 1 file changed, 61 insertions(+), 39 deletions(-)

diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
index 266c1670beda..42e9927e316b 100644
--- a/drivers/net/virtio_net.c
+++ b/drivers/net/virtio_net.c
@@ -1319,6 +1319,63 @@ static void *mergeable_xdp_prepare(struct virtnet_info *vi,
 	return page_address(*page) + VIRTIO_XDP_HEADROOM;
 }
 
+static struct sk_buff *receive_mergeable_xdp(struct net_device *dev,
+					     struct virtnet_info *vi,
+					     struct receive_queue *rq,
+					     struct bpf_prog *xdp_prog,
+					     void *buf,
+					     void *ctx,
+					     unsigned int len,
+					     unsigned int *xdp_xmit,
+					     struct virtnet_rq_stats *stats)
+{
+	struct virtio_net_hdr_mrg_rxbuf *hdr = buf;
+	int num_buf = virtio16_to_cpu(vi->vdev, hdr->num_buffers);
+	struct page *page = virt_to_head_page(buf);
+	int offset = buf - page_address(page);
+	unsigned int xdp_frags_truesz = 0;
+	struct sk_buff *head_skb;
+	unsigned int frame_sz;
+	struct xdp_buff xdp;
+	void *data;
+	u32 act;
+	int err;
+
+	data = mergeable_xdp_prepare(vi, rq, xdp_prog, ctx, &frame_sz, &num_buf, &page,
+				     offset, &len, hdr);
+	if (unlikely(!data))
+		goto err_xdp;
+
+	err = virtnet_build_xdp_buff_mrg(dev, vi, rq, &xdp, data, len, frame_sz,
+					 &num_buf, &xdp_frags_truesz, stats);
+	if (unlikely(err))
+		goto err_xdp;
+
+	act = virtnet_xdp_handler(xdp_prog, &xdp, dev, xdp_xmit, stats);
+
+	switch (act) {
+	case VIRTNET_XDP_RES_PASS:
+		head_skb = build_skb_from_xdp_buff(dev, vi, &xdp, xdp_frags_truesz);
+		if (unlikely(!head_skb))
+			goto err_xdp;
+		return head_skb;
+
+	case VIRTNET_XDP_RES_CONSUMED:
+		return NULL;
+
+	case VIRTNET_XDP_RES_DROP:
+		break;
+	}
+
+err_xdp:
+	put_page(page);
+	mergeable_buf_free(rq, num_buf, dev, stats);
+
+	stats->xdp_drops++;
+	stats->drops++;
+	return NULL;
+}
+
 static struct sk_buff *receive_mergeable(struct net_device *dev,
 					 struct virtnet_info *vi,
 					 struct receive_queue *rq,
@@ -1338,8 +1395,6 @@ static struct sk_buff *receive_mergeable(struct net_device *dev,
 	unsigned int headroom = mergeable_ctx_to_headroom(ctx);
 	unsigned int tailroom = headroom ? sizeof(struct skb_shared_info) : 0;
 	unsigned int room = SKB_DATA_ALIGN(headroom + tailroom);
-	unsigned int frame_sz;
-	int err;
 
 	head_skb = NULL;
 	stats->bytes += len - vi->hdr_len;
@@ -1359,39 +1414,10 @@ static struct sk_buff *receive_mergeable(struct net_device *dev,
 	rcu_read_lock();
 	xdp_prog = rcu_dereference(rq->xdp_prog);
 	if (xdp_prog) {
-		unsigned int xdp_frags_truesz = 0;
-		struct xdp_buff xdp;
-		void *data;
-		u32 act;
-
-		data = mergeable_xdp_prepare(vi, rq, xdp_prog, ctx, &frame_sz,
-					     &num_buf, &page, offset, &len, hdr);
-		if (unlikely(!data))
-			goto err_xdp;
-
-		err = virtnet_build_xdp_buff_mrg(dev, vi, rq, &xdp, data, len, frame_sz,
-						 &num_buf, &xdp_frags_truesz, stats);
-		if (unlikely(err))
-			goto err_xdp;
-
-		act = virtnet_xdp_handler(xdp_prog, &xdp, dev, xdp_xmit, stats);
-
-		switch (act) {
-		case VIRTNET_XDP_RES_PASS:
-			head_skb = build_skb_from_xdp_buff(dev, vi, &xdp, xdp_frags_truesz);
-			if (unlikely(!head_skb))
-				goto err_xdp;
-
-			rcu_read_unlock();
-			return head_skb;
-
-		case VIRTNET_XDP_RES_CONSUMED:
-			rcu_read_unlock();
-			goto xdp_xmit;
-
-		case VIRTNET_XDP_RES_DROP:
-			goto err_xdp;
-		}
+		head_skb = receive_mergeable_xdp(dev, vi, rq, xdp_prog, buf, ctx,
+						 len, xdp_xmit, stats);
+		rcu_read_unlock();
+		return head_skb;
 	}
 	rcu_read_unlock();
 
@@ -1461,9 +1487,6 @@ static struct sk_buff *receive_mergeable(struct net_device *dev,
 	ewma_pkt_len_add(&rq->mrg_avg_pkt_len, head_skb->len);
 	return head_skb;
 
-err_xdp:
-	rcu_read_unlock();
-	stats->xdp_drops++;
 err_skb:
 	put_page(page);
 	mergeable_buf_free(rq, num_buf, dev, stats);
@@ -1471,7 +1494,6 @@ static struct sk_buff *receive_mergeable(struct net_device *dev,
 err_buf:
 	stats->drops++;
 	dev_kfree_skb(head_skb);
-xdp_xmit:
 	return NULL;
 }
 
-- 
2.32.0.3.g01195cf9f

_______________________________________________
Virtualization mailing list
Virtualization@lists.linux-foundation.org
https://lists.linuxfoundation.org/mailman/listinfo/virtualization

[PATCH net-next v2 10/14] virtio_net: merge: remove skip_xdp

[Xuan Zhuo]

Now, the logic of merge xdp process is simple, we can remove the
skip_xdp.

Signed-off-by: Xuan Zhuo <xuanzhuo@linux.alibaba.com>
---
 drivers/net/virtio_net.c | 23 ++++++++++-------------
 1 file changed, 10 insertions(+), 13 deletions(-)

diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
index 42e9927e316b..a4bb25f39f12 100644
--- a/drivers/net/virtio_net.c
+++ b/drivers/net/virtio_net.c
@@ -1390,7 +1390,6 @@ static struct sk_buff *receive_mergeable(struct net_device *dev,
 	struct page *page = virt_to_head_page(buf);
 	int offset = buf - page_address(page);
 	struct sk_buff *head_skb, *curr_skb;
-	struct bpf_prog *xdp_prog;
 	unsigned int truesize = mergeable_ctx_to_truesize(ctx);
 	unsigned int headroom = mergeable_ctx_to_headroom(ctx);
 	unsigned int tailroom = headroom ? sizeof(struct skb_shared_info) : 0;
@@ -1406,22 +1405,20 @@ static struct sk_buff *receive_mergeable(struct net_device *dev,
 		goto err_skb;
 	}
 
-	if (likely(!vi->xdp_enabled)) {
-		xdp_prog = NULL;
-		goto skip_xdp;
-	}
+	if (unlikely(vi->xdp_enabled)) {
+		struct bpf_prog *xdp_prog;
 
-	rcu_read_lock();
-	xdp_prog = rcu_dereference(rq->xdp_prog);
-	if (xdp_prog) {
-		head_skb = receive_mergeable_xdp(dev, vi, rq, xdp_prog, buf, ctx,
-						 len, xdp_xmit, stats);
+		rcu_read_lock();
+		xdp_prog = rcu_dereference(rq->xdp_prog);
+		if (xdp_prog) {
+			head_skb = receive_mergeable_xdp(dev, vi, rq, xdp_prog, buf, ctx,
+							 len, xdp_xmit, stats);
+			rcu_read_unlock();
+			return head_skb;
+		}
 		rcu_read_unlock();
-		return head_skb;
 	}
-	rcu_read_unlock();
 
-skip_xdp:
 	head_skb = page_to_skb(vi, rq, page, offset, len, truesize, headroom);
 	curr_skb = head_skb;
 
-- 
2.32.0.3.g01195cf9f

_______________________________________________
Virtualization mailing list
Virtualization@lists.linux-foundation.org
https://lists.linuxfoundation.org/mailman/listinfo/virtualization

[PATCH net-next v2 11/14] virtio_net: introduce receive_small_xdp()

[Xuan Zhuo]

The purpose of this patch is to simplify the receive_small().
Separate all the logic of XDP of small into a function.

Signed-off-by: Xuan Zhuo <xuanzhuo@linux.alibaba.com>
---
 drivers/net/virtio_net.c | 165 +++++++++++++++++++++++----------------
 1 file changed, 99 insertions(+), 66 deletions(-)

diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
index a4bb25f39f12..34220f5f27d1 100644
--- a/drivers/net/virtio_net.c
+++ b/drivers/net/virtio_net.c
@@ -941,6 +941,98 @@ static struct page *xdp_linearize_page(struct receive_queue *rq,
 	return NULL;
 }
 
+static struct sk_buff *receive_small_xdp(struct net_device *dev,
+					 struct virtnet_info *vi,
+					 struct receive_queue *rq,
+					 struct bpf_prog *xdp_prog,
+					 void *buf,
+					 unsigned int xdp_headroom,
+					 unsigned int len,
+					 unsigned int *xdp_xmit,
+					 struct virtnet_rq_stats *stats)
+{
+	unsigned int header_offset = VIRTNET_RX_PAD + xdp_headroom;
+	unsigned int headroom = vi->hdr_len + header_offset;
+	struct virtio_net_hdr_mrg_rxbuf *hdr = buf + header_offset;
+	struct page *page = virt_to_head_page(buf);
+	struct page *xdp_page;
+	unsigned int buflen;
+	struct xdp_buff xdp;
+	struct sk_buff *skb;
+	unsigned int delta = 0;
+	unsigned int metasize = 0;
+	void *orig_data;
+	u32 act;
+
+	if (unlikely(hdr->hdr.gso_type))
+		goto err_xdp;
+
+	buflen = SKB_DATA_ALIGN(GOOD_PACKET_LEN + headroom) +
+		SKB_DATA_ALIGN(sizeof(struct skb_shared_info));
+
+	if (unlikely(xdp_headroom < virtnet_get_headroom(vi))) {
+		int offset = buf - page_address(page) + header_offset;
+		unsigned int tlen = len + vi->hdr_len;
+		int num_buf = 1;
+
+		xdp_headroom = virtnet_get_headroom(vi);
+		header_offset = VIRTNET_RX_PAD + xdp_headroom;
+		headroom = vi->hdr_len + header_offset;
+		buflen = SKB_DATA_ALIGN(GOOD_PACKET_LEN + headroom) +
+			SKB_DATA_ALIGN(sizeof(struct skb_shared_info));
+		xdp_page = xdp_linearize_page(rq, &num_buf, page,
+					      offset, header_offset,
+					      &tlen);
+		if (!xdp_page)
+			goto err_xdp;
+
+		buf = page_address(xdp_page);
+		put_page(page);
+		page = xdp_page;
+	}
+
+	xdp_init_buff(&xdp, buflen, &rq->xdp_rxq);
+	xdp_prepare_buff(&xdp, buf + VIRTNET_RX_PAD + vi->hdr_len,
+			 xdp_headroom, len, true);
+	orig_data = xdp.data;
+
+	act = virtnet_xdp_handler(xdp_prog, &xdp, dev, xdp_xmit, stats);
+
+	switch (act) {
+	case VIRTNET_XDP_RES_PASS:
+		/* Recalculate length in case bpf program changed it */
+		delta = orig_data - xdp.data;
+		len = xdp.data_end - xdp.data;
+		metasize = xdp.data - xdp.data_meta;
+		break;
+
+	case VIRTNET_XDP_RES_CONSUMED:
+		goto xdp_xmit;
+
+	case VIRTNET_XDP_RES_DROP:
+		goto err_xdp;
+	}
+
+	skb = build_skb(buf, buflen);
+	if (!skb)
+		goto err;
+
+	skb_reserve(skb, headroom - delta);
+	skb_put(skb, len);
+	if (metasize)
+		skb_metadata_set(skb, metasize);
+
+	return skb;
+
+err_xdp:
+	stats->xdp_drops++;
+err:
+	stats->drops++;
+	put_page(page);
+xdp_xmit:
+	return NULL;
+}
+
 static struct sk_buff *receive_small(struct net_device *dev,
 				     struct virtnet_info *vi,
 				     struct receive_queue *rq,
@@ -957,9 +1049,6 @@ static struct sk_buff *receive_small(struct net_device *dev,
 	unsigned int buflen = SKB_DATA_ALIGN(GOOD_PACKET_LEN + headroom) +
 			      SKB_DATA_ALIGN(sizeof(struct skb_shared_info));
 	struct page *page = virt_to_head_page(buf);
-	unsigned int delta = 0;
-	struct page *xdp_page;
-	unsigned int metasize = 0;
 
 	len -= vi->hdr_len;
 	stats->bytes += len;
@@ -979,57 +1068,10 @@ static struct sk_buff *receive_small(struct net_device *dev,
 	rcu_read_lock();
 	xdp_prog = rcu_dereference(rq->xdp_prog);
 	if (xdp_prog) {
-		struct virtio_net_hdr_mrg_rxbuf *hdr = buf + header_offset;
-		struct xdp_buff xdp;
-		void *orig_data;
-		u32 act;
-
-		if (unlikely(hdr->hdr.gso_type))
-			goto err_xdp;
-
-		if (unlikely(xdp_headroom < virtnet_get_headroom(vi))) {
-			int offset = buf - page_address(page) + header_offset;
-			unsigned int tlen = len + vi->hdr_len;
-			int num_buf = 1;
-
-			xdp_headroom = virtnet_get_headroom(vi);
-			header_offset = VIRTNET_RX_PAD + xdp_headroom;
-			headroom = vi->hdr_len + header_offset;
-			buflen = SKB_DATA_ALIGN(GOOD_PACKET_LEN + headroom) +
-				 SKB_DATA_ALIGN(sizeof(struct skb_shared_info));
-			xdp_page = xdp_linearize_page(rq, &num_buf, page,
-						      offset, header_offset,
-						      &tlen);
-			if (!xdp_page)
-				goto err_xdp;
-
-			buf = page_address(xdp_page);
-			put_page(page);
-			page = xdp_page;
-		}
-
-		xdp_init_buff(&xdp, buflen, &rq->xdp_rxq);
-		xdp_prepare_buff(&xdp, buf + VIRTNET_RX_PAD + vi->hdr_len,
-				 xdp_headroom, len, true);
-		orig_data = xdp.data;
-
-		act = virtnet_xdp_handler(xdp_prog, &xdp, dev, xdp_xmit, stats);
-
-		switch (act) {
-		case VIRTNET_XDP_RES_PASS:
-			/* Recalculate length in case bpf program changed it */
-			delta = orig_data - xdp.data;
-			len = xdp.data_end - xdp.data;
-			metasize = xdp.data - xdp.data_meta;
-			break;
-
-		case VIRTNET_XDP_RES_CONSUMED:
-			rcu_read_unlock();
-			goto xdp_xmit;
-
-		case VIRTNET_XDP_RES_DROP:
-			goto err_xdp;
-		}
+		skb = receive_small_xdp(dev, vi, rq, xdp_prog, buf, xdp_headroom,
+					len, xdp_xmit, stats);
+		rcu_read_unlock();
+		return skb;
 	}
 	rcu_read_unlock();
 
@@ -1037,25 +1079,16 @@ static struct sk_buff *receive_small(struct net_device *dev,
 	skb = build_skb(buf, buflen);
 	if (!skb)
 		goto err;
-	skb_reserve(skb, headroom - delta);
+	skb_reserve(skb, headroom);
 	skb_put(skb, len);
-	if (!xdp_prog) {
-		buf += header_offset;
-		memcpy(skb_vnet_hdr(skb), buf, vi->hdr_len);
-	} /* keep zeroed vnet hdr since XDP is loaded */
-
-	if (metasize)
-		skb_metadata_set(skb, metasize);
 
+	buf += header_offset;
+	memcpy(skb_vnet_hdr(skb), buf, vi->hdr_len);
 	return skb;
 
-err_xdp:
-	rcu_read_unlock();
-	stats->xdp_drops++;
 err:
 	stats->drops++;
 	put_page(page);
-xdp_xmit:
 	return NULL;
 }
 
-- 
2.32.0.3.g01195cf9f

_______________________________________________
Virtualization mailing list
Virtualization@lists.linux-foundation.org
https://lists.linuxfoundation.org/mailman/listinfo/virtualization

[PATCH net-next v2 12/14] virtio_net: small: optimize code

[Xuan Zhuo]

In the case of XDP-PASS, skb_reserve uses the delta to compatible
non-XDP, now remove this logic.

Signed-off-by: Xuan Zhuo <xuanzhuo@linux.alibaba.com>
---
 drivers/net/virtio_net.c | 6 +-----
 1 file changed, 1 insertion(+), 5 deletions(-)

diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
index 34220f5f27d1..f6f5903face2 100644
--- a/drivers/net/virtio_net.c
+++ b/drivers/net/virtio_net.c
@@ -959,9 +959,7 @@ static struct sk_buff *receive_small_xdp(struct net_device *dev,
 	unsigned int buflen;
 	struct xdp_buff xdp;
 	struct sk_buff *skb;
-	unsigned int delta = 0;
 	unsigned int metasize = 0;
-	void *orig_data;
 	u32 act;
 
 	if (unlikely(hdr->hdr.gso_type))
@@ -994,14 +992,12 @@ static struct sk_buff *receive_small_xdp(struct net_device *dev,
 	xdp_init_buff(&xdp, buflen, &rq->xdp_rxq);
 	xdp_prepare_buff(&xdp, buf + VIRTNET_RX_PAD + vi->hdr_len,
 			 xdp_headroom, len, true);
-	orig_data = xdp.data;
 
 	act = virtnet_xdp_handler(xdp_prog, &xdp, dev, xdp_xmit, stats);
 
 	switch (act) {
 	case VIRTNET_XDP_RES_PASS:
 		/* Recalculate length in case bpf program changed it */
-		delta = orig_data - xdp.data;
 		len = xdp.data_end - xdp.data;
 		metasize = xdp.data - xdp.data_meta;
 		break;
@@ -1017,7 +1013,7 @@ static struct sk_buff *receive_small_xdp(struct net_device *dev,
 	if (!skb)
 		goto err;
 
-	skb_reserve(skb, headroom - delta);
+	skb_reserve(skb, xdp.data - buf);
 	skb_put(skb, len);
 	if (metasize)
 		skb_metadata_set(skb, metasize);
-- 
2.32.0.3.g01195cf9f

_______________________________________________
Virtualization mailing list
Virtualization@lists.linux-foundation.org
https://lists.linuxfoundation.org/mailman/listinfo/virtualization

[PATCH net-next v2 13/14] virtio_net: small: optimize code

[Xuan Zhuo]

Avoid the problem that some variables(headroom and so on) will repeat
the calculation when process xdp.

Signed-off-by: Xuan Zhuo <xuanzhuo@linux.alibaba.com>
---
 drivers/net/virtio_net.c | 12 ++++++++----
 1 file changed, 8 insertions(+), 4 deletions(-)

diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
index f6f5903face2..5a5636178bd3 100644
--- a/drivers/net/virtio_net.c
+++ b/drivers/net/virtio_net.c
@@ -1040,11 +1040,10 @@ static struct sk_buff *receive_small(struct net_device *dev,
 	struct sk_buff *skb;
 	struct bpf_prog *xdp_prog;
 	unsigned int xdp_headroom = (unsigned long)ctx;
-	unsigned int header_offset = VIRTNET_RX_PAD + xdp_headroom;
-	unsigned int headroom = vi->hdr_len + header_offset;
-	unsigned int buflen = SKB_DATA_ALIGN(GOOD_PACKET_LEN + headroom) +
-			      SKB_DATA_ALIGN(sizeof(struct skb_shared_info));
 	struct page *page = virt_to_head_page(buf);
+	unsigned int header_offset;
+	unsigned int headroom;
+	unsigned int buflen;
 
 	len -= vi->hdr_len;
 	stats->bytes += len;
@@ -1072,6 +1071,11 @@ static struct sk_buff *receive_small(struct net_device *dev,
 	rcu_read_unlock();
 
 skip_xdp:
+	header_offset = VIRTNET_RX_PAD + xdp_headroom;
+	headroom = vi->hdr_len + header_offset;
+	buflen = SKB_DATA_ALIGN(GOOD_PACKET_LEN + headroom) +
+		SKB_DATA_ALIGN(sizeof(struct skb_shared_info));
+
 	skb = build_skb(buf, buflen);
 	if (!skb)
 		goto err;
-- 
2.32.0.3.g01195cf9f

_______________________________________________
Virtualization mailing list
Virtualization@lists.linux-foundation.org
https://lists.linuxfoundation.org/mailman/listinfo/virtualization

[PATCH net-next v2 14/14] virtio_net: small: remove skip_xdp

[Xuan Zhuo]

now, the process of xdp is simple, we can remove the skip_xdp.

Signed-off-by: Xuan Zhuo <xuanzhuo@linux.alibaba.com>
---
 drivers/net/virtio_net.c | 26 ++++++++++++--------------
 1 file changed, 12 insertions(+), 14 deletions(-)

diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
index 5a5636178bd3..19f7a8367c17 100644
--- a/drivers/net/virtio_net.c
+++ b/drivers/net/virtio_net.c
@@ -1037,13 +1037,12 @@ static struct sk_buff *receive_small(struct net_device *dev,
 				     unsigned int *xdp_xmit,
 				     struct virtnet_rq_stats *stats)
 {
-	struct sk_buff *skb;
-	struct bpf_prog *xdp_prog;
 	unsigned int xdp_headroom = (unsigned long)ctx;
 	struct page *page = virt_to_head_page(buf);
 	unsigned int header_offset;
 	unsigned int headroom;
 	unsigned int buflen;
+	struct sk_buff *skb;
 
 	len -= vi->hdr_len;
 	stats->bytes += len;
@@ -1055,22 +1054,21 @@ static struct sk_buff *receive_small(struct net_device *dev,
 		goto err;
 	}
 
-	if (likely(!vi->xdp_enabled)) {
-		xdp_prog = NULL;
-		goto skip_xdp;
-	}
+	if (unlikely(vi->xdp_enabled)) {
+		struct bpf_prog *xdp_prog;
 
-	rcu_read_lock();
-	xdp_prog = rcu_dereference(rq->xdp_prog);
-	if (xdp_prog) {
-		skb = receive_small_xdp(dev, vi, rq, xdp_prog, buf, xdp_headroom,
-					len, xdp_xmit, stats);
+		rcu_read_lock();
+		xdp_prog = rcu_dereference(rq->xdp_prog);
+		if (xdp_prog) {
+			skb = receive_small_xdp(dev, vi, rq, xdp_prog, buf,
+						xdp_headroom, len, xdp_xmit,
+						stats);
+			rcu_read_unlock();
+			return skb;
+		}
 		rcu_read_unlock();
-		return skb;
 	}
-	rcu_read_unlock();
 
-skip_xdp:
 	header_offset = VIRTNET_RX_PAD + xdp_headroom;
 	headroom = vi->hdr_len + header_offset;
 	buflen = SKB_DATA_ALIGN(GOOD_PACKET_LEN + headroom) +
-- 
2.32.0.3.g01195cf9f

_______________________________________________
Virtualization mailing list
Virtualization@lists.linux-foundation.org
https://lists.linuxfoundation.org/mailman/listinfo/virtualization

Re: [PATCH net-next v2 00/14] virtio_net: refactor xdp codes

[Michael S. Tsirkin]

On Tue, Apr 18, 2023 at 02:53:13PM +0800, Xuan Zhuo wrote:
> Due to historical reasons, the implementation of XDP in virtio-net is relatively
> chaotic. For example, the processing of XDP actions has two copies of similar
> code. Such as page, xdp_page processing, etc.
> 
> The purpose of this patch set is to refactor these code. Reduce the difficulty
> of subsequent maintenance. Subsequent developers will not introduce new bugs
> because of some complex logical relationships.
> 
> In addition, the supporting to AF_XDP that I want to submit later will also need
> to reuse the logic of XDP, such as the processing of actions, I don't want to
> introduce a new similar code. In this way, I can reuse these codes in the
> future.
> 
> Please review.
> 
> Thanks.

Big refactoring, pls allow a bit more time for review. Thanks!

> v2:
>     1. re-split to make review more convenient
> 
> v1:
>     1. fix some variables are uninitialized
> 
> Xuan Zhuo (14):
>   virtio_net: mergeable xdp: put old page immediately
>   virtio_net: introduce mergeable_xdp_prepare()
>   virtio_net: optimize mergeable_xdp_prepare()
>   virtio_net: introduce virtnet_xdp_handler() to seprate the logic of
>     run xdp
>   virtio_net: introduce xdp res enums
>   virtio_net: separate the logic of freeing xdp shinfo
>   virtio_net: separate the logic of freeing the rest mergeable buf
>   virtio_net: auto release xdp shinfo
>   virtio_net: introduce receive_mergeable_xdp()
>   virtio_net: merge: remove skip_xdp
>   virtio_net: introduce receive_small_xdp()
>   virtio_net: small: optimize code
>   virtio_net: small: optimize code
>   virtio_net: small: remove skip_xdp
> 
>  drivers/net/virtio_net.c | 625 +++++++++++++++++++++++----------------
>  1 file changed, 362 insertions(+), 263 deletions(-)
> 
> --
> 2.32.0.3.g01195cf9f

_______________________________________________
Virtualization mailing list
Virtualization@lists.linux-foundation.org
https://lists.linuxfoundation.org/mailman/listinfo/virtualization

Re: [PATCH net-next v2 01/14] virtio_net: mergeable xdp: put old page immediately

[Jason Wang]

On Tue, Apr 18, 2023 at 2:53 PM Xuan Zhuo <xuanzhuo@linux.alibaba.com> wrote:
>
> In the xdp implementation of virtio-net mergeable, it always checks
> whether two page is used and a page is selected to release. This is
> complicated for the processing of action, and be careful.
>
> In the entire process, we have such principles:
> * If xdp_page is used (PASS, TX, Redirect), then we release the old
>   page.
> * If it is a drop case, we will release two. The old page obtained from
>   buf is release inside err_xdp, and xdp_page needs be relased by us.
>
> But in fact, when we allocate a new page, we can release the old page
> immediately. Then just one is using, we just need to release the new
> page for drop case. On the drop path, err_xdp will release the variable
> "page", so we only need to let "page" point to the new xdp_page in
> advance.
>
> Signed-off-by: Xuan Zhuo <xuanzhuo@linux.alibaba.com>

Acked-by: Jason Wang <jasowang@redhat.com>

Thanks


> ---
>  drivers/net/virtio_net.c | 19 +++++++------------
>  1 file changed, 7 insertions(+), 12 deletions(-)
>
> diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
> index e2560b6f7980..42435e762d72 100644
> --- a/drivers/net/virtio_net.c
> +++ b/drivers/net/virtio_net.c
> @@ -1245,6 +1245,9 @@ static struct sk_buff *receive_mergeable(struct net_device *dev,
>                         if (!xdp_page)
>                                 goto err_xdp;
>                         offset = VIRTIO_XDP_HEADROOM;
> +
> +                       put_page(page);
> +                       page = xdp_page;
>                 } else if (unlikely(headroom < virtnet_get_headroom(vi))) {
>                         xdp_room = SKB_DATA_ALIGN(VIRTIO_XDP_HEADROOM +
>                                                   sizeof(struct skb_shared_info));
> @@ -1259,11 +1262,12 @@ static struct sk_buff *receive_mergeable(struct net_device *dev,
>                                page_address(page) + offset, len);
>                         frame_sz = PAGE_SIZE;
>                         offset = VIRTIO_XDP_HEADROOM;
> -               } else {
> -                       xdp_page = page;
> +
> +                       put_page(page);
> +                       page = xdp_page;
>                 }
>
> -               data = page_address(xdp_page) + offset;
> +               data = page_address(page) + offset;
>                 err = virtnet_build_xdp_buff_mrg(dev, vi, rq, &xdp, data, len, frame_sz,
>                                                  &num_buf, &xdp_frags_truesz, stats);
>                 if (unlikely(err))
> @@ -1278,8 +1282,6 @@ static struct sk_buff *receive_mergeable(struct net_device *dev,
>                         if (unlikely(!head_skb))
>                                 goto err_xdp_frags;
>
> -                       if (unlikely(xdp_page != page))
> -                               put_page(page);
>                         rcu_read_unlock();
>                         return head_skb;
>                 case XDP_TX:
> @@ -1297,8 +1299,6 @@ static struct sk_buff *receive_mergeable(struct net_device *dev,
>                                 goto err_xdp_frags;
>                         }
>                         *xdp_xmit |= VIRTIO_XDP_TX;
> -                       if (unlikely(xdp_page != page))
> -                               put_page(page);
>                         rcu_read_unlock();
>                         goto xdp_xmit;
>                 case XDP_REDIRECT:
> @@ -1307,8 +1307,6 @@ static struct sk_buff *receive_mergeable(struct net_device *dev,
>                         if (err)
>                                 goto err_xdp_frags;
>                         *xdp_xmit |= VIRTIO_XDP_REDIR;
> -                       if (unlikely(xdp_page != page))
> -                               put_page(page);
>                         rcu_read_unlock();
>                         goto xdp_xmit;
>                 default:
> @@ -1321,9 +1319,6 @@ static struct sk_buff *receive_mergeable(struct net_device *dev,
>                         goto err_xdp_frags;
>                 }
>  err_xdp_frags:
> -               if (unlikely(xdp_page != page))
> -                       __free_pages(xdp_page, 0);
> -
>                 if (xdp_buff_has_frags(&xdp)) {
>                         shinfo = xdp_get_shared_info_from_buff(&xdp);
>                         for (i = 0; i < shinfo->nr_frags; i++) {
> --
> 2.32.0.3.g01195cf9f
>

_______________________________________________
Virtualization mailing list
Virtualization@lists.linux-foundation.org
https://lists.linuxfoundation.org/mailman/listinfo/virtualization

Re: [PATCH net-next v2 02/14] virtio_net: introduce mergeable_xdp_prepare()

[Jason Wang]

On Tue, Apr 18, 2023 at 2:53 PM Xuan Zhuo <xuanzhuo@linux.alibaba.com> wrote:
>
> Separating the logic of preparation for xdp from receive_mergeable.
>
> The purpose of this is to simplify the logic of execution of XDP.
>
> The main logic here is that when headroom is insufficient, we need to
> allocate a new page and calculate offset. It should be noted that if
> there is new page, the variable page will refer to the new page.
>
> Signed-off-by: Xuan Zhuo <xuanzhuo@linux.alibaba.com>
> ---
>  drivers/net/virtio_net.c | 135 +++++++++++++++++++++++----------------
>  1 file changed, 79 insertions(+), 56 deletions(-)
>
> diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
> index 42435e762d72..12559062ffb6 100644
> --- a/drivers/net/virtio_net.c
> +++ b/drivers/net/virtio_net.c
> @@ -1162,6 +1162,81 @@ static int virtnet_build_xdp_buff_mrg(struct net_device *dev,
>         return 0;
>  }
>
> +static void *mergeable_xdp_prepare(struct virtnet_info *vi,
> +                                  struct receive_queue *rq,
> +                                  struct bpf_prog *xdp_prog,
> +                                  void *ctx,
> +                                  unsigned int *frame_sz,
> +                                  int *num_buf,
> +                                  struct page **page,
> +                                  int offset,
> +                                  unsigned int *len,
> +                                  struct virtio_net_hdr_mrg_rxbuf *hdr)

Nit: I think we probably need a better name. (mergeable_xdp_get_buf()?)

Other than this:

Acked-by: Jason Wang <jasowang@redhat.com>

Thanks


> +{
> +       unsigned int truesize = mergeable_ctx_to_truesize(ctx);
> +       unsigned int headroom = mergeable_ctx_to_headroom(ctx);
> +       struct page *xdp_page;
> +       unsigned int xdp_room;
> +
> +       /* Transient failure which in theory could occur if
> +        * in-flight packets from before XDP was enabled reach
> +        * the receive path after XDP is loaded.
> +        */
> +       if (unlikely(hdr->hdr.gso_type))
> +               return NULL;
> +
> +       /* Now XDP core assumes frag size is PAGE_SIZE, but buffers
> +        * with headroom may add hole in truesize, which
> +        * make their length exceed PAGE_SIZE. So we disabled the
> +        * hole mechanism for xdp. See add_recvbuf_mergeable().
> +        */
> +       *frame_sz = truesize;
> +
> +       /* This happens when headroom is not enough because
> +        * of the buffer was prefilled before XDP is set.
> +        * This should only happen for the first several packets.
> +        * In fact, vq reset can be used here to help us clean up
> +        * the prefilled buffers, but many existing devices do not
> +        * support it, and we don't want to bother users who are
> +        * using xdp normally.
> +        */
> +       if (!xdp_prog->aux->xdp_has_frags &&
> +           (*num_buf > 1 || headroom < virtnet_get_headroom(vi))) {
> +               /* linearize data for XDP */
> +               xdp_page = xdp_linearize_page(rq, num_buf,
> +                                             *page, offset,
> +                                             VIRTIO_XDP_HEADROOM,
> +                                             len);
> +               *frame_sz = PAGE_SIZE;
> +
> +               if (!xdp_page)
> +                       return NULL;
> +               offset = VIRTIO_XDP_HEADROOM;
> +
> +               put_page(*page);
> +               *page = xdp_page;
> +       } else if (unlikely(headroom < virtnet_get_headroom(vi))) {
> +               xdp_room = SKB_DATA_ALIGN(VIRTIO_XDP_HEADROOM +
> +                                         sizeof(struct skb_shared_info));
> +               if (*len + xdp_room > PAGE_SIZE)
> +                       return NULL;
> +
> +               xdp_page = alloc_page(GFP_ATOMIC);
> +               if (!xdp_page)
> +                       return NULL;
> +
> +               memcpy(page_address(xdp_page) + VIRTIO_XDP_HEADROOM,
> +                      page_address(*page) + offset, *len);
> +               *frame_sz = PAGE_SIZE;
> +               offset = VIRTIO_XDP_HEADROOM;
> +
> +               put_page(*page);
> +               *page = xdp_page;
> +       }
> +
> +       return page_address(*page) + offset;
> +}
> +
>  static struct sk_buff *receive_mergeable(struct net_device *dev,
>                                          struct virtnet_info *vi,
>                                          struct receive_queue *rq,
> @@ -1181,7 +1256,7 @@ static struct sk_buff *receive_mergeable(struct net_device *dev,
>         unsigned int headroom = mergeable_ctx_to_headroom(ctx);
>         unsigned int tailroom = headroom ? sizeof(struct skb_shared_info) : 0;
>         unsigned int room = SKB_DATA_ALIGN(headroom + tailroom);
> -       unsigned int frame_sz, xdp_room;
> +       unsigned int frame_sz;
>         int err;
>
>         head_skb = NULL;
> @@ -1211,63 +1286,11 @@ static struct sk_buff *receive_mergeable(struct net_device *dev,
>                 u32 act;
>                 int i;
>
> -               /* Transient failure which in theory could occur if
> -                * in-flight packets from before XDP was enabled reach
> -                * the receive path after XDP is loaded.
> -                */
> -               if (unlikely(hdr->hdr.gso_type))
> +               data = mergeable_xdp_prepare(vi, rq, xdp_prog, ctx, &frame_sz,
> +                                            &num_buf, &page, offset, &len, hdr);
> +               if (unlikely(!data))
>                         goto err_xdp;
>
> -               /* Now XDP core assumes frag size is PAGE_SIZE, but buffers
> -                * with headroom may add hole in truesize, which
> -                * make their length exceed PAGE_SIZE. So we disabled the
> -                * hole mechanism for xdp. See add_recvbuf_mergeable().
> -                */
> -               frame_sz = truesize;
> -
> -               /* This happens when headroom is not enough because
> -                * of the buffer was prefilled before XDP is set.
> -                * This should only happen for the first several packets.
> -                * In fact, vq reset can be used here to help us clean up
> -                * the prefilled buffers, but many existing devices do not
> -                * support it, and we don't want to bother users who are
> -                * using xdp normally.
> -                */
> -               if (!xdp_prog->aux->xdp_has_frags &&
> -                   (num_buf > 1 || headroom < virtnet_get_headroom(vi))) {
> -                       /* linearize data for XDP */
> -                       xdp_page = xdp_linearize_page(rq, &num_buf,
> -                                                     page, offset,
> -                                                     VIRTIO_XDP_HEADROOM,
> -                                                     &len);
> -                       frame_sz = PAGE_SIZE;
> -
> -                       if (!xdp_page)
> -                               goto err_xdp;
> -                       offset = VIRTIO_XDP_HEADROOM;
> -
> -                       put_page(page);
> -                       page = xdp_page;
> -               } else if (unlikely(headroom < virtnet_get_headroom(vi))) {
> -                       xdp_room = SKB_DATA_ALIGN(VIRTIO_XDP_HEADROOM +
> -                                                 sizeof(struct skb_shared_info));
> -                       if (len + xdp_room > PAGE_SIZE)
> -                               goto err_xdp;
> -
> -                       xdp_page = alloc_page(GFP_ATOMIC);
> -                       if (!xdp_page)
> -                               goto err_xdp;
> -
> -                       memcpy(page_address(xdp_page) + VIRTIO_XDP_HEADROOM,
> -                              page_address(page) + offset, len);
> -                       frame_sz = PAGE_SIZE;
> -                       offset = VIRTIO_XDP_HEADROOM;
> -
> -                       put_page(page);
> -                       page = xdp_page;
> -               }
> -
> -               data = page_address(page) + offset;
>                 err = virtnet_build_xdp_buff_mrg(dev, vi, rq, &xdp, data, len, frame_sz,
>                                                  &num_buf, &xdp_frags_truesz, stats);
>                 if (unlikely(err))
> --
> 2.32.0.3.g01195cf9f
>

_______________________________________________
Virtualization mailing list
Virtualization@lists.linux-foundation.org
https://lists.linuxfoundation.org/mailman/listinfo/virtualization

Re: [PATCH net-next v2 03/14] virtio_net: optimize mergeable_xdp_prepare()

[Jason Wang]

On Tue, Apr 18, 2023 at 2:53 PM Xuan Zhuo <xuanzhuo@linux.alibaba.com> wrote:
>
> The previous patch, in order to facilitate review, I do not do any
> modification. This patch has made some optimization on the top.
>
> * remove some repeated logics in this function.
> * add fast check for passing without any alloc.
>
> Signed-off-by: Xuan Zhuo <xuanzhuo@linux.alibaba.com>

Acked-by: Jason Wang <jasowang@redhat.com>

Thanks


> ---
>  drivers/net/virtio_net.c | 29 ++++++++++++++---------------
>  1 file changed, 14 insertions(+), 15 deletions(-)
>
> diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
> index 12559062ffb6..50dc64d80d3b 100644
> --- a/drivers/net/virtio_net.c
> +++ b/drivers/net/virtio_net.c
> @@ -1192,6 +1192,11 @@ static void *mergeable_xdp_prepare(struct virtnet_info *vi,
>          */
>         *frame_sz = truesize;
>
> +       if (likely(headroom >= virtnet_get_headroom(vi) &&
> +                  (*num_buf == 1 || xdp_prog->aux->xdp_has_frags))) {
> +               return page_address(*page) + offset;
> +       }
> +
>         /* This happens when headroom is not enough because
>          * of the buffer was prefilled before XDP is set.
>          * This should only happen for the first several packets.
> @@ -1200,22 +1205,15 @@ static void *mergeable_xdp_prepare(struct virtnet_info *vi,
>          * support it, and we don't want to bother users who are
>          * using xdp normally.
>          */
> -       if (!xdp_prog->aux->xdp_has_frags &&
> -           (*num_buf > 1 || headroom < virtnet_get_headroom(vi))) {
> +       if (!xdp_prog->aux->xdp_has_frags) {
>                 /* linearize data for XDP */
>                 xdp_page = xdp_linearize_page(rq, num_buf,
>                                               *page, offset,
>                                               VIRTIO_XDP_HEADROOM,
>                                               len);
> -               *frame_sz = PAGE_SIZE;
> -
>                 if (!xdp_page)
>                         return NULL;
> -               offset = VIRTIO_XDP_HEADROOM;
> -
> -               put_page(*page);
> -               *page = xdp_page;
> -       } else if (unlikely(headroom < virtnet_get_headroom(vi))) {
> +       } else {
>                 xdp_room = SKB_DATA_ALIGN(VIRTIO_XDP_HEADROOM +
>                                           sizeof(struct skb_shared_info));
>                 if (*len + xdp_room > PAGE_SIZE)
> @@ -1227,14 +1225,15 @@ static void *mergeable_xdp_prepare(struct virtnet_info *vi,
>
>                 memcpy(page_address(xdp_page) + VIRTIO_XDP_HEADROOM,
>                        page_address(*page) + offset, *len);
> -               *frame_sz = PAGE_SIZE;
> -               offset = VIRTIO_XDP_HEADROOM;
> -
> -               put_page(*page);
> -               *page = xdp_page;
>         }
>
> -       return page_address(*page) + offset;
> +       *frame_sz = PAGE_SIZE;
> +
> +       put_page(*page);
> +
> +       *page = xdp_page;
> +
> +       return page_address(*page) + VIRTIO_XDP_HEADROOM;
>  }
>
>  static struct sk_buff *receive_mergeable(struct net_device *dev,
> --
> 2.32.0.3.g01195cf9f
>

_______________________________________________
Virtualization mailing list
Virtualization@lists.linux-foundation.org
https://lists.linuxfoundation.org/mailman/listinfo/virtualization

Re: [PATCH net-next v2 04/14] virtio_net: introduce virtnet_xdp_handler() to seprate the logic of run xdp

[Jason Wang]

On Tue, Apr 18, 2023 at 2:53 PM Xuan Zhuo <xuanzhuo@linux.alibaba.com> wrote:
>
> At present, we have two similar logic to perform the XDP prog.
>
> Therefore, this PATCH separates the code of executing XDP, which is

Any reason for using upper case for "PATCH" ?


> conducive to later maintenance.
>
> Signed-off-by: Xuan Zhuo <xuanzhuo@linux.alibaba.com>

Acked-by: Jason Wang <jasowang@redhat.com>

Thanks


> ---
>  drivers/net/virtio_net.c | 118 +++++++++++++++++++--------------------
>  1 file changed, 58 insertions(+), 60 deletions(-)
>
> diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
> index 50dc64d80d3b..0fa64c314ea7 100644
> --- a/drivers/net/virtio_net.c
> +++ b/drivers/net/virtio_net.c
> @@ -789,6 +789,60 @@ static int virtnet_xdp_xmit(struct net_device *dev,
>         return ret;
>  }
>
> +static int virtnet_xdp_handler(struct bpf_prog *xdp_prog, struct xdp_buff *xdp,
> +                              struct net_device *dev,
> +                              unsigned int *xdp_xmit,
> +                              struct virtnet_rq_stats *stats)
> +{
> +       struct xdp_frame *xdpf;
> +       int err;
> +       u32 act;
> +
> +       act = bpf_prog_run_xdp(xdp_prog, xdp);
> +       stats->xdp_packets++;
> +
> +       switch (act) {
> +       case XDP_PASS:
> +               return act;
> +
> +       case XDP_TX:
> +               stats->xdp_tx++;
> +               xdpf = xdp_convert_buff_to_frame(xdp);
> +               if (unlikely(!xdpf)) {
> +                       netdev_dbg(dev, "convert buff to frame failed for xdp\n");
> +                       return XDP_DROP;
> +               }
> +
> +               err = virtnet_xdp_xmit(dev, 1, &xdpf, 0);
> +               if (unlikely(!err)) {
> +                       xdp_return_frame_rx_napi(xdpf);
> +               } else if (unlikely(err < 0)) {
> +                       trace_xdp_exception(dev, xdp_prog, act);
> +                       return XDP_DROP;
> +               }
> +               *xdp_xmit |= VIRTIO_XDP_TX;
> +               return act;
> +
> +       case XDP_REDIRECT:
> +               stats->xdp_redirects++;
> +               err = xdp_do_redirect(dev, xdp, xdp_prog);
> +               if (err)
> +                       return XDP_DROP;
> +
> +               *xdp_xmit |= VIRTIO_XDP_REDIR;
> +               return act;
> +
> +       default:
> +               bpf_warn_invalid_xdp_action(dev, xdp_prog, act);
> +               fallthrough;
> +       case XDP_ABORTED:
> +               trace_xdp_exception(dev, xdp_prog, act);
> +               fallthrough;
> +       case XDP_DROP:
> +               return XDP_DROP;
> +       }
> +}
> +
>  static unsigned int virtnet_get_headroom(struct virtnet_info *vi)
>  {
>         return vi->xdp_enabled ? VIRTIO_XDP_HEADROOM : 0;
> @@ -876,7 +930,6 @@ static struct sk_buff *receive_small(struct net_device *dev,
>         struct page *page = virt_to_head_page(buf);
>         unsigned int delta = 0;
>         struct page *xdp_page;
> -       int err;
>         unsigned int metasize = 0;
>
>         len -= vi->hdr_len;
> @@ -898,7 +951,6 @@ static struct sk_buff *receive_small(struct net_device *dev,
>         xdp_prog = rcu_dereference(rq->xdp_prog);
>         if (xdp_prog) {
>                 struct virtio_net_hdr_mrg_rxbuf *hdr = buf + header_offset;
> -               struct xdp_frame *xdpf;
>                 struct xdp_buff xdp;
>                 void *orig_data;
>                 u32 act;
> @@ -931,8 +983,8 @@ static struct sk_buff *receive_small(struct net_device *dev,
>                 xdp_prepare_buff(&xdp, buf + VIRTNET_RX_PAD + vi->hdr_len,
>                                  xdp_headroom, len, true);
>                 orig_data = xdp.data;
> -               act = bpf_prog_run_xdp(xdp_prog, &xdp);
> -               stats->xdp_packets++;
> +
> +               act = virtnet_xdp_handler(xdp_prog, &xdp, dev, xdp_xmit, stats);
>
>                 switch (act) {
>                 case XDP_PASS:
> @@ -942,35 +994,10 @@ static struct sk_buff *receive_small(struct net_device *dev,
>                         metasize = xdp.data - xdp.data_meta;
>                         break;
>                 case XDP_TX:
> -                       stats->xdp_tx++;
> -                       xdpf = xdp_convert_buff_to_frame(&xdp);
> -                       if (unlikely(!xdpf))
> -                               goto err_xdp;
> -                       err = virtnet_xdp_xmit(dev, 1, &xdpf, 0);
> -                       if (unlikely(!err)) {
> -                               xdp_return_frame_rx_napi(xdpf);
> -                       } else if (unlikely(err < 0)) {
> -                               trace_xdp_exception(vi->dev, xdp_prog, act);
> -                               goto err_xdp;
> -                       }
> -                       *xdp_xmit |= VIRTIO_XDP_TX;
> -                       rcu_read_unlock();
> -                       goto xdp_xmit;
>                 case XDP_REDIRECT:
> -                       stats->xdp_redirects++;
> -                       err = xdp_do_redirect(dev, &xdp, xdp_prog);
> -                       if (err)
> -                               goto err_xdp;
> -                       *xdp_xmit |= VIRTIO_XDP_REDIR;
>                         rcu_read_unlock();
>                         goto xdp_xmit;
>                 default:
> -                       bpf_warn_invalid_xdp_action(vi->dev, xdp_prog, act);
> -                       fallthrough;
> -               case XDP_ABORTED:
> -                       trace_xdp_exception(vi->dev, xdp_prog, act);
> -                       goto err_xdp;
> -               case XDP_DROP:
>                         goto err_xdp;
>                 }
>         }
> @@ -1278,7 +1305,6 @@ static struct sk_buff *receive_mergeable(struct net_device *dev,
>         if (xdp_prog) {
>                 unsigned int xdp_frags_truesz = 0;
>                 struct skb_shared_info *shinfo;
> -               struct xdp_frame *xdpf;
>                 struct page *xdp_page;
>                 struct xdp_buff xdp;
>                 void *data;
> @@ -1295,8 +1321,7 @@ static struct sk_buff *receive_mergeable(struct net_device *dev,
>                 if (unlikely(err))
>                         goto err_xdp_frags;
>
> -               act = bpf_prog_run_xdp(xdp_prog, &xdp);
> -               stats->xdp_packets++;
> +               act = virtnet_xdp_handler(xdp_prog, &xdp, dev, xdp_xmit, stats);
>
>                 switch (act) {
>                 case XDP_PASS:
> @@ -1307,38 +1332,11 @@ static struct sk_buff *receive_mergeable(struct net_device *dev,
>                         rcu_read_unlock();
>                         return head_skb;
>                 case XDP_TX:
> -                       stats->xdp_tx++;
> -                       xdpf = xdp_convert_buff_to_frame(&xdp);
> -                       if (unlikely(!xdpf)) {
> -                               netdev_dbg(dev, "convert buff to frame failed for xdp\n");
> -                               goto err_xdp_frags;
> -                       }
> -                       err = virtnet_xdp_xmit(dev, 1, &xdpf, 0);
> -                       if (unlikely(!err)) {
> -                               xdp_return_frame_rx_napi(xdpf);
> -                       } else if (unlikely(err < 0)) {
> -                               trace_xdp_exception(vi->dev, xdp_prog, act);
> -                               goto err_xdp_frags;
> -                       }
> -                       *xdp_xmit |= VIRTIO_XDP_TX;
> -                       rcu_read_unlock();
> -                       goto xdp_xmit;
>                 case XDP_REDIRECT:
> -                       stats->xdp_redirects++;
> -                       err = xdp_do_redirect(dev, &xdp, xdp_prog);
> -                       if (err)
> -                               goto err_xdp_frags;
> -                       *xdp_xmit |= VIRTIO_XDP_REDIR;
>                         rcu_read_unlock();
>                         goto xdp_xmit;
>                 default:
> -                       bpf_warn_invalid_xdp_action(vi->dev, xdp_prog, act);
> -                       fallthrough;
> -               case XDP_ABORTED:
> -                       trace_xdp_exception(vi->dev, xdp_prog, act);
> -                       fallthrough;
> -               case XDP_DROP:
> -                       goto err_xdp_frags;
> +                       break;
>                 }
>  err_xdp_frags:
>                 if (xdp_buff_has_frags(&xdp)) {
> --
> 2.32.0.3.g01195cf9f
>

_______________________________________________
Virtualization mailing list
Virtualization@lists.linux-foundation.org
https://lists.linuxfoundation.org/mailman/listinfo/virtualization

Re: [PATCH net-next v2 05/14] virtio_net: introduce xdp res enums

[Jason Wang]

On Tue, Apr 18, 2023 at 2:53 PM Xuan Zhuo <xuanzhuo@linux.alibaba.com> wrote:
>
> virtnet_xdp_handler() is to process all the logic related to XDP. The
> caller only needs to care about how to deal with the buf. So this commit
> introduces new enums:
>
> 1. VIRTNET_XDP_RES_PASS: make skb by the buf
> 2. VIRTNET_XDP_RES_DROP: xdp return drop action or some error, caller
>    should release the buf
> 3. VIRTNET_XDP_RES_CONSUMED: xdp consumed the buf, the caller doesnot to

Typo, should be "does not"

>    do anything
>

I think it's better if you could explain if this can help anything
(e.g simplify the patches on top?).

Thanks


> Signed-off-by: Xuan Zhuo <xuanzhuo@linux.alibaba.com>
> ---
>  drivers/net/virtio_net.c | 42 ++++++++++++++++++++++++++--------------
>  1 file changed, 27 insertions(+), 15 deletions(-)
>
> diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
> index 0fa64c314ea7..4dfdc211d355 100644
> --- a/drivers/net/virtio_net.c
> +++ b/drivers/net/virtio_net.c
> @@ -301,6 +301,15 @@ struct padded_vnet_hdr {
>         char padding[12];
>  };
>
> +enum {
> +       /* xdp pass */
> +       VIRTNET_XDP_RES_PASS,
> +       /* drop packet. the caller needs to release the page. */
> +       VIRTNET_XDP_RES_DROP,
> +       /* packet is consumed by xdp. the caller needs to do nothing. */
> +       VIRTNET_XDP_RES_CONSUMED,
> +};
> +
>  static void virtnet_rq_free_unused_buf(struct virtqueue *vq, void *buf);
>  static void virtnet_sq_free_unused_buf(struct virtqueue *vq, void *buf);
>
> @@ -803,14 +812,14 @@ static int virtnet_xdp_handler(struct bpf_prog *xdp_prog, struct xdp_buff *xdp,
>
>         switch (act) {
>         case XDP_PASS:
> -               return act;
> +               return VIRTNET_XDP_RES_PASS;
>
>         case XDP_TX:
>                 stats->xdp_tx++;
>                 xdpf = xdp_convert_buff_to_frame(xdp);
>                 if (unlikely(!xdpf)) {
>                         netdev_dbg(dev, "convert buff to frame failed for xdp\n");
> -                       return XDP_DROP;
> +                       return VIRTNET_XDP_RES_DROP;
>                 }
>
>                 err = virtnet_xdp_xmit(dev, 1, &xdpf, 0);
> @@ -818,19 +827,20 @@ static int virtnet_xdp_handler(struct bpf_prog *xdp_prog, struct xdp_buff *xdp,
>                         xdp_return_frame_rx_napi(xdpf);
>                 } else if (unlikely(err < 0)) {
>                         trace_xdp_exception(dev, xdp_prog, act);
> -                       return XDP_DROP;
> +                       return VIRTNET_XDP_RES_DROP;
>                 }
> +
>                 *xdp_xmit |= VIRTIO_XDP_TX;
> -               return act;
> +               return VIRTNET_XDP_RES_CONSUMED;
>
>         case XDP_REDIRECT:
>                 stats->xdp_redirects++;
>                 err = xdp_do_redirect(dev, xdp, xdp_prog);
>                 if (err)
> -                       return XDP_DROP;
> +                       return VIRTNET_XDP_RES_DROP;
>
>                 *xdp_xmit |= VIRTIO_XDP_REDIR;
> -               return act;
> +               return VIRTNET_XDP_RES_CONSUMED;
>
>         default:
>                 bpf_warn_invalid_xdp_action(dev, xdp_prog, act);
> @@ -839,7 +849,7 @@ static int virtnet_xdp_handler(struct bpf_prog *xdp_prog, struct xdp_buff *xdp,
>                 trace_xdp_exception(dev, xdp_prog, act);
>                 fallthrough;
>         case XDP_DROP:
> -               return XDP_DROP;
> +               return VIRTNET_XDP_RES_DROP;
>         }
>  }
>
> @@ -987,17 +997,18 @@ static struct sk_buff *receive_small(struct net_device *dev,
>                 act = virtnet_xdp_handler(xdp_prog, &xdp, dev, xdp_xmit, stats);
>
>                 switch (act) {
> -               case XDP_PASS:
> +               case VIRTNET_XDP_RES_PASS:
>                         /* Recalculate length in case bpf program changed it */
>                         delta = orig_data - xdp.data;
>                         len = xdp.data_end - xdp.data;
>                         metasize = xdp.data - xdp.data_meta;
>                         break;
> -               case XDP_TX:
> -               case XDP_REDIRECT:
> +
> +               case VIRTNET_XDP_RES_CONSUMED:
>                         rcu_read_unlock();
>                         goto xdp_xmit;
> -               default:
> +
> +               case VIRTNET_XDP_RES_DROP:
>                         goto err_xdp;
>                 }
>         }
> @@ -1324,18 +1335,19 @@ static struct sk_buff *receive_mergeable(struct net_device *dev,
>                 act = virtnet_xdp_handler(xdp_prog, &xdp, dev, xdp_xmit, stats);
>
>                 switch (act) {
> -               case XDP_PASS:
> +               case VIRTNET_XDP_RES_PASS:
>                         head_skb = build_skb_from_xdp_buff(dev, vi, &xdp, xdp_frags_truesz);
>                         if (unlikely(!head_skb))
>                                 goto err_xdp_frags;
>
>                         rcu_read_unlock();
>                         return head_skb;
> -               case XDP_TX:
> -               case XDP_REDIRECT:
> +
> +               case VIRTNET_XDP_RES_CONSUMED:
>                         rcu_read_unlock();
>                         goto xdp_xmit;
> -               default:
> +
> +               case VIRTNET_XDP_RES_DROP:
>                         break;
>                 }
>  err_xdp_frags:
> --
> 2.32.0.3.g01195cf9f
>

_______________________________________________
Virtualization mailing list
Virtualization@lists.linux-foundation.org
https://lists.linuxfoundation.org/mailman/listinfo/virtualization

Re: [PATCH net-next v2 08/14] virtio_net: auto release xdp shinfo

[Jason Wang]

On Tue, Apr 18, 2023 at 2:53 PM Xuan Zhuo <xuanzhuo@linux.alibaba.com> wrote:
>
> virtnet_build_xdp_buff_mrg() and virtnet_xdp_handler() auto
> release xdp shinfo then the caller no need to careful the xdp shinfo.
>
> Signed-off-by: Xuan Zhuo <xuanzhuo@linux.alibaba.com>
> ---
>  drivers/net/virtio_net.c | 29 +++++++++++++++++------------
>  1 file changed, 17 insertions(+), 12 deletions(-)
>
> diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
> index e2eade87d2d4..266c1670beda 100644
> --- a/drivers/net/virtio_net.c
> +++ b/drivers/net/virtio_net.c
> @@ -834,7 +834,7 @@ static int virtnet_xdp_handler(struct bpf_prog *xdp_prog, struct xdp_buff *xdp,
>                 xdpf = xdp_convert_buff_to_frame(xdp);
>                 if (unlikely(!xdpf)) {
>                         netdev_dbg(dev, "convert buff to frame failed for xdp\n");
> -                       return VIRTNET_XDP_RES_DROP;
> +                       goto drop;
>                 }
>
>                 err = virtnet_xdp_xmit(dev, 1, &xdpf, 0);
> @@ -842,7 +842,7 @@ static int virtnet_xdp_handler(struct bpf_prog *xdp_prog, struct xdp_buff *xdp,
>                         xdp_return_frame_rx_napi(xdpf);
>                 } else if (unlikely(err < 0)) {
>                         trace_xdp_exception(dev, xdp_prog, act);
> -                       return VIRTNET_XDP_RES_DROP;
> +                       goto drop;
>                 }
>
>                 *xdp_xmit |= VIRTIO_XDP_TX;
> @@ -852,7 +852,7 @@ static int virtnet_xdp_handler(struct bpf_prog *xdp_prog, struct xdp_buff *xdp,
>                 stats->xdp_redirects++;
>                 err = xdp_do_redirect(dev, xdp, xdp_prog);
>                 if (err)
> -                       return VIRTNET_XDP_RES_DROP;
> +                       goto drop;
>
>                 *xdp_xmit |= VIRTIO_XDP_REDIR;
>                 return VIRTNET_XDP_RES_CONSUMED;
> @@ -864,8 +864,12 @@ static int virtnet_xdp_handler(struct bpf_prog *xdp_prog, struct xdp_buff *xdp,
>                 trace_xdp_exception(dev, xdp_prog, act);
>                 fallthrough;
>         case XDP_DROP:
> -               return VIRTNET_XDP_RES_DROP;
> +               break;
>         }
> +
> +drop:
> +       put_xdp_frags(xdp);
> +       return VIRTNET_XDP_RES_DROP;
>  }

Patch looks correct but we end up some inconsistency here.

frags are automatically released.

but the page still needs to be freed by the caller?

Thanks


>
>  static unsigned int virtnet_get_headroom(struct virtnet_info *vi)
> @@ -1201,7 +1205,7 @@ static int virtnet_build_xdp_buff_mrg(struct net_device *dev,
>                                  dev->name, *num_buf,
>                                  virtio16_to_cpu(vi->vdev, hdr->num_buffers));
>                         dev->stats.rx_length_errors++;
> -                       return -EINVAL;
> +                       goto err;
>                 }
>
>                 stats->bytes += len;
> @@ -1220,7 +1224,7 @@ static int virtnet_build_xdp_buff_mrg(struct net_device *dev,
>                         pr_debug("%s: rx error: len %u exceeds truesize %lu\n",
>                                  dev->name, len, (unsigned long)(truesize - room));
>                         dev->stats.rx_length_errors++;
> -                       return -EINVAL;
> +                       goto err;
>                 }
>
>                 frag = &shinfo->frags[shinfo->nr_frags++];
> @@ -1235,6 +1239,10 @@ static int virtnet_build_xdp_buff_mrg(struct net_device *dev,
>
>         *xdp_frags_truesize = xdp_frags_truesz;
>         return 0;
> +
> +err:
> +       put_xdp_frags(xdp);
> +       return -EINVAL;
>  }
>
>  static void *mergeable_xdp_prepare(struct virtnet_info *vi,
> @@ -1364,7 +1372,7 @@ static struct sk_buff *receive_mergeable(struct net_device *dev,
>                 err = virtnet_build_xdp_buff_mrg(dev, vi, rq, &xdp, data, len, frame_sz,
>                                                  &num_buf, &xdp_frags_truesz, stats);
>                 if (unlikely(err))
> -                       goto err_xdp_frags;
> +                       goto err_xdp;
>
>                 act = virtnet_xdp_handler(xdp_prog, &xdp, dev, xdp_xmit, stats);
>
> @@ -1372,7 +1380,7 @@ static struct sk_buff *receive_mergeable(struct net_device *dev,
>                 case VIRTNET_XDP_RES_PASS:
>                         head_skb = build_skb_from_xdp_buff(dev, vi, &xdp, xdp_frags_truesz);
>                         if (unlikely(!head_skb))
> -                               goto err_xdp_frags;
> +                               goto err_xdp;
>
>                         rcu_read_unlock();
>                         return head_skb;
> @@ -1382,11 +1390,8 @@ static struct sk_buff *receive_mergeable(struct net_device *dev,
>                         goto xdp_xmit;
>
>                 case VIRTNET_XDP_RES_DROP:
> -                       break;
> +                       goto err_xdp;
>                 }
> -err_xdp_frags:
> -               put_xdp_frags(&xdp);
> -               goto err_xdp;
>         }
>         rcu_read_unlock();
>
> --
> 2.32.0.3.g01195cf9f
>

_______________________________________________
Virtualization mailing list
Virtualization@lists.linux-foundation.org
https://lists.linuxfoundation.org/mailman/listinfo/virtualization

Re: [PATCH net-next v2 09/14] virtio_net: introduce receive_mergeable_xdp()

[Jason Wang]

On Tue, Apr 18, 2023 at 2:53 PM Xuan Zhuo <xuanzhuo@linux.alibaba.com> wrote:
>
> The purpose of this patch is to simplify the receive_mergeable().
> Separate all the logic of XDP into a function.
>
> Signed-off-by: Xuan Zhuo <xuanzhuo@linux.alibaba.com>

Acked-by: Jason Wang <jasowang@redhat.com>

Thanks

> ---
>  drivers/net/virtio_net.c | 100 ++++++++++++++++++++++++---------------
>  1 file changed, 61 insertions(+), 39 deletions(-)
>
> diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
> index 266c1670beda..42e9927e316b 100644
> --- a/drivers/net/virtio_net.c
> +++ b/drivers/net/virtio_net.c
> @@ -1319,6 +1319,63 @@ static void *mergeable_xdp_prepare(struct virtnet_info *vi,
>         return page_address(*page) + VIRTIO_XDP_HEADROOM;
>  }
>
> +static struct sk_buff *receive_mergeable_xdp(struct net_device *dev,
> +                                            struct virtnet_info *vi,
> +                                            struct receive_queue *rq,
> +                                            struct bpf_prog *xdp_prog,
> +                                            void *buf,
> +                                            void *ctx,
> +                                            unsigned int len,
> +                                            unsigned int *xdp_xmit,
> +                                            struct virtnet_rq_stats *stats)
> +{
> +       struct virtio_net_hdr_mrg_rxbuf *hdr = buf;
> +       int num_buf = virtio16_to_cpu(vi->vdev, hdr->num_buffers);
> +       struct page *page = virt_to_head_page(buf);
> +       int offset = buf - page_address(page);
> +       unsigned int xdp_frags_truesz = 0;
> +       struct sk_buff *head_skb;
> +       unsigned int frame_sz;
> +       struct xdp_buff xdp;
> +       void *data;
> +       u32 act;
> +       int err;
> +
> +       data = mergeable_xdp_prepare(vi, rq, xdp_prog, ctx, &frame_sz, &num_buf, &page,
> +                                    offset, &len, hdr);
> +       if (unlikely(!data))
> +               goto err_xdp;
> +
> +       err = virtnet_build_xdp_buff_mrg(dev, vi, rq, &xdp, data, len, frame_sz,
> +                                        &num_buf, &xdp_frags_truesz, stats);
> +       if (unlikely(err))
> +               goto err_xdp;
> +
> +       act = virtnet_xdp_handler(xdp_prog, &xdp, dev, xdp_xmit, stats);
> +
> +       switch (act) {
> +       case VIRTNET_XDP_RES_PASS:
> +               head_skb = build_skb_from_xdp_buff(dev, vi, &xdp, xdp_frags_truesz);
> +               if (unlikely(!head_skb))
> +                       goto err_xdp;
> +               return head_skb;
> +
> +       case VIRTNET_XDP_RES_CONSUMED:
> +               return NULL;
> +
> +       case VIRTNET_XDP_RES_DROP:
> +               break;
> +       }
> +
> +err_xdp:
> +       put_page(page);
> +       mergeable_buf_free(rq, num_buf, dev, stats);
> +
> +       stats->xdp_drops++;
> +       stats->drops++;
> +       return NULL;
> +}
> +
>  static struct sk_buff *receive_mergeable(struct net_device *dev,
>                                          struct virtnet_info *vi,
>                                          struct receive_queue *rq,
> @@ -1338,8 +1395,6 @@ static struct sk_buff *receive_mergeable(struct net_device *dev,
>         unsigned int headroom = mergeable_ctx_to_headroom(ctx);
>         unsigned int tailroom = headroom ? sizeof(struct skb_shared_info) : 0;
>         unsigned int room = SKB_DATA_ALIGN(headroom + tailroom);
> -       unsigned int frame_sz;
> -       int err;
>
>         head_skb = NULL;
>         stats->bytes += len - vi->hdr_len;
> @@ -1359,39 +1414,10 @@ static struct sk_buff *receive_mergeable(struct net_device *dev,
>         rcu_read_lock();
>         xdp_prog = rcu_dereference(rq->xdp_prog);
>         if (xdp_prog) {
> -               unsigned int xdp_frags_truesz = 0;
> -               struct xdp_buff xdp;
> -               void *data;
> -               u32 act;
> -
> -               data = mergeable_xdp_prepare(vi, rq, xdp_prog, ctx, &frame_sz,
> -                                            &num_buf, &page, offset, &len, hdr);
> -               if (unlikely(!data))
> -                       goto err_xdp;
> -
> -               err = virtnet_build_xdp_buff_mrg(dev, vi, rq, &xdp, data, len, frame_sz,
> -                                                &num_buf, &xdp_frags_truesz, stats);
> -               if (unlikely(err))
> -                       goto err_xdp;
> -
> -               act = virtnet_xdp_handler(xdp_prog, &xdp, dev, xdp_xmit, stats);
> -
> -               switch (act) {
> -               case VIRTNET_XDP_RES_PASS:
> -                       head_skb = build_skb_from_xdp_buff(dev, vi, &xdp, xdp_frags_truesz);
> -                       if (unlikely(!head_skb))
> -                               goto err_xdp;
> -
> -                       rcu_read_unlock();
> -                       return head_skb;
> -
> -               case VIRTNET_XDP_RES_CONSUMED:
> -                       rcu_read_unlock();
> -                       goto xdp_xmit;
> -
> -               case VIRTNET_XDP_RES_DROP:
> -                       goto err_xdp;
> -               }
> +               head_skb = receive_mergeable_xdp(dev, vi, rq, xdp_prog, buf, ctx,
> +                                                len, xdp_xmit, stats);
> +               rcu_read_unlock();
> +               return head_skb;
>         }
>         rcu_read_unlock();
>
> @@ -1461,9 +1487,6 @@ static struct sk_buff *receive_mergeable(struct net_device *dev,
>         ewma_pkt_len_add(&rq->mrg_avg_pkt_len, head_skb->len);
>         return head_skb;
>
> -err_xdp:
> -       rcu_read_unlock();
> -       stats->xdp_drops++;
>  err_skb:
>         put_page(page);
>         mergeable_buf_free(rq, num_buf, dev, stats);
> @@ -1471,7 +1494,6 @@ static struct sk_buff *receive_mergeable(struct net_device *dev,
>  err_buf:
>         stats->drops++;
>         dev_kfree_skb(head_skb);
> -xdp_xmit:
>         return NULL;
>  }
>
> --
> 2.32.0.3.g01195cf9f
>

_______________________________________________
Virtualization mailing list
Virtualization@lists.linux-foundation.org
https://lists.linuxfoundation.org/mailman/listinfo/virtualization

Re: [PATCH net-next v2 10/14] virtio_net: merge: remove skip_xdp

[Jason Wang]

On Tue, Apr 18, 2023 at 2:53 PM Xuan Zhuo <xuanzhuo@linux.alibaba.com> wrote:
>
> Now, the logic of merge xdp process is simple, we can remove the
> skip_xdp.
>
> Signed-off-by: Xuan Zhuo <xuanzhuo@linux.alibaba.com>

Acked-by: Jason Wang <jasowang@redhat.com>

Thanks

> ---
>  drivers/net/virtio_net.c | 23 ++++++++++-------------
>  1 file changed, 10 insertions(+), 13 deletions(-)
>
> diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
> index 42e9927e316b..a4bb25f39f12 100644
> --- a/drivers/net/virtio_net.c
> +++ b/drivers/net/virtio_net.c
> @@ -1390,7 +1390,6 @@ static struct sk_buff *receive_mergeable(struct net_device *dev,
>         struct page *page = virt_to_head_page(buf);
>         int offset = buf - page_address(page);
>         struct sk_buff *head_skb, *curr_skb;
> -       struct bpf_prog *xdp_prog;
>         unsigned int truesize = mergeable_ctx_to_truesize(ctx);
>         unsigned int headroom = mergeable_ctx_to_headroom(ctx);
>         unsigned int tailroom = headroom ? sizeof(struct skb_shared_info) : 0;
> @@ -1406,22 +1405,20 @@ static struct sk_buff *receive_mergeable(struct net_device *dev,
>                 goto err_skb;
>         }
>
> -       if (likely(!vi->xdp_enabled)) {
> -               xdp_prog = NULL;
> -               goto skip_xdp;
> -       }
> +       if (unlikely(vi->xdp_enabled)) {
> +               struct bpf_prog *xdp_prog;
>
> -       rcu_read_lock();
> -       xdp_prog = rcu_dereference(rq->xdp_prog);
> -       if (xdp_prog) {
> -               head_skb = receive_mergeable_xdp(dev, vi, rq, xdp_prog, buf, ctx,
> -                                                len, xdp_xmit, stats);
> +               rcu_read_lock();
> +               xdp_prog = rcu_dereference(rq->xdp_prog);
> +               if (xdp_prog) {
> +                       head_skb = receive_mergeable_xdp(dev, vi, rq, xdp_prog, buf, ctx,
> +                                                        len, xdp_xmit, stats);
> +                       rcu_read_unlock();
> +                       return head_skb;
> +               }
>                 rcu_read_unlock();
> -               return head_skb;
>         }
> -       rcu_read_unlock();
>
> -skip_xdp:
>         head_skb = page_to_skb(vi, rq, page, offset, len, truesize, headroom);
>         curr_skb = head_skb;
>
> --
> 2.32.0.3.g01195cf9f
>

_______________________________________________
Virtualization mailing list
Virtualization@lists.linux-foundation.org
https://lists.linuxfoundation.org/mailman/listinfo/virtualization

Re: [PATCH net-next v2 11/14] virtio_net: introduce receive_small_xdp()

[Jason Wang]

On Tue, Apr 18, 2023 at 2:53 PM Xuan Zhuo <xuanzhuo@linux.alibaba.com> wrote:
>
> The purpose of this patch is to simplify the receive_small().
> Separate all the logic of XDP of small into a function.
>
> Signed-off-by: Xuan Zhuo <xuanzhuo@linux.alibaba.com>
> ---
>  drivers/net/virtio_net.c | 165 +++++++++++++++++++++++----------------
>  1 file changed, 99 insertions(+), 66 deletions(-)
>
> diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
> index a4bb25f39f12..34220f5f27d1 100644
> --- a/drivers/net/virtio_net.c
> +++ b/drivers/net/virtio_net.c
> @@ -941,6 +941,98 @@ static struct page *xdp_linearize_page(struct receive_queue *rq,
>         return NULL;
>  }
>
> +static struct sk_buff *receive_small_xdp(struct net_device *dev,
> +                                        struct virtnet_info *vi,
> +                                        struct receive_queue *rq,
> +                                        struct bpf_prog *xdp_prog,
> +                                        void *buf,
> +                                        unsigned int xdp_headroom,
> +                                        unsigned int len,
> +                                        unsigned int *xdp_xmit,
> +                                        struct virtnet_rq_stats *stats)
> +{
> +       unsigned int header_offset = VIRTNET_RX_PAD + xdp_headroom;
> +       unsigned int headroom = vi->hdr_len + header_offset;
> +       struct virtio_net_hdr_mrg_rxbuf *hdr = buf + header_offset;
> +       struct page *page = virt_to_head_page(buf);
> +       struct page *xdp_page;
> +       unsigned int buflen;
> +       struct xdp_buff xdp;
> +       struct sk_buff *skb;
> +       unsigned int delta = 0;
> +       unsigned int metasize = 0;
> +       void *orig_data;
> +       u32 act;
> +
> +       if (unlikely(hdr->hdr.gso_type))
> +               goto err_xdp;
> +
> +       buflen = SKB_DATA_ALIGN(GOOD_PACKET_LEN + headroom) +
> +               SKB_DATA_ALIGN(sizeof(struct skb_shared_info));
> +
> +       if (unlikely(xdp_headroom < virtnet_get_headroom(vi))) {
> +               int offset = buf - page_address(page) + header_offset;
> +               unsigned int tlen = len + vi->hdr_len;
> +               int num_buf = 1;
> +
> +               xdp_headroom = virtnet_get_headroom(vi);
> +               header_offset = VIRTNET_RX_PAD + xdp_headroom;
> +               headroom = vi->hdr_len + header_offset;
> +               buflen = SKB_DATA_ALIGN(GOOD_PACKET_LEN + headroom) +
> +                       SKB_DATA_ALIGN(sizeof(struct skb_shared_info));
> +               xdp_page = xdp_linearize_page(rq, &num_buf, page,
> +                                             offset, header_offset,
> +                                             &tlen);
> +               if (!xdp_page)
> +                       goto err_xdp;
> +
> +               buf = page_address(xdp_page);
> +               put_page(page);
> +               page = xdp_page;
> +       }
> +
> +       xdp_init_buff(&xdp, buflen, &rq->xdp_rxq);
> +       xdp_prepare_buff(&xdp, buf + VIRTNET_RX_PAD + vi->hdr_len,
> +                        xdp_headroom, len, true);
> +       orig_data = xdp.data;
> +
> +       act = virtnet_xdp_handler(xdp_prog, &xdp, dev, xdp_xmit, stats);
> +
> +       switch (act) {
> +       case VIRTNET_XDP_RES_PASS:
> +               /* Recalculate length in case bpf program changed it */
> +               delta = orig_data - xdp.data;
> +               len = xdp.data_end - xdp.data;
> +               metasize = xdp.data - xdp.data_meta;
> +               break;
> +
> +       case VIRTNET_XDP_RES_CONSUMED:
> +               goto xdp_xmit;
> +
> +       case VIRTNET_XDP_RES_DROP:
> +               goto err_xdp;
> +       }
> +
> +       skb = build_skb(buf, buflen);
> +       if (!skb)
> +               goto err;
> +
> +       skb_reserve(skb, headroom - delta);
> +       skb_put(skb, len);
> +       if (metasize)
> +               skb_metadata_set(skb, metasize);
> +
> +       return skb;
> +
> +err_xdp:
> +       stats->xdp_drops++;
> +err:
> +       stats->drops++;
> +       put_page(page);
> +xdp_xmit:
> +       return NULL;
> +}
> +
>  static struct sk_buff *receive_small(struct net_device *dev,
>                                      struct virtnet_info *vi,
>                                      struct receive_queue *rq,
> @@ -957,9 +1049,6 @@ static struct sk_buff *receive_small(struct net_device *dev,
>         unsigned int buflen = SKB_DATA_ALIGN(GOOD_PACKET_LEN + headroom) +
>                               SKB_DATA_ALIGN(sizeof(struct skb_shared_info));
>         struct page *page = virt_to_head_page(buf);
> -       unsigned int delta = 0;
> -       struct page *xdp_page;
> -       unsigned int metasize = 0;
>
>         len -= vi->hdr_len;
>         stats->bytes += len;
> @@ -979,57 +1068,10 @@ static struct sk_buff *receive_small(struct net_device *dev,
>         rcu_read_lock();
>         xdp_prog = rcu_dereference(rq->xdp_prog);
>         if (xdp_prog) {
> -               struct virtio_net_hdr_mrg_rxbuf *hdr = buf + header_offset;
> -               struct xdp_buff xdp;
> -               void *orig_data;
> -               u32 act;
> -
> -               if (unlikely(hdr->hdr.gso_type))
> -                       goto err_xdp;
> -
> -               if (unlikely(xdp_headroom < virtnet_get_headroom(vi))) {
> -                       int offset = buf - page_address(page) + header_offset;
> -                       unsigned int tlen = len + vi->hdr_len;
> -                       int num_buf = 1;
> -
> -                       xdp_headroom = virtnet_get_headroom(vi);
> -                       header_offset = VIRTNET_RX_PAD + xdp_headroom;
> -                       headroom = vi->hdr_len + header_offset;
> -                       buflen = SKB_DATA_ALIGN(GOOD_PACKET_LEN + headroom) +
> -                                SKB_DATA_ALIGN(sizeof(struct skb_shared_info));
> -                       xdp_page = xdp_linearize_page(rq, &num_buf, page,
> -                                                     offset, header_offset,
> -                                                     &tlen);
> -                       if (!xdp_page)
> -                               goto err_xdp;
> -
> -                       buf = page_address(xdp_page);
> -                       put_page(page);
> -                       page = xdp_page;
> -               }
> -
> -               xdp_init_buff(&xdp, buflen, &rq->xdp_rxq);
> -               xdp_prepare_buff(&xdp, buf + VIRTNET_RX_PAD + vi->hdr_len,
> -                                xdp_headroom, len, true);
> -               orig_data = xdp.data;
> -
> -               act = virtnet_xdp_handler(xdp_prog, &xdp, dev, xdp_xmit, stats);
> -
> -               switch (act) {
> -               case VIRTNET_XDP_RES_PASS:
> -                       /* Recalculate length in case bpf program changed it */
> -                       delta = orig_data - xdp.data;
> -                       len = xdp.data_end - xdp.data;
> -                       metasize = xdp.data - xdp.data_meta;
> -                       break;
> -
> -               case VIRTNET_XDP_RES_CONSUMED:
> -                       rcu_read_unlock();
> -                       goto xdp_xmit;
> -
> -               case VIRTNET_XDP_RES_DROP:
> -                       goto err_xdp;
> -               }
> +               skb = receive_small_xdp(dev, vi, rq, xdp_prog, buf, xdp_headroom,
> +                                       len, xdp_xmit, stats);
> +               rcu_read_unlock();
> +               return skb;
>         }
>         rcu_read_unlock();
>
> @@ -1037,25 +1079,16 @@ static struct sk_buff *receive_small(struct net_device *dev,
>         skb = build_skb(buf, buflen);
>         if (!skb)
>                 goto err;
> -       skb_reserve(skb, headroom - delta);
> +       skb_reserve(skb, headroom);
>         skb_put(skb, len);
> -       if (!xdp_prog) {
> -               buf += header_offset;
> -               memcpy(skb_vnet_hdr(skb), buf, vi->hdr_len);
> -       } /* keep zeroed vnet hdr since XDP is loaded */
> -
> -       if (metasize)
> -               skb_metadata_set(skb, metasize);
>
> +       buf += header_offset;
> +       memcpy(skb_vnet_hdr(skb), buf, vi->hdr_len);

So we end up with some code duplication between receive_small() and
receive_small_xdp() on building skbs. Is this intended?

For mergeable, we're probably fine since XDP_PASS doesn't need to
get_buf(). But for the path of small, I wonder if it's better to keep
the original code for building skb.

Thanks

>         return skb;
>
> -err_xdp:
> -       rcu_read_unlock();
> -       stats->xdp_drops++;
>  err:
>         stats->drops++;
>         put_page(page);
> -xdp_xmit:
>         return NULL;
>  }
>
> --
> 2.32.0.3.g01195cf9f
>

_______________________________________________
Virtualization mailing list
Virtualization@lists.linux-foundation.org
https://lists.linuxfoundation.org/mailman/listinfo/virtualization

Re: [PATCH net-next v2 12/14] virtio_net: small: optimize code

[Jason Wang]

On Tue, Apr 18, 2023 at 2:53 PM Xuan Zhuo <xuanzhuo@linux.alibaba.com> wrote:
>
> In the case of XDP-PASS, skb_reserve uses the delta to compatible
> non-XDP, now remove this logic.
>
> Signed-off-by: Xuan Zhuo <xuanzhuo@linux.alibaba.com>

Acked-by: Jason Wang <jasowang@redhat.com>

Thanks

> ---
>  drivers/net/virtio_net.c | 6 +-----
>  1 file changed, 1 insertion(+), 5 deletions(-)
>
> diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
> index 34220f5f27d1..f6f5903face2 100644
> --- a/drivers/net/virtio_net.c
> +++ b/drivers/net/virtio_net.c
> @@ -959,9 +959,7 @@ static struct sk_buff *receive_small_xdp(struct net_device *dev,
>         unsigned int buflen;
>         struct xdp_buff xdp;
>         struct sk_buff *skb;
> -       unsigned int delta = 0;
>         unsigned int metasize = 0;
> -       void *orig_data;
>         u32 act;
>
>         if (unlikely(hdr->hdr.gso_type))
> @@ -994,14 +992,12 @@ static struct sk_buff *receive_small_xdp(struct net_device *dev,
>         xdp_init_buff(&xdp, buflen, &rq->xdp_rxq);
>         xdp_prepare_buff(&xdp, buf + VIRTNET_RX_PAD + vi->hdr_len,
>                          xdp_headroom, len, true);
> -       orig_data = xdp.data;
>
>         act = virtnet_xdp_handler(xdp_prog, &xdp, dev, xdp_xmit, stats);
>
>         switch (act) {
>         case VIRTNET_XDP_RES_PASS:
>                 /* Recalculate length in case bpf program changed it */
> -               delta = orig_data - xdp.data;
>                 len = xdp.data_end - xdp.data;
>                 metasize = xdp.data - xdp.data_meta;
>                 break;
> @@ -1017,7 +1013,7 @@ static struct sk_buff *receive_small_xdp(struct net_device *dev,
>         if (!skb)
>                 goto err;
>
> -       skb_reserve(skb, headroom - delta);
> +       skb_reserve(skb, xdp.data - buf);
>         skb_put(skb, len);
>         if (metasize)
>                 skb_metadata_set(skb, metasize);
> --
> 2.32.0.3.g01195cf9f
>

_______________________________________________
Virtualization mailing list
Virtualization@lists.linux-foundation.org
https://lists.linuxfoundation.org/mailman/listinfo/virtualization

Re: [PATCH net-next v2 13/14] virtio_net: small: optimize code

[Jason Wang]

On Tue, Apr 18, 2023 at 2:53 PM Xuan Zhuo <xuanzhuo@linux.alibaba.com> wrote:
>
> Avoid the problem that some variables(headroom and so on) will repeat
> the calculation when process xdp.

While at it, if we agree to use separate code paths for building skbs.

It would be better to have a helper for building skb for non XDP
cases, then we can hide those calculation there.

Thanks

>
> Signed-off-by: Xuan Zhuo <xuanzhuo@linux.alibaba.com>
> ---
>  drivers/net/virtio_net.c | 12 ++++++++----
>  1 file changed, 8 insertions(+), 4 deletions(-)
>
> diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
> index f6f5903face2..5a5636178bd3 100644
> --- a/drivers/net/virtio_net.c
> +++ b/drivers/net/virtio_net.c
> @@ -1040,11 +1040,10 @@ static struct sk_buff *receive_small(struct net_device *dev,
>         struct sk_buff *skb;
>         struct bpf_prog *xdp_prog;
>         unsigned int xdp_headroom = (unsigned long)ctx;
> -       unsigned int header_offset = VIRTNET_RX_PAD + xdp_headroom;
> -       unsigned int headroom = vi->hdr_len + header_offset;
> -       unsigned int buflen = SKB_DATA_ALIGN(GOOD_PACKET_LEN + headroom) +
> -                             SKB_DATA_ALIGN(sizeof(struct skb_shared_info));
>         struct page *page = virt_to_head_page(buf);
> +       unsigned int header_offset;
> +       unsigned int headroom;
> +       unsigned int buflen;
>
>         len -= vi->hdr_len;
>         stats->bytes += len;
> @@ -1072,6 +1071,11 @@ static struct sk_buff *receive_small(struct net_device *dev,
>         rcu_read_unlock();
>
>  skip_xdp:
> +       header_offset = VIRTNET_RX_PAD + xdp_headroom;
> +       headroom = vi->hdr_len + header_offset;
> +       buflen = SKB_DATA_ALIGN(GOOD_PACKET_LEN + headroom) +
> +               SKB_DATA_ALIGN(sizeof(struct skb_shared_info));
> +
>         skb = build_skb(buf, buflen);
>         if (!skb)
>                 goto err;
> --
> 2.32.0.3.g01195cf9f
>

_______________________________________________
Virtualization mailing list
Virtualization@lists.linux-foundation.org
https://lists.linuxfoundation.org/mailman/listinfo/virtualization

Re: [PATCH net-next v2 13/14] virtio_net: small: optimize code

[Xuan Zhuo]

On Thu, 20 Apr 2023 14:32:37 +0800, Jason Wang <jasowang@redhat.com> wrote:
> On Tue, Apr 18, 2023 at 2:53 PM Xuan Zhuo <xuanzhuo@linux.alibaba.com> wrote:
> >
> > Avoid the problem that some variables(headroom and so on) will repeat
> > the calculation when process xdp.
>
> While at it, if we agree to use separate code paths for building skbs.
>
> It would be better to have a helper for building skb for non XDP
> cases, then we can hide those calculation there.


Yes, we can introduce one helper, then receive_small will be more simple.
But these calculation can not shared with xdp case, because xdp case needs to
get these vars before running xdp.

Then the code "copy vnet hdr" and "set metadata" should stay in their own
function.

Only the virtnet_build_skb()[build_skb, skb_reserve, skb_put] can be shared.

static struct sk_buff *virtnet_build_skb(void *buf, unsigned int buflen,
					 unsigned int headroom,
					 unsigned int len)
{
	struct sk_buff *skb;

	skb = build_skb(buf, buflen);
	if (!skb)
		return NULL;

	skb_reserve(skb, headroom);
	skb_put(skb, len);

	return skb;
}

static struct sk_buff *receive_small_build_skb(struct virtnet_info *vi,
					       unsigned int xdp_headroom,
					       void *buf,
					       unsigned int len)
{
	unsigned int header_offset;
	unsigned int headroom;
	unsigned int buflen;
	struct sk_buff *skb;

	header_offset = VIRTNET_RX_PAD + xdp_headroom;
	headroom = vi->hdr_len + header_offset;
	buflen = SKB_DATA_ALIGN(GOOD_PACKET_LEN + headroom) +
		SKB_DATA_ALIGN(sizeof(struct skb_shared_info));

	skb = virtnet_build_skb(buf, buflen, headroom, len);
	if (unlikely(!skb))
		return NULL;

	buf += header_offset;
	memcpy(skb_vnet_hdr(skb), buf, vi->hdr_len);

	return skb;
}

Thanks

>
> Thanks
>
> >
> > Signed-off-by: Xuan Zhuo <xuanzhuo@linux.alibaba.com>
> > ---
> >  drivers/net/virtio_net.c | 12 ++++++++----
> >  1 file changed, 8 insertions(+), 4 deletions(-)
> >
> > diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
> > index f6f5903face2..5a5636178bd3 100644
> > --- a/drivers/net/virtio_net.c
> > +++ b/drivers/net/virtio_net.c
> > @@ -1040,11 +1040,10 @@ static struct sk_buff *receive_small(struct net_device *dev,
> >         struct sk_buff *skb;
> >         struct bpf_prog *xdp_prog;
> >         unsigned int xdp_headroom = (unsigned long)ctx;
> > -       unsigned int header_offset = VIRTNET_RX_PAD + xdp_headroom;
> > -       unsigned int headroom = vi->hdr_len + header_offset;
> > -       unsigned int buflen = SKB_DATA_ALIGN(GOOD_PACKET_LEN + headroom) +
> > -                             SKB_DATA_ALIGN(sizeof(struct skb_shared_info));
> >         struct page *page = virt_to_head_page(buf);
> > +       unsigned int header_offset;
> > +       unsigned int headroom;
> > +       unsigned int buflen;
> >
> >         len -= vi->hdr_len;
> >         stats->bytes += len;
> > @@ -1072,6 +1071,11 @@ static struct sk_buff *receive_small(struct net_device *dev,
> >         rcu_read_unlock();
> >
> >  skip_xdp:
> > +       header_offset = VIRTNET_RX_PAD + xdp_headroom;
> > +       headroom = vi->hdr_len + header_offset;
> > +       buflen = SKB_DATA_ALIGN(GOOD_PACKET_LEN + headroom) +
> > +               SKB_DATA_ALIGN(sizeof(struct skb_shared_info));
> > +
> >         skb = build_skb(buf, buflen);
> >         if (!skb)
> >                 goto err;
> > --
> > 2.32.0.3.g01195cf9f
> >
>
_______________________________________________
Virtualization mailing list
Virtualization@lists.linux-foundation.org
https://lists.linuxfoundation.org/mailman/listinfo/virtualization

Re: [PATCH net-next v2 08/14] virtio_net: auto release xdp shinfo

[Xuan Zhuo]

On Thu, 20 Apr 2023 13:59:30 +0800, Jason Wang <jasowang@redhat.com> wrote:
> On Tue, Apr 18, 2023 at 2:53 PM Xuan Zhuo <xuanzhuo@linux.alibaba.com> wrote:
> >
> > virtnet_build_xdp_buff_mrg() and virtnet_xdp_handler() auto
> > release xdp shinfo then the caller no need to careful the xdp shinfo.
> >
> > Signed-off-by: Xuan Zhuo <xuanzhuo@linux.alibaba.com>
> > ---
> >  drivers/net/virtio_net.c | 29 +++++++++++++++++------------
> >  1 file changed, 17 insertions(+), 12 deletions(-)
> >
> > diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
> > index e2eade87d2d4..266c1670beda 100644
> > --- a/drivers/net/virtio_net.c
> > +++ b/drivers/net/virtio_net.c
> > @@ -834,7 +834,7 @@ static int virtnet_xdp_handler(struct bpf_prog *xdp_prog, struct xdp_buff *xdp,
> >                 xdpf = xdp_convert_buff_to_frame(xdp);
> >                 if (unlikely(!xdpf)) {
> >                         netdev_dbg(dev, "convert buff to frame failed for xdp\n");
> > -                       return VIRTNET_XDP_RES_DROP;
> > +                       goto drop;
> >                 }
> >
> >                 err = virtnet_xdp_xmit(dev, 1, &xdpf, 0);
> > @@ -842,7 +842,7 @@ static int virtnet_xdp_handler(struct bpf_prog *xdp_prog, struct xdp_buff *xdp,
> >                         xdp_return_frame_rx_napi(xdpf);
> >                 } else if (unlikely(err < 0)) {
> >                         trace_xdp_exception(dev, xdp_prog, act);
> > -                       return VIRTNET_XDP_RES_DROP;
> > +                       goto drop;
> >                 }
> >
> >                 *xdp_xmit |= VIRTIO_XDP_TX;
> > @@ -852,7 +852,7 @@ static int virtnet_xdp_handler(struct bpf_prog *xdp_prog, struct xdp_buff *xdp,
> >                 stats->xdp_redirects++;
> >                 err = xdp_do_redirect(dev, xdp, xdp_prog);
> >                 if (err)
> > -                       return VIRTNET_XDP_RES_DROP;
> > +                       goto drop;
> >
> >                 *xdp_xmit |= VIRTIO_XDP_REDIR;
> >                 return VIRTNET_XDP_RES_CONSUMED;
> > @@ -864,8 +864,12 @@ static int virtnet_xdp_handler(struct bpf_prog *xdp_prog, struct xdp_buff *xdp,
> >                 trace_xdp_exception(dev, xdp_prog, act);
> >                 fallthrough;
> >         case XDP_DROP:
> > -               return VIRTNET_XDP_RES_DROP;
> > +               break;
> >         }
> > +
> > +drop:
> > +       put_xdp_frags(xdp);
> > +       return VIRTNET_XDP_RES_DROP;
> >  }
>
> Patch looks correct but we end up some inconsistency here.
>
> frags are automatically released.
>
> but the page still needs to be freed by the caller?


Yes.

Do you want to auto free page?

Thanks.



>
> Thanks
>
>
> >
> >  static unsigned int virtnet_get_headroom(struct virtnet_info *vi)
> > @@ -1201,7 +1205,7 @@ static int virtnet_build_xdp_buff_mrg(struct net_device *dev,
> >                                  dev->name, *num_buf,
> >                                  virtio16_to_cpu(vi->vdev, hdr->num_buffers));
> >                         dev->stats.rx_length_errors++;
> > -                       return -EINVAL;
> > +                       goto err;
> >                 }
> >
> >                 stats->bytes += len;
> > @@ -1220,7 +1224,7 @@ static int virtnet_build_xdp_buff_mrg(struct net_device *dev,
> >                         pr_debug("%s: rx error: len %u exceeds truesize %lu\n",
> >                                  dev->name, len, (unsigned long)(truesize - room));
> >                         dev->stats.rx_length_errors++;
> > -                       return -EINVAL;
> > +                       goto err;
> >                 }
> >
> >                 frag = &shinfo->frags[shinfo->nr_frags++];
> > @@ -1235,6 +1239,10 @@ static int virtnet_build_xdp_buff_mrg(struct net_device *dev,
> >
> >         *xdp_frags_truesize = xdp_frags_truesz;
> >         return 0;
> > +
> > +err:
> > +       put_xdp_frags(xdp);
> > +       return -EINVAL;
> >  }
> >
> >  static void *mergeable_xdp_prepare(struct virtnet_info *vi,
> > @@ -1364,7 +1372,7 @@ static struct sk_buff *receive_mergeable(struct net_device *dev,
> >                 err = virtnet_build_xdp_buff_mrg(dev, vi, rq, &xdp, data, len, frame_sz,
> >                                                  &num_buf, &xdp_frags_truesz, stats);
> >                 if (unlikely(err))
> > -                       goto err_xdp_frags;
> > +                       goto err_xdp;
> >
> >                 act = virtnet_xdp_handler(xdp_prog, &xdp, dev, xdp_xmit, stats);
> >
> > @@ -1372,7 +1380,7 @@ static struct sk_buff *receive_mergeable(struct net_device *dev,
> >                 case VIRTNET_XDP_RES_PASS:
> >                         head_skb = build_skb_from_xdp_buff(dev, vi, &xdp, xdp_frags_truesz);
> >                         if (unlikely(!head_skb))
> > -                               goto err_xdp_frags;
> > +                               goto err_xdp;
> >
> >                         rcu_read_unlock();
> >                         return head_skb;
> > @@ -1382,11 +1390,8 @@ static struct sk_buff *receive_mergeable(struct net_device *dev,
> >                         goto xdp_xmit;
> >
> >                 case VIRTNET_XDP_RES_DROP:
> > -                       break;
> > +                       goto err_xdp;
> >                 }
> > -err_xdp_frags:
> > -               put_xdp_frags(&xdp);
> > -               goto err_xdp;
> >         }
> >         rcu_read_unlock();
> >
> > --
> > 2.32.0.3.g01195cf9f
> >
>
_______________________________________________
Virtualization mailing list
Virtualization@lists.linux-foundation.org
https://lists.linuxfoundation.org/mailman/listinfo/virtualization

Re: [PATCH net-next v2 05/14] virtio_net: introduce xdp res enums

[Michael S. Tsirkin]

On Tue, Apr 18, 2023 at 02:53:18PM +0800, Xuan Zhuo wrote:
> virtnet_xdp_handler() is to process all the logic related to XDP. The
> caller only needs to care about how to deal with the buf. So this commit
> introduces new enums:
> 
> 1. VIRTNET_XDP_RES_PASS: make skb by the buf
> 2. VIRTNET_XDP_RES_DROP: xdp return drop action or some error, caller
>    should release the buf
> 3. VIRTNET_XDP_RES_CONSUMED: xdp consumed the buf, the caller doesnot to
>    do anything
> 
> Signed-off-by: Xuan Zhuo <xuanzhuo@linux.alibaba.com>


I am not excited about using virtio specific enums then translating
to standard ones.

> ---
>  drivers/net/virtio_net.c | 42 ++++++++++++++++++++++++++--------------
>  1 file changed, 27 insertions(+), 15 deletions(-)
> 
> diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
> index 0fa64c314ea7..4dfdc211d355 100644
> --- a/drivers/net/virtio_net.c
> +++ b/drivers/net/virtio_net.c
> @@ -301,6 +301,15 @@ struct padded_vnet_hdr {
>  	char padding[12];
>  };
>  
> +enum {
> +	/* xdp pass */
> +	VIRTNET_XDP_RES_PASS,
> +	/* drop packet. the caller needs to release the page. */
> +	VIRTNET_XDP_RES_DROP,
> +	/* packet is consumed by xdp. the caller needs to do nothing. */
> +	VIRTNET_XDP_RES_CONSUMED,
> +};
> +
>  static void virtnet_rq_free_unused_buf(struct virtqueue *vq, void *buf);
>  static void virtnet_sq_free_unused_buf(struct virtqueue *vq, void *buf);
>  
> @@ -803,14 +812,14 @@ static int virtnet_xdp_handler(struct bpf_prog *xdp_prog, struct xdp_buff *xdp,
>  
>  	switch (act) {
>  	case XDP_PASS:
> -		return act;
> +		return VIRTNET_XDP_RES_PASS;
>  
>  	case XDP_TX:
>  		stats->xdp_tx++;
>  		xdpf = xdp_convert_buff_to_frame(xdp);
>  		if (unlikely(!xdpf)) {
>  			netdev_dbg(dev, "convert buff to frame failed for xdp\n");
> -			return XDP_DROP;
> +			return VIRTNET_XDP_RES_DROP;
>  		}
>  
>  		err = virtnet_xdp_xmit(dev, 1, &xdpf, 0);
> @@ -818,19 +827,20 @@ static int virtnet_xdp_handler(struct bpf_prog *xdp_prog, struct xdp_buff *xdp,
>  			xdp_return_frame_rx_napi(xdpf);
>  		} else if (unlikely(err < 0)) {
>  			trace_xdp_exception(dev, xdp_prog, act);
> -			return XDP_DROP;
> +			return VIRTNET_XDP_RES_DROP;
>  		}
> +
>  		*xdp_xmit |= VIRTIO_XDP_TX;
> -		return act;
> +		return VIRTNET_XDP_RES_CONSUMED;
>  
>  	case XDP_REDIRECT:
>  		stats->xdp_redirects++;
>  		err = xdp_do_redirect(dev, xdp, xdp_prog);
>  		if (err)
> -			return XDP_DROP;
> +			return VIRTNET_XDP_RES_DROP;
>  
>  		*xdp_xmit |= VIRTIO_XDP_REDIR;
> -		return act;
> +		return VIRTNET_XDP_RES_CONSUMED;
>  
>  	default:
>  		bpf_warn_invalid_xdp_action(dev, xdp_prog, act);
> @@ -839,7 +849,7 @@ static int virtnet_xdp_handler(struct bpf_prog *xdp_prog, struct xdp_buff *xdp,
>  		trace_xdp_exception(dev, xdp_prog, act);
>  		fallthrough;
>  	case XDP_DROP:
> -		return XDP_DROP;
> +		return VIRTNET_XDP_RES_DROP;
>  	}
>  }
>  
> @@ -987,17 +997,18 @@ static struct sk_buff *receive_small(struct net_device *dev,
>  		act = virtnet_xdp_handler(xdp_prog, &xdp, dev, xdp_xmit, stats);
>  
>  		switch (act) {
> -		case XDP_PASS:
> +		case VIRTNET_XDP_RES_PASS:
>  			/* Recalculate length in case bpf program changed it */
>  			delta = orig_data - xdp.data;
>  			len = xdp.data_end - xdp.data;
>  			metasize = xdp.data - xdp.data_meta;
>  			break;
> -		case XDP_TX:
> -		case XDP_REDIRECT:
> +
> +		case VIRTNET_XDP_RES_CONSUMED:
>  			rcu_read_unlock();
>  			goto xdp_xmit;
> -		default:
> +
> +		case VIRTNET_XDP_RES_DROP:
>  			goto err_xdp;
>  		}
>  	}
> @@ -1324,18 +1335,19 @@ static struct sk_buff *receive_mergeable(struct net_device *dev,
>  		act = virtnet_xdp_handler(xdp_prog, &xdp, dev, xdp_xmit, stats);
>  
>  		switch (act) {
> -		case XDP_PASS:
> +		case VIRTNET_XDP_RES_PASS:
>  			head_skb = build_skb_from_xdp_buff(dev, vi, &xdp, xdp_frags_truesz);
>  			if (unlikely(!head_skb))
>  				goto err_xdp_frags;
>  
>  			rcu_read_unlock();
>  			return head_skb;
> -		case XDP_TX:
> -		case XDP_REDIRECT:
> +
> +		case VIRTNET_XDP_RES_CONSUMED:
>  			rcu_read_unlock();
>  			goto xdp_xmit;
> -		default:
> +
> +		case VIRTNET_XDP_RES_DROP:
>  			break;
>  		}
>  err_xdp_frags:
> -- 
> 2.32.0.3.g01195cf9f

_______________________________________________
Virtualization mailing list
Virtualization@lists.linux-foundation.org
https://lists.linuxfoundation.org/mailman/listinfo/virtualization

Re: [PATCH net-next v2 05/14] virtio_net: introduce xdp res enums

[Xuan Zhuo]

On Fri, 21 Apr 2023 03:00:15 -0400, "Michael S. Tsirkin" <mst@redhat.com> wrote:
> On Tue, Apr 18, 2023 at 02:53:18PM +0800, Xuan Zhuo wrote:
> > virtnet_xdp_handler() is to process all the logic related to XDP. The
> > caller only needs to care about how to deal with the buf. So this commit
> > introduces new enums:
> >
> > 1. VIRTNET_XDP_RES_PASS: make skb by the buf
> > 2. VIRTNET_XDP_RES_DROP: xdp return drop action or some error, caller
> >    should release the buf
> > 3. VIRTNET_XDP_RES_CONSUMED: xdp consumed the buf, the caller doesnot to
> >    do anything
> >
> > Signed-off-by: Xuan Zhuo <xuanzhuo@linux.alibaba.com>
>
>
> I am not excited about using virtio specific enums then translating
> to standard ones.


My fault, my expression is not very complete.

This is not a replacement, but just want to say, there are only three cases of
virtnet_xdp_handler. Caller only needs to handle this three cases. Instead
of paying attention to the detailed return results of XDP.

In addition, virtnet_xdp_handler returns XDP_TX, but in fact, the work of XDP_TX
is already done in Virtnet_xdp_handler. Caller does not need to do anything for
XDP_TX, giving people a feeling, XDP_TX does not need to be processed. I think
it is not good.

Thanks.



>
> > ---
> >  drivers/net/virtio_net.c | 42 ++++++++++++++++++++++++++--------------
> >  1 file changed, 27 insertions(+), 15 deletions(-)
> >
> > diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
> > index 0fa64c314ea7..4dfdc211d355 100644
> > --- a/drivers/net/virtio_net.c
> > +++ b/drivers/net/virtio_net.c
> > @@ -301,6 +301,15 @@ struct padded_vnet_hdr {
> >  	char padding[12];
> >  };
> >
> > +enum {
> > +	/* xdp pass */
> > +	VIRTNET_XDP_RES_PASS,
> > +	/* drop packet. the caller needs to release the page. */
> > +	VIRTNET_XDP_RES_DROP,
> > +	/* packet is consumed by xdp. the caller needs to do nothing. */
> > +	VIRTNET_XDP_RES_CONSUMED,
> > +};
> > +
> >  static void virtnet_rq_free_unused_buf(struct virtqueue *vq, void *buf);
> >  static void virtnet_sq_free_unused_buf(struct virtqueue *vq, void *buf);
> >
> > @@ -803,14 +812,14 @@ static int virtnet_xdp_handler(struct bpf_prog *xdp_prog, struct xdp_buff *xdp,
> >
> >  	switch (act) {
> >  	case XDP_PASS:
> > -		return act;
> > +		return VIRTNET_XDP_RES_PASS;
> >
> >  	case XDP_TX:
> >  		stats->xdp_tx++;
> >  		xdpf = xdp_convert_buff_to_frame(xdp);
> >  		if (unlikely(!xdpf)) {
> >  			netdev_dbg(dev, "convert buff to frame failed for xdp\n");
> > -			return XDP_DROP;
> > +			return VIRTNET_XDP_RES_DROP;
> >  		}
> >
> >  		err = virtnet_xdp_xmit(dev, 1, &xdpf, 0);
> > @@ -818,19 +827,20 @@ static int virtnet_xdp_handler(struct bpf_prog *xdp_prog, struct xdp_buff *xdp,
> >  			xdp_return_frame_rx_napi(xdpf);
> >  		} else if (unlikely(err < 0)) {
> >  			trace_xdp_exception(dev, xdp_prog, act);
> > -			return XDP_DROP;
> > +			return VIRTNET_XDP_RES_DROP;
> >  		}
> > +
> >  		*xdp_xmit |= VIRTIO_XDP_TX;
> > -		return act;
> > +		return VIRTNET_XDP_RES_CONSUMED;
> >
> >  	case XDP_REDIRECT:
> >  		stats->xdp_redirects++;
> >  		err = xdp_do_redirect(dev, xdp, xdp_prog);
> >  		if (err)
> > -			return XDP_DROP;
> > +			return VIRTNET_XDP_RES_DROP;
> >
> >  		*xdp_xmit |= VIRTIO_XDP_REDIR;
> > -		return act;
> > +		return VIRTNET_XDP_RES_CONSUMED;
> >
> >  	default:
> >  		bpf_warn_invalid_xdp_action(dev, xdp_prog, act);
> > @@ -839,7 +849,7 @@ static int virtnet_xdp_handler(struct bpf_prog *xdp_prog, struct xdp_buff *xdp,
> >  		trace_xdp_exception(dev, xdp_prog, act);
> >  		fallthrough;
> >  	case XDP_DROP:
> > -		return XDP_DROP;
> > +		return VIRTNET_XDP_RES_DROP;
> >  	}
> >  }
> >
> > @@ -987,17 +997,18 @@ static struct sk_buff *receive_small(struct net_device *dev,
> >  		act = virtnet_xdp_handler(xdp_prog, &xdp, dev, xdp_xmit, stats);
> >
> >  		switch (act) {
> > -		case XDP_PASS:
> > +		case VIRTNET_XDP_RES_PASS:
> >  			/* Recalculate length in case bpf program changed it */
> >  			delta = orig_data - xdp.data;
> >  			len = xdp.data_end - xdp.data;
> >  			metasize = xdp.data - xdp.data_meta;
> >  			break;
> > -		case XDP_TX:
> > -		case XDP_REDIRECT:
> > +
> > +		case VIRTNET_XDP_RES_CONSUMED:
> >  			rcu_read_unlock();
> >  			goto xdp_xmit;
> > -		default:
> > +
> > +		case VIRTNET_XDP_RES_DROP:
> >  			goto err_xdp;
> >  		}
> >  	}
> > @@ -1324,18 +1335,19 @@ static struct sk_buff *receive_mergeable(struct net_device *dev,
> >  		act = virtnet_xdp_handler(xdp_prog, &xdp, dev, xdp_xmit, stats);
> >
> >  		switch (act) {
> > -		case XDP_PASS:
> > +		case VIRTNET_XDP_RES_PASS:
> >  			head_skb = build_skb_from_xdp_buff(dev, vi, &xdp, xdp_frags_truesz);
> >  			if (unlikely(!head_skb))
> >  				goto err_xdp_frags;
> >
> >  			rcu_read_unlock();
> >  			return head_skb;
> > -		case XDP_TX:
> > -		case XDP_REDIRECT:
> > +
> > +		case VIRTNET_XDP_RES_CONSUMED:
> >  			rcu_read_unlock();
> >  			goto xdp_xmit;
> > -		default:
> > +
> > +		case VIRTNET_XDP_RES_DROP:
> >  			break;
> >  		}
> >  err_xdp_frags:
> > --
> > 2.32.0.3.g01195cf9f
>
_______________________________________________
Virtualization mailing list
Virtualization@lists.linux-foundation.org
https://lists.linuxfoundation.org/mailman/listinfo/virtualization

Re: [PATCH net-next v2 05/14] virtio_net: introduce xdp res enums

[Michael S. Tsirkin]

On Fri, Apr 21, 2023 at 03:24:00PM +0800, Xuan Zhuo wrote:
> On Fri, 21 Apr 2023 03:00:15 -0400, "Michael S. Tsirkin" <mst@redhat.com> wrote:
> > On Tue, Apr 18, 2023 at 02:53:18PM +0800, Xuan Zhuo wrote:
> > > virtnet_xdp_handler() is to process all the logic related to XDP. The
> > > caller only needs to care about how to deal with the buf. So this commit
> > > introduces new enums:
> > >
> > > 1. VIRTNET_XDP_RES_PASS: make skb by the buf
> > > 2. VIRTNET_XDP_RES_DROP: xdp return drop action or some error, caller
> > >    should release the buf
> > > 3. VIRTNET_XDP_RES_CONSUMED: xdp consumed the buf, the caller doesnot to
> > >    do anything
> > >
> > > Signed-off-by: Xuan Zhuo <xuanzhuo@linux.alibaba.com>
> >
> >
> > I am not excited about using virtio specific enums then translating
> > to standard ones.
> 
> 
> My fault, my expression is not very complete.
> 
> This is not a replacement, but just want to say, there are only three cases of
> virtnet_xdp_handler. Caller only needs to handle this three cases. Instead
> of paying attention to the detailed return results of XDP.
> 
> In addition, virtnet_xdp_handler returns XDP_TX, but in fact, the work of XDP_TX
> is already done in Virtnet_xdp_handler. Caller does not need to do anything for
> XDP_TX, giving people a feeling, XDP_TX does not need to be processed. I think
> it is not good.
> 
> Thanks.

I don't really get it, sorry. If it's possible to stick to
XDP return codes, that is preferable.

> 
> 
> >
> > > ---
> > >  drivers/net/virtio_net.c | 42 ++++++++++++++++++++++++++--------------
> > >  1 file changed, 27 insertions(+), 15 deletions(-)
> > >
> > > diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
> > > index 0fa64c314ea7..4dfdc211d355 100644
> > > --- a/drivers/net/virtio_net.c
> > > +++ b/drivers/net/virtio_net.c
> > > @@ -301,6 +301,15 @@ struct padded_vnet_hdr {
> > >  	char padding[12];
> > >  };
> > >
> > > +enum {
> > > +	/* xdp pass */
> > > +	VIRTNET_XDP_RES_PASS,
> > > +	/* drop packet. the caller needs to release the page. */
> > > +	VIRTNET_XDP_RES_DROP,
> > > +	/* packet is consumed by xdp. the caller needs to do nothing. */
> > > +	VIRTNET_XDP_RES_CONSUMED,
> > > +};
> > > +
> > >  static void virtnet_rq_free_unused_buf(struct virtqueue *vq, void *buf);
> > >  static void virtnet_sq_free_unused_buf(struct virtqueue *vq, void *buf);
> > >
> > > @@ -803,14 +812,14 @@ static int virtnet_xdp_handler(struct bpf_prog *xdp_prog, struct xdp_buff *xdp,
> > >
> > >  	switch (act) {
> > >  	case XDP_PASS:
> > > -		return act;
> > > +		return VIRTNET_XDP_RES_PASS;
> > >
> > >  	case XDP_TX:
> > >  		stats->xdp_tx++;
> > >  		xdpf = xdp_convert_buff_to_frame(xdp);
> > >  		if (unlikely(!xdpf)) {
> > >  			netdev_dbg(dev, "convert buff to frame failed for xdp\n");
> > > -			return XDP_DROP;
> > > +			return VIRTNET_XDP_RES_DROP;
> > >  		}
> > >
> > >  		err = virtnet_xdp_xmit(dev, 1, &xdpf, 0);
> > > @@ -818,19 +827,20 @@ static int virtnet_xdp_handler(struct bpf_prog *xdp_prog, struct xdp_buff *xdp,
> > >  			xdp_return_frame_rx_napi(xdpf);
> > >  		} else if (unlikely(err < 0)) {
> > >  			trace_xdp_exception(dev, xdp_prog, act);
> > > -			return XDP_DROP;
> > > +			return VIRTNET_XDP_RES_DROP;
> > >  		}
> > > +
> > >  		*xdp_xmit |= VIRTIO_XDP_TX;
> > > -		return act;
> > > +		return VIRTNET_XDP_RES_CONSUMED;
> > >
> > >  	case XDP_REDIRECT:
> > >  		stats->xdp_redirects++;
> > >  		err = xdp_do_redirect(dev, xdp, xdp_prog);
> > >  		if (err)
> > > -			return XDP_DROP;
> > > +			return VIRTNET_XDP_RES_DROP;
> > >
> > >  		*xdp_xmit |= VIRTIO_XDP_REDIR;
> > > -		return act;
> > > +		return VIRTNET_XDP_RES_CONSUMED;
> > >
> > >  	default:
> > >  		bpf_warn_invalid_xdp_action(dev, xdp_prog, act);
> > > @@ -839,7 +849,7 @@ static int virtnet_xdp_handler(struct bpf_prog *xdp_prog, struct xdp_buff *xdp,
> > >  		trace_xdp_exception(dev, xdp_prog, act);
> > >  		fallthrough;
> > >  	case XDP_DROP:
> > > -		return XDP_DROP;
> > > +		return VIRTNET_XDP_RES_DROP;
> > >  	}
> > >  }
> > >
> > > @@ -987,17 +997,18 @@ static struct sk_buff *receive_small(struct net_device *dev,
> > >  		act = virtnet_xdp_handler(xdp_prog, &xdp, dev, xdp_xmit, stats);
> > >
> > >  		switch (act) {
> > > -		case XDP_PASS:
> > > +		case VIRTNET_XDP_RES_PASS:
> > >  			/* Recalculate length in case bpf program changed it */
> > >  			delta = orig_data - xdp.data;
> > >  			len = xdp.data_end - xdp.data;
> > >  			metasize = xdp.data - xdp.data_meta;
> > >  			break;
> > > -		case XDP_TX:
> > > -		case XDP_REDIRECT:
> > > +
> > > +		case VIRTNET_XDP_RES_CONSUMED:
> > >  			rcu_read_unlock();
> > >  			goto xdp_xmit;
> > > -		default:
> > > +
> > > +		case VIRTNET_XDP_RES_DROP:
> > >  			goto err_xdp;
> > >  		}
> > >  	}
> > > @@ -1324,18 +1335,19 @@ static struct sk_buff *receive_mergeable(struct net_device *dev,
> > >  		act = virtnet_xdp_handler(xdp_prog, &xdp, dev, xdp_xmit, stats);
> > >
> > >  		switch (act) {
> > > -		case XDP_PASS:
> > > +		case VIRTNET_XDP_RES_PASS:
> > >  			head_skb = build_skb_from_xdp_buff(dev, vi, &xdp, xdp_frags_truesz);
> > >  			if (unlikely(!head_skb))
> > >  				goto err_xdp_frags;
> > >
> > >  			rcu_read_unlock();
> > >  			return head_skb;
> > > -		case XDP_TX:
> > > -		case XDP_REDIRECT:
> > > +
> > > +		case VIRTNET_XDP_RES_CONSUMED:
> > >  			rcu_read_unlock();
> > >  			goto xdp_xmit;
> > > -		default:
> > > +
> > > +		case VIRTNET_XDP_RES_DROP:
> > >  			break;
> > >  		}
> > >  err_xdp_frags:
> > > --
> > > 2.32.0.3.g01195cf9f
> >

_______________________________________________
Virtualization mailing list
Virtualization@lists.linux-foundation.org
https://lists.linuxfoundation.org/mailman/listinfo/virtualization

Re: [PATCH net-next v2 05/14] virtio_net: introduce xdp res enums

[Xuan Zhuo]

On Fri, 21 Apr 2023 07:54:11 -0400, "Michael S. Tsirkin" <mst@redhat.com> wrote:
> On Fri, Apr 21, 2023 at 03:24:00PM +0800, Xuan Zhuo wrote:
> > On Fri, 21 Apr 2023 03:00:15 -0400, "Michael S. Tsirkin" <mst@redhat.com> wrote:
> > > On Tue, Apr 18, 2023 at 02:53:18PM +0800, Xuan Zhuo wrote:
> > > > virtnet_xdp_handler() is to process all the logic related to XDP. The
> > > > caller only needs to care about how to deal with the buf. So this commit
> > > > introduces new enums:
> > > >
> > > > 1. VIRTNET_XDP_RES_PASS: make skb by the buf
> > > > 2. VIRTNET_XDP_RES_DROP: xdp return drop action or some error, caller
> > > >    should release the buf
> > > > 3. VIRTNET_XDP_RES_CONSUMED: xdp consumed the buf, the caller doesnot to
> > > >    do anything
> > > >
> > > > Signed-off-by: Xuan Zhuo <xuanzhuo@linux.alibaba.com>
> > >
> > >
> > > I am not excited about using virtio specific enums then translating
> > > to standard ones.
> >
> >
> > My fault, my expression is not very complete.
> >
> > This is not a replacement, but just want to say, there are only three cases of
> > virtnet_xdp_handler. Caller only needs to handle this three cases. Instead
> > of paying attention to the detailed return results of XDP.
> >
> > In addition, virtnet_xdp_handler returns XDP_TX, but in fact, the work of XDP_TX
> > is already done in Virtnet_xdp_handler. Caller does not need to do anything for
> > XDP_TX, giving people a feeling, XDP_TX does not need to be processed. I think
> > it is not good.
> >
> > Thanks.
>
> I don't really get it, sorry. If it's possible to stick to
> XDP return codes, that is preferable.

Although, I still think that it would be better to use VIRTNET_XDP_RES_*, and
other drivers did the same thing. If you still insist, I can remove this commit.

In addition, I think squashing this patch to the previous patch may be more
suitable. Just like I done in last patchset. @Jason

Because what I want to express is that virtnet_xdp_handler() has only three
kinds of returns, caller does not need to pay attention to the details of XDP
returns.

Thanks.


>
> >
> >
> > >
> > > > ---
> > > >  drivers/net/virtio_net.c | 42 ++++++++++++++++++++++++++--------------
> > > >  1 file changed, 27 insertions(+), 15 deletions(-)
> > > >
> > > > diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
> > > > index 0fa64c314ea7..4dfdc211d355 100644
> > > > --- a/drivers/net/virtio_net.c
> > > > +++ b/drivers/net/virtio_net.c
> > > > @@ -301,6 +301,15 @@ struct padded_vnet_hdr {
> > > >  	char padding[12];
> > > >  };
> > > >
> > > > +enum {
> > > > +	/* xdp pass */
> > > > +	VIRTNET_XDP_RES_PASS,
> > > > +	/* drop packet. the caller needs to release the page. */
> > > > +	VIRTNET_XDP_RES_DROP,
> > > > +	/* packet is consumed by xdp. the caller needs to do nothing. */
> > > > +	VIRTNET_XDP_RES_CONSUMED,
> > > > +};
> > > > +
> > > >  static void virtnet_rq_free_unused_buf(struct virtqueue *vq, void *buf);
> > > >  static void virtnet_sq_free_unused_buf(struct virtqueue *vq, void *buf);
> > > >
> > > > @@ -803,14 +812,14 @@ static int virtnet_xdp_handler(struct bpf_prog *xdp_prog, struct xdp_buff *xdp,
> > > >
> > > >  	switch (act) {
> > > >  	case XDP_PASS:
> > > > -		return act;
> > > > +		return VIRTNET_XDP_RES_PASS;
> > > >
> > > >  	case XDP_TX:
> > > >  		stats->xdp_tx++;
> > > >  		xdpf = xdp_convert_buff_to_frame(xdp);
> > > >  		if (unlikely(!xdpf)) {
> > > >  			netdev_dbg(dev, "convert buff to frame failed for xdp\n");
> > > > -			return XDP_DROP;
> > > > +			return VIRTNET_XDP_RES_DROP;
> > > >  		}
> > > >
> > > >  		err = virtnet_xdp_xmit(dev, 1, &xdpf, 0);
> > > > @@ -818,19 +827,20 @@ static int virtnet_xdp_handler(struct bpf_prog *xdp_prog, struct xdp_buff *xdp,
> > > >  			xdp_return_frame_rx_napi(xdpf);
> > > >  		} else if (unlikely(err < 0)) {
> > > >  			trace_xdp_exception(dev, xdp_prog, act);
> > > > -			return XDP_DROP;
> > > > +			return VIRTNET_XDP_RES_DROP;
> > > >  		}
> > > > +
> > > >  		*xdp_xmit |= VIRTIO_XDP_TX;
> > > > -		return act;
> > > > +		return VIRTNET_XDP_RES_CONSUMED;
> > > >
> > > >  	case XDP_REDIRECT:
> > > >  		stats->xdp_redirects++;
> > > >  		err = xdp_do_redirect(dev, xdp, xdp_prog);
> > > >  		if (err)
> > > > -			return XDP_DROP;
> > > > +			return VIRTNET_XDP_RES_DROP;
> > > >
> > > >  		*xdp_xmit |= VIRTIO_XDP_REDIR;
> > > > -		return act;
> > > > +		return VIRTNET_XDP_RES_CONSUMED;
> > > >
> > > >  	default:
> > > >  		bpf_warn_invalid_xdp_action(dev, xdp_prog, act);
> > > > @@ -839,7 +849,7 @@ static int virtnet_xdp_handler(struct bpf_prog *xdp_prog, struct xdp_buff *xdp,
> > > >  		trace_xdp_exception(dev, xdp_prog, act);
> > > >  		fallthrough;
> > > >  	case XDP_DROP:
> > > > -		return XDP_DROP;
> > > > +		return VIRTNET_XDP_RES_DROP;
> > > >  	}
> > > >  }
> > > >
> > > > @@ -987,17 +997,18 @@ static struct sk_buff *receive_small(struct net_device *dev,
> > > >  		act = virtnet_xdp_handler(xdp_prog, &xdp, dev, xdp_xmit, stats);
> > > >
> > > >  		switch (act) {
> > > > -		case XDP_PASS:
> > > > +		case VIRTNET_XDP_RES_PASS:
> > > >  			/* Recalculate length in case bpf program changed it */
> > > >  			delta = orig_data - xdp.data;
> > > >  			len = xdp.data_end - xdp.data;
> > > >  			metasize = xdp.data - xdp.data_meta;
> > > >  			break;
> > > > -		case XDP_TX:
> > > > -		case XDP_REDIRECT:
> > > > +
> > > > +		case VIRTNET_XDP_RES_CONSUMED:
> > > >  			rcu_read_unlock();
> > > >  			goto xdp_xmit;
> > > > -		default:
> > > > +
> > > > +		case VIRTNET_XDP_RES_DROP:
> > > >  			goto err_xdp;
> > > >  		}
> > > >  	}
> > > > @@ -1324,18 +1335,19 @@ static struct sk_buff *receive_mergeable(struct net_device *dev,
> > > >  		act = virtnet_xdp_handler(xdp_prog, &xdp, dev, xdp_xmit, stats);
> > > >
> > > >  		switch (act) {
> > > > -		case XDP_PASS:
> > > > +		case VIRTNET_XDP_RES_PASS:
> > > >  			head_skb = build_skb_from_xdp_buff(dev, vi, &xdp, xdp_frags_truesz);
> > > >  			if (unlikely(!head_skb))
> > > >  				goto err_xdp_frags;
> > > >
> > > >  			rcu_read_unlock();
> > > >  			return head_skb;
> > > > -		case XDP_TX:
> > > > -		case XDP_REDIRECT:
> > > > +
> > > > +		case VIRTNET_XDP_RES_CONSUMED:
> > > >  			rcu_read_unlock();
> > > >  			goto xdp_xmit;
> > > > -		default:
> > > > +
> > > > +		case VIRTNET_XDP_RES_DROP:
> > > >  			break;
> > > >  		}
> > > >  err_xdp_frags:
> > > > --
> > > > 2.32.0.3.g01195cf9f
> > >
>
_______________________________________________
Virtualization mailing list
Virtualization@lists.linux-foundation.org
https://lists.linuxfoundation.org/mailman/listinfo/virtualization

Re: [PATCH net-next v2 05/14] virtio_net: introduce xdp res enums

[Jason Wang]

On Sun, Apr 23, 2023 at 10:05 AM Xuan Zhuo <xuanzhuo@linux.alibaba.com> wrote:
>
> On Fri, 21 Apr 2023 07:54:11 -0400, "Michael S. Tsirkin" <mst@redhat.com> wrote:
> > On Fri, Apr 21, 2023 at 03:24:00PM +0800, Xuan Zhuo wrote:
> > > On Fri, 21 Apr 2023 03:00:15 -0400, "Michael S. Tsirkin" <mst@redhat.com> wrote:
> > > > On Tue, Apr 18, 2023 at 02:53:18PM +0800, Xuan Zhuo wrote:
> > > > > virtnet_xdp_handler() is to process all the logic related to XDP. The
> > > > > caller only needs to care about how to deal with the buf. So this commit
> > > > > introduces new enums:
> > > > >
> > > > > 1. VIRTNET_XDP_RES_PASS: make skb by the buf
> > > > > 2. VIRTNET_XDP_RES_DROP: xdp return drop action or some error, caller
> > > > >    should release the buf
> > > > > 3. VIRTNET_XDP_RES_CONSUMED: xdp consumed the buf, the caller doesnot to
> > > > >    do anything
> > > > >
> > > > > Signed-off-by: Xuan Zhuo <xuanzhuo@linux.alibaba.com>
> > > >
> > > >
> > > > I am not excited about using virtio specific enums then translating
> > > > to standard ones.
> > >
> > >
> > > My fault, my expression is not very complete.
> > >
> > > This is not a replacement, but just want to say, there are only three cases of
> > > virtnet_xdp_handler. Caller only needs to handle this three cases. Instead
> > > of paying attention to the detailed return results of XDP.
> > >
> > > In addition, virtnet_xdp_handler returns XDP_TX, but in fact, the work of XDP_TX
> > > is already done in Virtnet_xdp_handler. Caller does not need to do anything for
> > > XDP_TX, giving people a feeling, XDP_TX does not need to be processed. I think
> > > it is not good.
> > >
> > > Thanks.
> >
> > I don't really get it, sorry. If it's possible to stick to
> > XDP return codes, that is preferable.
>
> Although, I still think that it would be better to use VIRTNET_XDP_RES_*, and
> other drivers did the same thing. If you still insist, I can remove this commit.
>

I second for sticking to XDP return codes.

(Since I didn't really get what's wrong with that).

Thanks

> In addition, I think squashing this patch to the previous patch may be more
> suitable. Just like I done in last patchset. @Jason
>
> Because what I want to express is that virtnet_xdp_handler() has only three
> kinds of returns, caller does not need to pay attention to the details of XDP
> returns.
>
> Thanks.
>
>
> >
> > >
> > >
> > > >
> > > > > ---
> > > > >  drivers/net/virtio_net.c | 42 ++++++++++++++++++++++++++--------------
> > > > >  1 file changed, 27 insertions(+), 15 deletions(-)
> > > > >
> > > > > diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
> > > > > index 0fa64c314ea7..4dfdc211d355 100644
> > > > > --- a/drivers/net/virtio_net.c
> > > > > +++ b/drivers/net/virtio_net.c
> > > > > @@ -301,6 +301,15 @@ struct padded_vnet_hdr {
> > > > >         char padding[12];
> > > > >  };
> > > > >
> > > > > +enum {
> > > > > +       /* xdp pass */
> > > > > +       VIRTNET_XDP_RES_PASS,
> > > > > +       /* drop packet. the caller needs to release the page. */
> > > > > +       VIRTNET_XDP_RES_DROP,
> > > > > +       /* packet is consumed by xdp. the caller needs to do nothing. */
> > > > > +       VIRTNET_XDP_RES_CONSUMED,
> > > > > +};
> > > > > +
> > > > >  static void virtnet_rq_free_unused_buf(struct virtqueue *vq, void *buf);
> > > > >  static void virtnet_sq_free_unused_buf(struct virtqueue *vq, void *buf);
> > > > >
> > > > > @@ -803,14 +812,14 @@ static int virtnet_xdp_handler(struct bpf_prog *xdp_prog, struct xdp_buff *xdp,
> > > > >
> > > > >         switch (act) {
> > > > >         case XDP_PASS:
> > > > > -               return act;
> > > > > +               return VIRTNET_XDP_RES_PASS;
> > > > >
> > > > >         case XDP_TX:
> > > > >                 stats->xdp_tx++;
> > > > >                 xdpf = xdp_convert_buff_to_frame(xdp);
> > > > >                 if (unlikely(!xdpf)) {
> > > > >                         netdev_dbg(dev, "convert buff to frame failed for xdp\n");
> > > > > -                       return XDP_DROP;
> > > > > +                       return VIRTNET_XDP_RES_DROP;
> > > > >                 }
> > > > >
> > > > >                 err = virtnet_xdp_xmit(dev, 1, &xdpf, 0);
> > > > > @@ -818,19 +827,20 @@ static int virtnet_xdp_handler(struct bpf_prog *xdp_prog, struct xdp_buff *xdp,
> > > > >                         xdp_return_frame_rx_napi(xdpf);
> > > > >                 } else if (unlikely(err < 0)) {
> > > > >                         trace_xdp_exception(dev, xdp_prog, act);
> > > > > -                       return XDP_DROP;
> > > > > +                       return VIRTNET_XDP_RES_DROP;
> > > > >                 }
> > > > > +
> > > > >                 *xdp_xmit |= VIRTIO_XDP_TX;
> > > > > -               return act;
> > > > > +               return VIRTNET_XDP_RES_CONSUMED;
> > > > >
> > > > >         case XDP_REDIRECT:
> > > > >                 stats->xdp_redirects++;
> > > > >                 err = xdp_do_redirect(dev, xdp, xdp_prog);
> > > > >                 if (err)
> > > > > -                       return XDP_DROP;
> > > > > +                       return VIRTNET_XDP_RES_DROP;
> > > > >
> > > > >                 *xdp_xmit |= VIRTIO_XDP_REDIR;
> > > > > -               return act;
> > > > > +               return VIRTNET_XDP_RES_CONSUMED;
> > > > >
> > > > >         default:
> > > > >                 bpf_warn_invalid_xdp_action(dev, xdp_prog, act);
> > > > > @@ -839,7 +849,7 @@ static int virtnet_xdp_handler(struct bpf_prog *xdp_prog, struct xdp_buff *xdp,
> > > > >                 trace_xdp_exception(dev, xdp_prog, act);
> > > > >                 fallthrough;
> > > > >         case XDP_DROP:
> > > > > -               return XDP_DROP;
> > > > > +               return VIRTNET_XDP_RES_DROP;
> > > > >         }
> > > > >  }
> > > > >
> > > > > @@ -987,17 +997,18 @@ static struct sk_buff *receive_small(struct net_device *dev,
> > > > >                 act = virtnet_xdp_handler(xdp_prog, &xdp, dev, xdp_xmit, stats);
> > > > >
> > > > >                 switch (act) {
> > > > > -               case XDP_PASS:
> > > > > +               case VIRTNET_XDP_RES_PASS:
> > > > >                         /* Recalculate length in case bpf program changed it */
> > > > >                         delta = orig_data - xdp.data;
> > > > >                         len = xdp.data_end - xdp.data;
> > > > >                         metasize = xdp.data - xdp.data_meta;
> > > > >                         break;
> > > > > -               case XDP_TX:
> > > > > -               case XDP_REDIRECT:
> > > > > +
> > > > > +               case VIRTNET_XDP_RES_CONSUMED:
> > > > >                         rcu_read_unlock();
> > > > >                         goto xdp_xmit;
> > > > > -               default:
> > > > > +
> > > > > +               case VIRTNET_XDP_RES_DROP:
> > > > >                         goto err_xdp;
> > > > >                 }
> > > > >         }
> > > > > @@ -1324,18 +1335,19 @@ static struct sk_buff *receive_mergeable(struct net_device *dev,
> > > > >                 act = virtnet_xdp_handler(xdp_prog, &xdp, dev, xdp_xmit, stats);
> > > > >
> > > > >                 switch (act) {
> > > > > -               case XDP_PASS:
> > > > > +               case VIRTNET_XDP_RES_PASS:
> > > > >                         head_skb = build_skb_from_xdp_buff(dev, vi, &xdp, xdp_frags_truesz);
> > > > >                         if (unlikely(!head_skb))
> > > > >                                 goto err_xdp_frags;
> > > > >
> > > > >                         rcu_read_unlock();
> > > > >                         return head_skb;
> > > > > -               case XDP_TX:
> > > > > -               case XDP_REDIRECT:
> > > > > +
> > > > > +               case VIRTNET_XDP_RES_CONSUMED:
> > > > >                         rcu_read_unlock();
> > > > >                         goto xdp_xmit;
> > > > > -               default:
> > > > > +
> > > > > +               case VIRTNET_XDP_RES_DROP:
> > > > >                         break;
> > > > >                 }
> > > > >  err_xdp_frags:
> > > > > --
> > > > > 2.32.0.3.g01195cf9f
> > > >
> >
>

_______________________________________________
Virtualization mailing list
Virtualization@lists.linux-foundation.org
https://lists.linuxfoundation.org/mailman/listinfo/virtualization

Re: [PATCH net-next v2 05/14] virtio_net: introduce xdp res enums

[Xuan Zhuo]

On Sun, 23 Apr 2023 13:28:07 +0800, Jason Wang <jasowang@redhat.com> wrote:
> On Sun, Apr 23, 2023 at 10:05 AM Xuan Zhuo <xuanzhuo@linux.alibaba.com> wrote:
> >
> > On Fri, 21 Apr 2023 07:54:11 -0400, "Michael S. Tsirkin" <mst@redhat.com> wrote:
> > > On Fri, Apr 21, 2023 at 03:24:00PM +0800, Xuan Zhuo wrote:
> > > > On Fri, 21 Apr 2023 03:00:15 -0400, "Michael S. Tsirkin" <mst@redhat.com> wrote:
> > > > > On Tue, Apr 18, 2023 at 02:53:18PM +0800, Xuan Zhuo wrote:
> > > > > > virtnet_xdp_handler() is to process all the logic related to XDP. The
> > > > > > caller only needs to care about how to deal with the buf. So this commit
> > > > > > introduces new enums:
> > > > > >
> > > > > > 1. VIRTNET_XDP_RES_PASS: make skb by the buf
> > > > > > 2. VIRTNET_XDP_RES_DROP: xdp return drop action or some error, caller
> > > > > >    should release the buf
> > > > > > 3. VIRTNET_XDP_RES_CONSUMED: xdp consumed the buf, the caller doesnot to
> > > > > >    do anything
> > > > > >
> > > > > > Signed-off-by: Xuan Zhuo <xuanzhuo@linux.alibaba.com>
> > > > >
> > > > >
> > > > > I am not excited about using virtio specific enums then translating
> > > > > to standard ones.
> > > >
> > > >
> > > > My fault, my expression is not very complete.
> > > >
> > > > This is not a replacement, but just want to say, there are only three cases of
> > > > virtnet_xdp_handler. Caller only needs to handle this three cases. Instead
> > > > of paying attention to the detailed return results of XDP.
> > > >
> > > > In addition, virtnet_xdp_handler returns XDP_TX, but in fact, the work of XDP_TX
> > > > is already done in Virtnet_xdp_handler. Caller does not need to do anything for
> > > > XDP_TX, giving people a feeling, XDP_TX does not need to be processed. I think
> > > > it is not good.
> > > >
> > > > Thanks.
> > >
> > > I don't really get it, sorry. If it's possible to stick to
> > > XDP return codes, that is preferable.
> >
> > Although, I still think that it would be better to use VIRTNET_XDP_RES_*, and
> > other drivers did the same thing. If you still insist, I can remove this commit.
> >
>
> I second for sticking to XDP return codes.
>
> (Since I didn't really get what's wrong with that).


OK. If nobody has other advice I will remove this commit in next verion.

Thanks.


>
> Thanks
>
> > In addition, I think squashing this patch to the previous patch may be more
> > suitable. Just like I done in last patchset. @Jason
> >
> > Because what I want to express is that virtnet_xdp_handler() has only three
> > kinds of returns, caller does not need to pay attention to the details of XDP
> > returns.
> >
> > Thanks.
> >
> >
> > >
> > > >
> > > >
> > > > >
> > > > > > ---
> > > > > >  drivers/net/virtio_net.c | 42 ++++++++++++++++++++++++++--------------
> > > > > >  1 file changed, 27 insertions(+), 15 deletions(-)
> > > > > >
> > > > > > diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
> > > > > > index 0fa64c314ea7..4dfdc211d355 100644
> > > > > > --- a/drivers/net/virtio_net.c
> > > > > > +++ b/drivers/net/virtio_net.c
> > > > > > @@ -301,6 +301,15 @@ struct padded_vnet_hdr {
> > > > > >         char padding[12];
> > > > > >  };
> > > > > >
> > > > > > +enum {
> > > > > > +       /* xdp pass */
> > > > > > +       VIRTNET_XDP_RES_PASS,
> > > > > > +       /* drop packet. the caller needs to release the page. */
> > > > > > +       VIRTNET_XDP_RES_DROP,
> > > > > > +       /* packet is consumed by xdp. the caller needs to do nothing. */
> > > > > > +       VIRTNET_XDP_RES_CONSUMED,
> > > > > > +};
> > > > > > +
> > > > > >  static void virtnet_rq_free_unused_buf(struct virtqueue *vq, void *buf);
> > > > > >  static void virtnet_sq_free_unused_buf(struct virtqueue *vq, void *buf);
> > > > > >
> > > > > > @@ -803,14 +812,14 @@ static int virtnet_xdp_handler(struct bpf_prog *xdp_prog, struct xdp_buff *xdp,
> > > > > >
> > > > > >         switch (act) {
> > > > > >         case XDP_PASS:
> > > > > > -               return act;
> > > > > > +               return VIRTNET_XDP_RES_PASS;
> > > > > >
> > > > > >         case XDP_TX:
> > > > > >                 stats->xdp_tx++;
> > > > > >                 xdpf = xdp_convert_buff_to_frame(xdp);
> > > > > >                 if (unlikely(!xdpf)) {
> > > > > >                         netdev_dbg(dev, "convert buff to frame failed for xdp\n");
> > > > > > -                       return XDP_DROP;
> > > > > > +                       return VIRTNET_XDP_RES_DROP;
> > > > > >                 }
> > > > > >
> > > > > >                 err = virtnet_xdp_xmit(dev, 1, &xdpf, 0);
> > > > > > @@ -818,19 +827,20 @@ static int virtnet_xdp_handler(struct bpf_prog *xdp_prog, struct xdp_buff *xdp,
> > > > > >                         xdp_return_frame_rx_napi(xdpf);
> > > > > >                 } else if (unlikely(err < 0)) {
> > > > > >                         trace_xdp_exception(dev, xdp_prog, act);
> > > > > > -                       return XDP_DROP;
> > > > > > +                       return VIRTNET_XDP_RES_DROP;
> > > > > >                 }
> > > > > > +
> > > > > >                 *xdp_xmit |= VIRTIO_XDP_TX;
> > > > > > -               return act;
> > > > > > +               return VIRTNET_XDP_RES_CONSUMED;
> > > > > >
> > > > > >         case XDP_REDIRECT:
> > > > > >                 stats->xdp_redirects++;
> > > > > >                 err = xdp_do_redirect(dev, xdp, xdp_prog);
> > > > > >                 if (err)
> > > > > > -                       return XDP_DROP;
> > > > > > +                       return VIRTNET_XDP_RES_DROP;
> > > > > >
> > > > > >                 *xdp_xmit |= VIRTIO_XDP_REDIR;
> > > > > > -               return act;
> > > > > > +               return VIRTNET_XDP_RES_CONSUMED;
> > > > > >
> > > > > >         default:
> > > > > >                 bpf_warn_invalid_xdp_action(dev, xdp_prog, act);
> > > > > > @@ -839,7 +849,7 @@ static int virtnet_xdp_handler(struct bpf_prog *xdp_prog, struct xdp_buff *xdp,
> > > > > >                 trace_xdp_exception(dev, xdp_prog, act);
> > > > > >                 fallthrough;
> > > > > >         case XDP_DROP:
> > > > > > -               return XDP_DROP;
> > > > > > +               return VIRTNET_XDP_RES_DROP;
> > > > > >         }
> > > > > >  }
> > > > > >
> > > > > > @@ -987,17 +997,18 @@ static struct sk_buff *receive_small(struct net_device *dev,
> > > > > >                 act = virtnet_xdp_handler(xdp_prog, &xdp, dev, xdp_xmit, stats);
> > > > > >
> > > > > >                 switch (act) {
> > > > > > -               case XDP_PASS:
> > > > > > +               case VIRTNET_XDP_RES_PASS:
> > > > > >                         /* Recalculate length in case bpf program changed it */
> > > > > >                         delta = orig_data - xdp.data;
> > > > > >                         len = xdp.data_end - xdp.data;
> > > > > >                         metasize = xdp.data - xdp.data_meta;
> > > > > >                         break;
> > > > > > -               case XDP_TX:
> > > > > > -               case XDP_REDIRECT:
> > > > > > +
> > > > > > +               case VIRTNET_XDP_RES_CONSUMED:
> > > > > >                         rcu_read_unlock();
> > > > > >                         goto xdp_xmit;
> > > > > > -               default:
> > > > > > +
> > > > > > +               case VIRTNET_XDP_RES_DROP:
> > > > > >                         goto err_xdp;
> > > > > >                 }
> > > > > >         }
> > > > > > @@ -1324,18 +1335,19 @@ static struct sk_buff *receive_mergeable(struct net_device *dev,
> > > > > >                 act = virtnet_xdp_handler(xdp_prog, &xdp, dev, xdp_xmit, stats);
> > > > > >
> > > > > >                 switch (act) {
> > > > > > -               case XDP_PASS:
> > > > > > +               case VIRTNET_XDP_RES_PASS:
> > > > > >                         head_skb = build_skb_from_xdp_buff(dev, vi, &xdp, xdp_frags_truesz);
> > > > > >                         if (unlikely(!head_skb))
> > > > > >                                 goto err_xdp_frags;
> > > > > >
> > > > > >                         rcu_read_unlock();
> > > > > >                         return head_skb;
> > > > > > -               case XDP_TX:
> > > > > > -               case XDP_REDIRECT:
> > > > > > +
> > > > > > +               case VIRTNET_XDP_RES_CONSUMED:
> > > > > >                         rcu_read_unlock();
> > > > > >                         goto xdp_xmit;
> > > > > > -               default:
> > > > > > +
> > > > > > +               case VIRTNET_XDP_RES_DROP:
> > > > > >                         break;
> > > > > >                 }
> > > > > >  err_xdp_frags:
> > > > > > --
> > > > > > 2.32.0.3.g01195cf9f
> > > > >
> > >
> >
>
_______________________________________________
Virtualization mailing list
Virtualization@lists.linux-foundation.org
https://lists.linuxfoundation.org/mailman/listinfo/virtualization