Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > linux.kernel > #1546878 > unrolled thread

[PATCH net 0/9] several fixups for virtio-net XDP

Back to article view | Back to linux.kernel

Contents

  [PATCH net 0/9] several fixups for virtio-net XDP Jason Wang <jasowang@redhat.com> - 2016-12-23 15:40 +0100
    [PATCH net 4/9] virtio-net: correctly handle XDP_PASS for linearized packets Jason Wang <jasowang@redhat.com> - 2016-12-23 15:40 +0100
      Re: [PATCH net 4/9] virtio-net: correctly handle XDP_PASS for  linearized packets John Fastabend <john.fastabend@gmail.com> - 2016-12-23 17:00 +0100
        Re: [PATCH net 4/9] virtio-net: correctly handle XDP_PASS for  linearized packets Jason Wang <jasowang@redhat.com> - 2016-12-26 03:40 +0100
    [PATCH net 2/9] virtio-net: correctly xmit linearized page on XDP_TX Jason Wang <jasowang@redhat.com> - 2016-12-23 15:40 +0100
      Re: [PATCH net 2/9] virtio-net: correctly xmit linearized page on  XDP_TX John Fastabend <john.fastabend@gmail.com> - 2016-12-23 16:50 +0100
    [PATCH net 8/9] virtio-net: remove big packet XDP codes Jason Wang <jasowang@redhat.com> - 2016-12-23 15:40 +0100
    [PATCH net 9/9] virtio-net: XDP support for small buffers Jason Wang <jasowang@redhat.com> - 2016-12-23 15:50 +0100
      Re: [PATCH net 9/9] virtio-net: XDP support for small buffers John Fastabend <john.fastabend@gmail.com> - 2016-12-23 18:00 +0100
      Re: [PATCH net 9/9] virtio-net: XDP support for small buffers John Fastabend <john.fastabend@gmail.com> - 2017-01-02 23:50 +0100
        Re: [PATCH net 9/9] virtio-net: XDP support for small buffers Jason Wang <jasowang@redhat.com> - 2017-01-03 07:20 +0100
          Re: [PATCH net 9/9] virtio-net: XDP support for small buffers John Fastabend <john.fastabend@gmail.com> - 2017-01-03 17:50 +0100
    [PATCH net 5/9] virtio-net: unbreak csumed packets for XDP_PASS Jason Wang <jasowang@redhat.com> - 2016-12-23 15:50 +0100
      Re: [PATCH net 5/9] virtio-net: unbreak csumed packets for XDP_PASS John Fastabend <john.fastabend@gmail.com> - 2016-12-23 17:00 +0100
    Re: [PATCH net 0/9] several fixups for virtio-net XDP John Fastabend <john.fastabend@gmail.com> - 2016-12-23 18:20 +0100
      Re: [PATCH net 0/9] several fixups for virtio-net XDP Jason Wang <jasowang@redhat.com> - 2016-12-26 03:40 +0100
    Re: [PATCH net 0/9] several fixups for virtio-net XDP David Miller <davem@davemloft.net> - 2016-12-23 19:50 +0100

#1546878 — [PATCH net 0/9] several fixups for virtio-net XDP

Merry Xmas and a Happy New year to all:

This series tries to fixes several issues for virtio-net XDP which
could be categorized into several parts:

- fix several issues during XDP linearizing
- allow csumed packet to work for XDP_PASS
- make EWMA rxbuf size estimation works for XDP
- forbid XDP when GUEST_UFO is support
- remove big packet XDP support
- add XDP support or small buffer

Please see individual patches for details.

Thanks

Jason Wang (9):
  virtio-net: remove the warning before XDP linearizing
  virtio-net: correctly xmit linearized page on XDP_TX
  virtio-net: fix page miscount during XDP linearizing
  virtio-net: correctly handle XDP_PASS for linearized packets
  virtio-net: unbreak csumed packets for XDP_PASS
  virtio-net: make rx buf size estimation works for XDP
  virtio-net: forbid XDP when VIRTIO_NET_F_GUEST_UFO is support
  virtio-net: remove big packet XDP codes
  virtio-net: XDP support for small buffers

 drivers/net/virtio_net.c | 172 ++++++++++++++++++++++++++++-------------------
 1 file changed, 102 insertions(+), 70 deletions(-)

-- 
2.7.4

[toc] | [next] | [standalone]

#1546879 — [PATCH net 4/9] virtio-net: correctly handle XDP_PASS for linearized packets

When XDP_PASS were determined for linearized packets, we try to get
new buffers in the virtqueue and build skbs from them. This is wrong,
we should create skbs based on existed buffers instead. Fixing them by
creating skb based on xdp_page.

With this patch "ping 192.168.100.4 -s 3900 -M do" works for XDP_PASS.

Cc: John Fastabend <john.r.fastabend@intel.com>
Signed-off-by: Jason Wang <jasowang@redhat.com>
---
 drivers/net/virtio_net.c | 10 ++++++++--
 1 file changed, 8 insertions(+), 2 deletions(-)

diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
index 58ad40e..470293e 100644
--- a/drivers/net/virtio_net.c
+++ b/drivers/net/virtio_net.c
@@ -578,8 +578,14 @@ static struct sk_buff *receive_mergeable(struct net_device *dev,
 		act = do_xdp_prog(vi, rq, xdp_prog, xdp_page, offset, len);
 		switch (act) {
 		case XDP_PASS:
-			if (unlikely(xdp_page != page))
-				__free_pages(xdp_page, 0);
+			/* We can only create skb based on xdp_page. */
+			if (unlikely(xdp_page != page)) {
+				rcu_read_unlock();
+				put_page(page);
+				head_skb = page_to_skb(vi, rq, xdp_page,
+						       0, len, PAGE_SIZE);
+				return head_skb;
+			}
 			break;
 		case XDP_TX:
 			if (unlikely(xdp_page != page))
-- 
2.7.4

[toc] | [prev] | [next] | [standalone]

#1546902 — Re: [PATCH net 4/9] virtio-net: correctly handle XDP_PASS for linearized packets

On 16-12-23 06:37 AM, Jason Wang wrote:
> When XDP_PASS were determined for linearized packets, we try to get
> new buffers in the virtqueue and build skbs from them. This is wrong,
> we should create skbs based on existed buffers instead. Fixing them by
> creating skb based on xdp_page.
> 
> With this patch "ping 192.168.100.4 -s 3900 -M do" works for XDP_PASS.
> 
> Cc: John Fastabend <john.r.fastabend@intel.com>
> Signed-off-by: Jason Wang <jasowang@redhat.com>
> ---
>  drivers/net/virtio_net.c | 10 ++++++++--
>  1 file changed, 8 insertions(+), 2 deletions(-)
> 
> diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
> index 58ad40e..470293e 100644
> --- a/drivers/net/virtio_net.c
> +++ b/drivers/net/virtio_net.c
> @@ -578,8 +578,14 @@ static struct sk_buff *receive_mergeable(struct net_device *dev,
>  		act = do_xdp_prog(vi, rq, xdp_prog, xdp_page, offset, len);
>  		switch (act) {
>  		case XDP_PASS:
> -			if (unlikely(xdp_page != page))
> -				__free_pages(xdp_page, 0);
> +			/* We can only create skb based on xdp_page. */
> +			if (unlikely(xdp_page != page)) {
> +				rcu_read_unlock();
> +				put_page(page);
> +				head_skb = page_to_skb(vi, rq, xdp_page,
> +						       0, len, PAGE_SIZE);
> +				return head_skb;
> +			}
>  			break;
>  		case XDP_TX:
>  			if (unlikely(xdp_page != page))
> 

Great thanks. This was likely working before because of the memory
leak fixed in 3/9.

Acked-by: John Fastabend <john.r.fastabend@intel.com>

[toc] | [prev] | [next] | [standalone]

#1547239 — Re: [PATCH net 4/9] virtio-net: correctly handle XDP_PASS for linearized packets

On 2016年12月23日 23:57, John Fastabend wrote:
> On 16-12-23 06:37 AM, Jason Wang wrote:
>> When XDP_PASS were determined for linearized packets, we try to get
>> new buffers in the virtqueue and build skbs from them. This is wrong,
>> we should create skbs based on existed buffers instead. Fixing them by
>> creating skb based on xdp_page.
>>
>> With this patch "ping 192.168.100.4 -s 3900 -M do" works for XDP_PASS.
>>
>> Cc: John Fastabend <john.r.fastabend@intel.com>
>> Signed-off-by: Jason Wang <jasowang@redhat.com>
>> ---
>>   drivers/net/virtio_net.c | 10 ++++++++--
>>   1 file changed, 8 insertions(+), 2 deletions(-)
>>
>> diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
>> index 58ad40e..470293e 100644
>> --- a/drivers/net/virtio_net.c
>> +++ b/drivers/net/virtio_net.c
>> @@ -578,8 +578,14 @@ static struct sk_buff *receive_mergeable(struct net_device *dev,
>>   		act = do_xdp_prog(vi, rq, xdp_prog, xdp_page, offset, len);
>>   		switch (act) {
>>   		case XDP_PASS:
>> -			if (unlikely(xdp_page != page))
>> -				__free_pages(xdp_page, 0);
>> +			/* We can only create skb based on xdp_page. */
>> +			if (unlikely(xdp_page != page)) {
>> +				rcu_read_unlock();
>> +				put_page(page);
>> +				head_skb = page_to_skb(vi, rq, xdp_page,
>> +						       0, len, PAGE_SIZE);
>> +				return head_skb;
>> +			}
>>   			break;
>>   		case XDP_TX:
>>   			if (unlikely(xdp_page != page))
>>
> Great thanks. This was likely working before because of the memory
> leak fixed in 3/9.

Looks not, without this and 3/9 the code will try to get buffers and 
build skb for a new packet instead of existed buffers.

Thanks

>
> Acked-by: John Fastabend <john.r.fastabend@intel.com>

[toc] | [prev] | [next] | [standalone]

#1546880 — [PATCH net 2/9] virtio-net: correctly xmit linearized page on XDP_TX

After we linearize page, we should xmit this page instead of the page
of first buffer which may lead unexpected result. With this patch, we
can see correct packet during XDP_TX.

Cc: John Fastabend <john.r.fastabend@intel.com>
Signed-off-by: Jason Wang <jasowang@redhat.com>
---
 drivers/net/virtio_net.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
index 1067253..fe4562d 100644
--- a/drivers/net/virtio_net.c
+++ b/drivers/net/virtio_net.c
@@ -572,7 +572,7 @@ static struct sk_buff *receive_mergeable(struct net_device *dev,
 		if (unlikely(hdr->hdr.gso_type || hdr->hdr.flags))
 			goto err_xdp;
 
-		act = do_xdp_prog(vi, rq, xdp_prog, page, offset, len);
+		act = do_xdp_prog(vi, rq, xdp_prog, xdp_page, offset, len);
 		switch (act) {
 		case XDP_PASS:
 			if (unlikely(xdp_page != page))
-- 
2.7.4

[toc] | [prev] | [next] | [standalone]

#1546899 — Re: [PATCH net 2/9] virtio-net: correctly xmit linearized page on XDP_TX

On 16-12-23 06:37 AM, Jason Wang wrote:
> After we linearize page, we should xmit this page instead of the page
> of first buffer which may lead unexpected result. With this patch, we
> can see correct packet during XDP_TX.
> 
> Cc: John Fastabend <john.r.fastabend@intel.com>
> Signed-off-by: Jason Wang <jasowang@redhat.com>
> ---
>  drivers/net/virtio_net.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
> index 1067253..fe4562d 100644
> --- a/drivers/net/virtio_net.c
> +++ b/drivers/net/virtio_net.c
> @@ -572,7 +572,7 @@ static struct sk_buff *receive_mergeable(struct net_device *dev,
>  		if (unlikely(hdr->hdr.gso_type || hdr->hdr.flags))
>  			goto err_xdp;
>  
> -		act = do_xdp_prog(vi, rq, xdp_prog, page, offset, len);
> +		act = do_xdp_prog(vi, rq, xdp_prog, xdp_page, offset, len);
>  		switch (act) {
>  		case XDP_PASS:
>  			if (unlikely(xdp_page != page))
> 

Thanks clumsy merge conflict resolution between v4 and v6 versions :/

Acked-by: John Fastabend <john.r.fastabend@intel.com>

[toc] | [prev] | [next] | [standalone]

#1546881 — [PATCH net 8/9] virtio-net: remove big packet XDP codes

Now we in fact don't allow XDP for big packets, remove its codes.

Cc: John Fastabend <john.r.fastabend@intel.com>
Signed-off-by: Jason Wang <jasowang@redhat.com>
---
 drivers/net/virtio_net.c | 44 +++-----------------------------------------
 1 file changed, 3 insertions(+), 41 deletions(-)

diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
index c1f66d8..e53365a8 100644
--- a/drivers/net/virtio_net.c
+++ b/drivers/net/virtio_net.c
@@ -344,11 +344,7 @@ static void virtnet_xdp_xmit(struct virtnet_info *vi,
 	/* Free up any pending old buffers before queueing new ones. */
 	while ((xdp_sent = virtqueue_get_buf(sq->vq, &len)) != NULL) {
 		struct page *sent_page = virt_to_head_page(xdp_sent);
-
-		if (vi->mergeable_rx_bufs)
-			put_page(sent_page);
-		else
-			give_pages(rq, sent_page);
+		put_page(sent_page);
 	}
 
 	/* Zero header and leave csum up to XDP layers */
@@ -360,15 +356,8 @@ static void virtnet_xdp_xmit(struct virtnet_info *vi,
 	err = virtqueue_add_outbuf(sq->vq, sq->sg, num_sg,
 				   xdp->data, GFP_ATOMIC);
 	if (unlikely(err)) {
-		if (vi->mergeable_rx_bufs)
-			put_page(page);
-		else
-			give_pages(rq, page);
+		put_page(page);
 		return; // On error abort to avoid unnecessary kick
-	} else if (!vi->mergeable_rx_bufs) {
-		/* If not mergeable bufs must be big packets so cleanup pages */
-		give_pages(rq, (struct page *)page->private);
-		page->private = 0;
 	}
 
 	virtqueue_kick(sq->vq);
@@ -430,44 +419,17 @@ static struct sk_buff *receive_big(struct net_device *dev,
 				   void *buf,
 				   unsigned int len)
 {
-	struct bpf_prog *xdp_prog;
 	struct page *page = buf;
-	struct sk_buff *skb;
-
-	rcu_read_lock();
-	xdp_prog = rcu_dereference(rq->xdp_prog);
-	if (xdp_prog) {
-		struct virtio_net_hdr_mrg_rxbuf *hdr = buf;
-		u32 act;
-
-		if (unlikely(hdr->hdr.gso_type))
-			goto err_xdp;
-		act = do_xdp_prog(vi, rq, xdp_prog, page, 0, len);
-		switch (act) {
-		case XDP_PASS:
-			break;
-		case XDP_TX:
-			rcu_read_unlock();
-			goto xdp_xmit;
-		case XDP_DROP:
-		default:
-			goto err_xdp;
-		}
-	}
-	rcu_read_unlock();
+	struct sk_buff *skb = page_to_skb(vi, rq, page, 0, len, PAGE_SIZE);
 
-	skb = page_to_skb(vi, rq, page, 0, len, PAGE_SIZE);
 	if (unlikely(!skb))
 		goto err;
 
 	return skb;
 
-err_xdp:
-	rcu_read_unlock();
 err:
 	dev->stats.rx_dropped++;
 	give_pages(rq, page);
-xdp_xmit:
 	return NULL;
 }
 
-- 
2.7.4

[toc] | [prev] | [next] | [standalone]

#1546882 — [PATCH net 9/9] virtio-net: XDP support for small buffers

Commit f600b6905015 ("virtio_net: Add XDP support") leaves the case of
small receive buffer untouched. This will confuse the user who want to
set XDP but use small buffers. Other than forbid XDP in small buffer
mode, let's make it work. XDP then can only work at skb->data since
virtio-net create skbs during refill, this is sub optimal which could
be optimized in the future.

Cc: John Fastabend <john.r.fastabend@intel.com>
Signed-off-by: Jason Wang <jasowang@redhat.com>
---
 drivers/net/virtio_net.c | 112 ++++++++++++++++++++++++++++++++++++-----------
 1 file changed, 87 insertions(+), 25 deletions(-)

diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
index e53365a8..5deeda6 100644
--- a/drivers/net/virtio_net.c
+++ b/drivers/net/virtio_net.c
@@ -333,9 +333,9 @@ static struct sk_buff *page_to_skb(struct virtnet_info *vi,
 static void virtnet_xdp_xmit(struct virtnet_info *vi,
 			     struct receive_queue *rq,
 			     struct send_queue *sq,
-			     struct xdp_buff *xdp)
+			     struct xdp_buff *xdp,
+			     void *data)
 {
-	struct page *page = virt_to_head_page(xdp->data);
 	struct virtio_net_hdr_mrg_rxbuf *hdr;
 	unsigned int num_sg, len;
 	void *xdp_sent;
@@ -343,20 +343,45 @@ static void virtnet_xdp_xmit(struct virtnet_info *vi,
 
 	/* Free up any pending old buffers before queueing new ones. */
 	while ((xdp_sent = virtqueue_get_buf(sq->vq, &len)) != NULL) {
-		struct page *sent_page = virt_to_head_page(xdp_sent);
-		put_page(sent_page);
+		if (vi->mergeable_rx_bufs) {
+			struct page *sent_page = virt_to_head_page(xdp_sent);
+
+			put_page(sent_page);
+		} else { /* small buffer */
+			struct sk_buff *skb = xdp_sent;
+
+			kfree_skb(skb);
+		}
 	}
 
-	/* Zero header and leave csum up to XDP layers */
-	hdr = xdp->data;
-	memset(hdr, 0, vi->hdr_len);
+	if (vi->mergeable_rx_bufs) {
+		/* Zero header and leave csum up to XDP layers */
+		hdr = xdp->data;
+		memset(hdr, 0, vi->hdr_len);
+
+		num_sg = 1;
+		sg_init_one(sq->sg, xdp->data, xdp->data_end - xdp->data);
+	} else { /* small buffer */
+		struct sk_buff *skb = data;
 
-	num_sg = 1;
-	sg_init_one(sq->sg, xdp->data, xdp->data_end - xdp->data);
+		/* Zero header and leave csum up to XDP layers */
+		hdr = skb_vnet_hdr(skb);
+		memset(hdr, 0, vi->hdr_len);
+
+		num_sg = 2;
+		sg_init_table(sq->sg, 2);
+		sg_set_buf(sq->sg, hdr, vi->hdr_len);
+		skb_to_sgvec(skb, sq->sg + 1, 0, skb->len);
+	}
 	err = virtqueue_add_outbuf(sq->vq, sq->sg, num_sg,
-				   xdp->data, GFP_ATOMIC);
+				   data, GFP_ATOMIC);
 	if (unlikely(err)) {
-		put_page(page);
+		if (vi->mergeable_rx_bufs) {
+			struct page *page = virt_to_head_page(xdp->data);
+
+			put_page(page);
+		} else /* small buffer */
+			kfree_skb(data);
 		return; // On error abort to avoid unnecessary kick
 	}
 
@@ -366,23 +391,26 @@ static void virtnet_xdp_xmit(struct virtnet_info *vi,
 static u32 do_xdp_prog(struct virtnet_info *vi,
 		       struct receive_queue *rq,
 		       struct bpf_prog *xdp_prog,
-		       struct page *page, int offset, int len)
+		       void *data, int len)
 {
 	int hdr_padded_len;
 	struct xdp_buff xdp;
+	void *buf;
 	unsigned int qp;
 	u32 act;
-	u8 *buf;
-
-	buf = page_address(page) + offset;
 
-	if (vi->mergeable_rx_bufs)
+	if (vi->mergeable_rx_bufs) {
 		hdr_padded_len = sizeof(struct virtio_net_hdr_mrg_rxbuf);
-	else
-		hdr_padded_len = sizeof(struct padded_vnet_hdr);
+		xdp.data = data + hdr_padded_len;
+		xdp.data_end = xdp.data + (len - vi->hdr_len);
+		buf = data;
+	} else { /* small buffers */
+		struct sk_buff *skb = data;
 
-	xdp.data = buf + hdr_padded_len;
-	xdp.data_end = xdp.data + (len - vi->hdr_len);
+		xdp.data = skb->data;
+		xdp.data_end = xdp.data + len;
+		buf = skb->data;
+	}
 
 	act = bpf_prog_run_xdp(xdp_prog, &xdp);
 	switch (act) {
@@ -392,8 +420,8 @@ static u32 do_xdp_prog(struct virtnet_info *vi,
 		qp = vi->curr_queue_pairs -
 			vi->xdp_queue_pairs +
 			smp_processor_id();
-		xdp.data = buf + (vi->mergeable_rx_bufs ? 0 : 4);
-		virtnet_xdp_xmit(vi, rq, &vi->sq[qp], &xdp);
+		xdp.data = buf;
+		virtnet_xdp_xmit(vi, rq, &vi->sq[qp], &xdp, data);
 		return XDP_TX;
 	default:
 		bpf_warn_invalid_xdp_action(act);
@@ -403,14 +431,47 @@ static u32 do_xdp_prog(struct virtnet_info *vi,
 	}
 }
 
-static struct sk_buff *receive_small(struct virtnet_info *vi, void *buf, unsigned int len)
+static struct sk_buff *receive_small(struct net_device *dev,
+				     struct virtnet_info *vi,
+				     struct receive_queue *rq,
+				     void *buf, unsigned int len)
 {
 	struct sk_buff * skb = buf;
+	struct bpf_prog *xdp_prog;
 
 	len -= vi->hdr_len;
 	skb_trim(skb, len);
 
+	rcu_read_lock();
+	xdp_prog = rcu_dereference(rq->xdp_prog);
+	if (xdp_prog) {
+		struct virtio_net_hdr_mrg_rxbuf *hdr = buf;
+		u32 act;
+
+		if (unlikely(hdr->hdr.gso_type || hdr->hdr.flags))
+			goto err_xdp;
+		act = do_xdp_prog(vi, rq, xdp_prog, skb, len);
+		switch (act) {
+		case XDP_PASS:
+			break;
+		case XDP_TX:
+			rcu_read_unlock();
+			goto xdp_xmit;
+		case XDP_DROP:
+		default:
+			goto err_xdp;
+		}
+	}
+	rcu_read_unlock();
+
 	return skb;
+
+err_xdp:
+	rcu_read_unlock();
+	dev->stats.rx_dropped++;
+	kfree_skb(skb);
+xdp_xmit:
+	return NULL;
 }
 
 static struct sk_buff *receive_big(struct net_device *dev,
@@ -537,7 +598,8 @@ static struct sk_buff *receive_mergeable(struct net_device *dev,
 		if (unlikely(hdr->hdr.gso_type))
 			goto err_xdp;
 
-		act = do_xdp_prog(vi, rq, xdp_prog, xdp_page, offset, len);
+		act = do_xdp_prog(vi, rq, xdp_prog,
+				  page_address(xdp_page) + offset, len);
 		switch (act) {
 		case XDP_PASS:
 			/* We can only create skb based on xdp_page. */
@@ -672,7 +734,7 @@ static void receive_buf(struct virtnet_info *vi, struct receive_queue *rq,
 	else if (vi->big_packets)
 		skb = receive_big(dev, vi, rq, buf, len);
 	else
-		skb = receive_small(vi, buf, len);
+		skb = receive_small(dev, vi, rq, buf, len);
 
 	if (unlikely(!skb))
 		return;
-- 
2.7.4

[toc] | [prev] | [next] | [standalone]

#1546919 — Re: [PATCH net 9/9] virtio-net: XDP support for small buffers

On 16-12-23 06:37 AM, Jason Wang wrote:
> Commit f600b6905015 ("virtio_net: Add XDP support") leaves the case of
> small receive buffer untouched. This will confuse the user who want to
> set XDP but use small buffers. Other than forbid XDP in small buffer
> mode, let's make it work. XDP then can only work at skb->data since
> virtio-net create skbs during refill, this is sub optimal which could
> be optimized in the future.
> 
> Cc: John Fastabend <john.r.fastabend@intel.com>
> Signed-off-by: Jason Wang <jasowang@redhat.com>
> ---


Looks good to me thanks!

Acked-by: John Fastabend <john.r.fastabend@intel.com>

[toc] | [prev] | [next] | [standalone]

#1549428 — Re: [PATCH net 9/9] virtio-net: XDP support for small buffers

On 16-12-23 06:37 AM, Jason Wang wrote:
> Commit f600b6905015 ("virtio_net: Add XDP support") leaves the case of
> small receive buffer untouched. This will confuse the user who want to
> set XDP but use small buffers. Other than forbid XDP in small buffer
> mode, let's make it work. XDP then can only work at skb->data since
> virtio-net create skbs during refill, this is sub optimal which could
> be optimized in the future.
> 
> Cc: John Fastabend <john.r.fastabend@intel.com>
> Signed-off-by: Jason Wang <jasowang@redhat.com>
> ---
>  drivers/net/virtio_net.c | 112 ++++++++++++++++++++++++++++++++++++-----------
>  1 file changed, 87 insertions(+), 25 deletions(-)
> 

Hi Jason,

I was doing some more testing on this what do you think about doing this
so that free_unused_bufs() handles the buffer free with dev_kfree_skb()
instead of put_page in small receive mode. Seems more correct to me.


diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
index 783e842..27ff76c 100644
--- a/drivers/net/virtio_net.c
+++ b/drivers/net/virtio_net.c
@@ -1898,6 +1898,10 @@ static void free_receive_page_frags(struct virtnet_info *vi)

 static bool is_xdp_queue(struct virtnet_info *vi, int q)
 {
+       /* For small receive mode always use kfree_skb variants */
+       if (!vi->mergeable_rx_bufs)
+               return false;
+
        if (q < (vi->curr_queue_pairs - vi->xdp_queue_pairs))
                return false;
        else if (q < vi->curr_queue_pairs)


patch is untested just spotted doing code review.

Thanks,
John

[toc] | [prev] | [next] | [standalone]

#1549551 — Re: [PATCH net 9/9] virtio-net: XDP support for small buffers

On 2017年01月03日 06:43, John Fastabend wrote:
> On 16-12-23 06:37 AM, Jason Wang wrote:
>> Commit f600b6905015 ("virtio_net: Add XDP support") leaves the case of
>> small receive buffer untouched. This will confuse the user who want to
>> set XDP but use small buffers. Other than forbid XDP in small buffer
>> mode, let's make it work. XDP then can only work at skb->data since
>> virtio-net create skbs during refill, this is sub optimal which could
>> be optimized in the future.
>>
>> Cc: John Fastabend <john.r.fastabend@intel.com>
>> Signed-off-by: Jason Wang <jasowang@redhat.com>
>> ---
>>   drivers/net/virtio_net.c | 112 ++++++++++++++++++++++++++++++++++++-----------
>>   1 file changed, 87 insertions(+), 25 deletions(-)
>>
> Hi Jason,
>
> I was doing some more testing on this what do you think about doing this
> so that free_unused_bufs() handles the buffer free with dev_kfree_skb()
> instead of put_page in small receive mode. Seems more correct to me.
>
>
> diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
> index 783e842..27ff76c 100644
> --- a/drivers/net/virtio_net.c
> +++ b/drivers/net/virtio_net.c
> @@ -1898,6 +1898,10 @@ static void free_receive_page_frags(struct virtnet_info *vi)
>
>   static bool is_xdp_queue(struct virtnet_info *vi, int q)
>   {
> +       /* For small receive mode always use kfree_skb variants */
> +       if (!vi->mergeable_rx_bufs)
> +               return false;
> +
>          if (q < (vi->curr_queue_pairs - vi->xdp_queue_pairs))
>                  return false;
>          else if (q < vi->curr_queue_pairs)
>
>
> patch is untested just spotted doing code review.
>
> Thanks,
> John

We probably need a better name for this function.

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

[toc] | [prev] | [next] | [standalone]

#1549945 — Re: [PATCH net 9/9] virtio-net: XDP support for small buffers

On 17-01-02 10:16 PM, Jason Wang wrote:
> 
> 
> On 2017年01月03日 06:43, John Fastabend wrote:
>> On 16-12-23 06:37 AM, Jason Wang wrote:
>>> Commit f600b6905015 ("virtio_net: Add XDP support") leaves the case of
>>> small receive buffer untouched. This will confuse the user who want to
>>> set XDP but use small buffers. Other than forbid XDP in small buffer
>>> mode, let's make it work. XDP then can only work at skb->data since
>>> virtio-net create skbs during refill, this is sub optimal which could
>>> be optimized in the future.
>>>
>>> Cc: John Fastabend <john.r.fastabend@intel.com>
>>> Signed-off-by: Jason Wang <jasowang@redhat.com>
>>> ---
>>>   drivers/net/virtio_net.c | 112 ++++++++++++++++++++++++++++++++++++-----------
>>>   1 file changed, 87 insertions(+), 25 deletions(-)
>>>
>> Hi Jason,
>>
>> I was doing some more testing on this what do you think about doing this
>> so that free_unused_bufs() handles the buffer free with dev_kfree_skb()
>> instead of put_page in small receive mode. Seems more correct to me.
>>
>>
>> diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
>> index 783e842..27ff76c 100644
>> --- a/drivers/net/virtio_net.c
>> +++ b/drivers/net/virtio_net.c
>> @@ -1898,6 +1898,10 @@ static void free_receive_page_frags(struct virtnet_info
>> *vi)
>>
>>   static bool is_xdp_queue(struct virtnet_info *vi, int q)
>>   {
>> +       /* For small receive mode always use kfree_skb variants */
>> +       if (!vi->mergeable_rx_bufs)
>> +               return false;
>> +
>>          if (q < (vi->curr_queue_pairs - vi->xdp_queue_pairs))
>>                  return false;
>>          else if (q < vi->curr_queue_pairs)
>>
>>
>> patch is untested just spotted doing code review.
>>
>> Thanks,
>> John
> 
> We probably need a better name for this function.
> 
> Acked-by: Jason Wang <jasowang@redhat.com>
> 

How about is_xdp_raw_buffer_queue()?

I'll submit a proper patch today.

[toc] | [prev] | [next] | [standalone]

#1546884 — [PATCH net 5/9] virtio-net: unbreak csumed packets for XDP_PASS

We drop csumed packet when do XDP for packets. This breaks
XDP_PASS when GUEST_CSUM is supported. Fix this by allowing csum flag
to be set. With this patch, simple TCP works for XDP_PASS.

Cc: John Fastabend <john.r.fastabend@intel.com>
Signed-off-by: Jason Wang <jasowang@redhat.com>
---
 drivers/net/virtio_net.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
index 470293e..0778dc8 100644
--- a/drivers/net/virtio_net.c
+++ b/drivers/net/virtio_net.c
@@ -440,7 +440,7 @@ static struct sk_buff *receive_big(struct net_device *dev,
 		struct virtio_net_hdr_mrg_rxbuf *hdr = buf;
 		u32 act;
 
-		if (unlikely(hdr->hdr.gso_type || hdr->hdr.flags))
+		if (unlikely(hdr->hdr.gso_type))
 			goto err_xdp;
 		act = do_xdp_prog(vi, rq, xdp_prog, page, 0, len);
 		switch (act) {
@@ -572,7 +572,7 @@ static struct sk_buff *receive_mergeable(struct net_device *dev,
 		 * the receive path after XDP is loaded. In practice I
 		 * was not able to create this condition.
 		 */
-		if (unlikely(hdr->hdr.gso_type || hdr->hdr.flags))
+		if (unlikely(hdr->hdr.gso_type))
 			goto err_xdp;
 
 		act = do_xdp_prog(vi, rq, xdp_prog, xdp_page, offset, len);
-- 
2.7.4

[toc] | [prev] | [next] | [standalone]

#1546904 — Re: [PATCH net 5/9] virtio-net: unbreak csumed packets for XDP_PASS

On 16-12-23 06:37 AM, Jason Wang wrote:
> We drop csumed packet when do XDP for packets. This breaks
> XDP_PASS when GUEST_CSUM is supported. Fix this by allowing csum flag
> to be set. With this patch, simple TCP works for XDP_PASS.
> 
> Cc: John Fastabend <john.r.fastabend@intel.com>
> Signed-off-by: Jason Wang <jasowang@redhat.com>
> ---
>  drivers/net/virtio_net.c | 4 ++--
>  1 file changed, 2 insertions(+), 2 deletions(-)
> 
> diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
> index 470293e..0778dc8 100644
> --- a/drivers/net/virtio_net.c
> +++ b/drivers/net/virtio_net.c
> @@ -440,7 +440,7 @@ static struct sk_buff *receive_big(struct net_device *dev,
>  		struct virtio_net_hdr_mrg_rxbuf *hdr = buf;
>  		u32 act;
>  
> -		if (unlikely(hdr->hdr.gso_type || hdr->hdr.flags))
> +		if (unlikely(hdr->hdr.gso_type))
>  			goto err_xdp;
>  		act = do_xdp_prog(vi, rq, xdp_prog, page, 0, len);
>  		switch (act) {
> @@ -572,7 +572,7 @@ static struct sk_buff *receive_mergeable(struct net_device *dev,
>  		 * the receive path after XDP is loaded. In practice I
>  		 * was not able to create this condition.
>  		 */
> -		if (unlikely(hdr->hdr.gso_type || hdr->hdr.flags))
> +		if (unlikely(hdr->hdr.gso_type))
>  			goto err_xdp;
>  
>  		act = do_xdp_prog(vi, rq, xdp_prog, xdp_page, offset, len);
> 

Acked-by: John Fastabend <john.r.fastabend@intel.com>

[toc] | [prev] | [next] | [standalone]

#1546923

On 16-12-23 06:37 AM, Jason Wang wrote:
> Merry Xmas and a Happy New year to all:
> 
> This series tries to fixes several issues for virtio-net XDP which
> could be categorized into several parts:
> 
> - fix several issues during XDP linearizing
> - allow csumed packet to work for XDP_PASS
> - make EWMA rxbuf size estimation works for XDP
> - forbid XDP when GUEST_UFO is support
> - remove big packet XDP support
> - add XDP support or small buffer
> 
> Please see individual patches for details.
> 
> Thanks
> 
> Jason Wang (9):
>   virtio-net: remove the warning before XDP linearizing
>   virtio-net: correctly xmit linearized page on XDP_TX
>   virtio-net: fix page miscount during XDP linearizing
>   virtio-net: correctly handle XDP_PASS for linearized packets
>   virtio-net: unbreak csumed packets for XDP_PASS
>   virtio-net: make rx buf size estimation works for XDP
>   virtio-net: forbid XDP when VIRTIO_NET_F_GUEST_UFO is support
>   virtio-net: remove big packet XDP codes
>   virtio-net: XDP support for small buffers
> 
>  drivers/net/virtio_net.c | 172 ++++++++++++++++++++++++++++-------------------
>  1 file changed, 102 insertions(+), 70 deletions(-)
> 

Thanks a lot Jason. The last piece that is needed is support to
complete XDP support is to get the adjust_head part correct. I'll
send out a patch in a bit but will need to merge it on top of this
set.

.John

[toc] | [prev] | [next] | [standalone]

#1547238

On 2016年12月24日 01:10, John Fastabend wrote:
> On 16-12-23 06:37 AM, Jason Wang wrote:
>> Merry Xmas and a Happy New year to all:
>>
>> This series tries to fixes several issues for virtio-net XDP which
>> could be categorized into several parts:
>>
>> - fix several issues during XDP linearizing
>> - allow csumed packet to work for XDP_PASS
>> - make EWMA rxbuf size estimation works for XDP
>> - forbid XDP when GUEST_UFO is support
>> - remove big packet XDP support
>> - add XDP support or small buffer
>>
>> Please see individual patches for details.
>>
>> Thanks
>>
>> Jason Wang (9):
>>    virtio-net: remove the warning before XDP linearizing
>>    virtio-net: correctly xmit linearized page on XDP_TX
>>    virtio-net: fix page miscount during XDP linearizing
>>    virtio-net: correctly handle XDP_PASS for linearized packets
>>    virtio-net: unbreak csumed packets for XDP_PASS
>>    virtio-net: make rx buf size estimation works for XDP
>>    virtio-net: forbid XDP when VIRTIO_NET_F_GUEST_UFO is support
>>    virtio-net: remove big packet XDP codes
>>    virtio-net: XDP support for small buffers
>>
>>   drivers/net/virtio_net.c | 172 ++++++++++++++++++++++++++++-------------------
>>   1 file changed, 102 insertions(+), 70 deletions(-)
>>
> Thanks a lot Jason. The last piece that is needed is support to
> complete XDP support is to get the adjust_head part correct. I'll
> send out a patch in a bit but will need to merge it on top of this
> set.
>
> .John

Yes, glad to see the your patch.

Thanks.

[toc] | [prev] | [next] | [standalone]

#1546944

From: Jason Wang <jasowang@redhat.com>
Date: Fri, 23 Dec 2016 22:37:23 +0800

> Merry Xmas and a Happy New year to all:
> 
> This series tries to fixes several issues for virtio-net XDP which
> could be categorized into several parts:
> 
> - fix several issues during XDP linearizing
> - allow csumed packet to work for XDP_PASS
> - make EWMA rxbuf size estimation works for XDP
> - forbid XDP when GUEST_UFO is support
> - remove big packet XDP support
> - add XDP support or small buffer
> 
> Please see individual patches for details.

Series applied, thanks Jason.

[toc] | [prev] | [standalone]

Back to top | Article view | linux.kernel

csiph-web