Linux Virtualization - Oct 2014 - [PATCH RFC 3/4] virtio_net: stricter short buffer length checks

Too many places poke at [rs]q->vq->vdev->priv just to get
the the vi structure.  Let's just pass the pointer around: seems
cleaner, and might even be faster.

Signed-off-by: Michael S. Tsirkin <mst at redhat.com>
---
 drivers/net/virtio_net.c | 36 +++++++++++++++++++-----------------
 1 file changed, 19 insertions(+), 17 deletions(-)

diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
index 57cbc7d..36f3dfc 100644
--- a/drivers/net/virtio_net.c
+++ b/drivers/net/virtio_net.c
@@ -241,11 +241,11 @@ static unsigned long mergeable_buf_to_ctx(void *buf,
unsigned int truesize)
 }
 
 /* Called from bottom half context */
-static struct sk_buff *page_to_skb(struct receive_queue *rq,
+static struct sk_buff *page_to_skb(struct virtnet_info *vi,
+				   struct receive_queue *rq,
 				   struct page *page, unsigned int offset,
 				   unsigned int len, unsigned int truesize)
 {
-	struct virtnet_info *vi = rq->vq->vdev->priv;
 	struct sk_buff *skb;
 	struct skb_vnet_hdr *hdr;
 	unsigned int copy, hdr_len, hdr_padded_len;
@@ -328,12 +328,13 @@ static struct sk_buff *receive_small(void *buf, unsigned
int len)
 }
 
 static struct sk_buff *receive_big(struct net_device *dev,
+				   struct virtnet_info *vi,
 				   struct receive_queue *rq,
 				   void *buf,
 				   unsigned int len)
 {
 	struct page *page = buf;
-	struct sk_buff *skb = page_to_skb(rq, page, 0, len, PAGE_SIZE);
+	struct sk_buff *skb = page_to_skb(vi, rq, page, 0, len, PAGE_SIZE);
 
 	if (unlikely(!skb))
 		goto err;
@@ -359,7 +360,8 @@ static struct sk_buff *receive_mergeable(struct net_device
*dev,
 	int offset = buf - page_address(page);
 	unsigned int truesize = max(len, mergeable_ctx_to_buf_truesize(ctx));
 
-	struct sk_buff *head_skb = page_to_skb(rq, page, offset, len, truesize);
+	struct sk_buff *head_skb = page_to_skb(vi, rq, page, offset, len,
+					       truesize);
 	struct sk_buff *curr_skb = head_skb;
 
 	if (unlikely(!curr_skb))
@@ -433,9 +435,9 @@ err_buf:
 	return NULL;
 }
 
-static void receive_buf(struct receive_queue *rq, void *buf, unsigned int len)
+static void receive_buf(struct virtnet_info *vi, struct receive_queue *rq,
+			void *buf, unsigned int len)
 {
-	struct virtnet_info *vi = rq->vq->vdev->priv;
 	struct net_device *dev = vi->dev;
 	struct virtnet_stats *stats = this_cpu_ptr(vi->stats);
 	struct sk_buff *skb;
@@ -459,9 +461,9 @@ static void receive_buf(struct receive_queue *rq, void *buf,
unsigned int len)
 	if (vi->mergeable_rx_bufs)
 		skb = receive_mergeable(dev, vi, rq, (unsigned long)buf, len);
 	else if (vi->big_packets)
-		skb = receive_big(dev, rq, buf, len);
+		skb = receive_big(dev, vi, rq, buf, len);
 	else
-		skb = receive_small(buf, len);
+		skb = receive_small(vi, buf, len);
 
 	if (unlikely(!skb))
 		return;
@@ -530,9 +532,9 @@ frame_err:
 	dev_kfree_skb(skb);
 }
 
-static int add_recvbuf_small(struct receive_queue *rq, gfp_t gfp)
+static int add_recvbuf_small(struct virtnet_info *vi, struct receive_queue *rq,
+			     gfp_t gfp)
 {
-	struct virtnet_info *vi = rq->vq->vdev->priv;
 	struct sk_buff *skb;
 	struct skb_vnet_hdr *hdr;
 	int err;
@@ -655,9 +657,9 @@ static int add_recvbuf_mergeable(struct receive_queue *rq,
gfp_t gfp)
  * before we're receiving packets, or from refill_work which is
  * careful to disable receiving (using napi_disable).
  */
-static bool try_fill_recv(struct receive_queue *rq, gfp_t gfp)
+static bool try_fill_recv(struct virtnet_info *vi, struct receive_queue *rq,
+			  gfp_t gfp)
 {
-	struct virtnet_info *vi = rq->vq->vdev->priv;
 	int err;
 	bool oom;
 
@@ -668,7 +670,7 @@ static bool try_fill_recv(struct receive_queue *rq, gfp_t
gfp)
 		else if (vi->big_packets)
 			err = add_recvbuf_big(rq, gfp);
 		else
-			err = add_recvbuf_small(rq, gfp);
+			err = add_recvbuf_small(vi, rq, gfp);
 
 		oom = err == -ENOMEM;
 		if (err)
@@ -717,7 +719,7 @@ static void refill_work(struct work_struct *work)
 		struct receive_queue *rq = &vi->rq[i];
 
 		napi_disable(&rq->napi);
-		still_empty = !try_fill_recv(rq, GFP_KERNEL);
+		still_empty = !try_fill_recv(vi, rq, GFP_KERNEL);
 		virtnet_napi_enable(rq);
 
 		/* In theory, this can happen: if we don't get any buffers in
@@ -736,12 +738,12 @@ static int virtnet_receive(struct receive_queue *rq, int
budget)
 
 	while (received < budget &&
 	       (buf = virtqueue_get_buf(rq->vq, &len)) != NULL) {
-		receive_buf(rq, buf, len);
+		receive_buf(vi, rq, buf, len);
 		received++;
 	}
 
 	if (rq->vq->num_free > virtqueue_get_vring_size(rq->vq) / 2) {
-		if (!try_fill_recv(rq, GFP_ATOMIC))
+		if (!try_fill_recv(vi, rq, GFP_ATOMIC))
 			schedule_delayed_work(&vi->refill, 0);
 	}
 
@@ -817,7 +819,7 @@ static int virtnet_open(struct net_device *dev)
 	for (i = 0; i < vi->max_queue_pairs; i++) {
 		if (i < vi->curr_queue_pairs)
 			/* Make sure we have some buffers: if oom use wq. */
-			if (!try_fill_recv(&vi->rq[i], GFP_KERNEL))
+			if (!try_fill_recv(vi, &vi->rq[i], GFP_KERNEL))
 				schedule_delayed_work(&vi->refill, 0);
 		virtnet_napi_enable(&vi->rq[i]);
 	}
-- 
MST

virtio 1.0 doesn't use virtio_net_hdr anymore, and in fact, it's not
really useful since virtio_net_hdr_mrg_rxbuf includes that as the first
field anyway.

Let's drop it, precalculate header len and store within vi instead.

This way we can also remove struct skb_vnet_hdr.

Signed-off-by: Michael S. Tsirkin <mst at redhat.com>
---
 drivers/net/virtio_net.c | 88 ++++++++++++++++++++++--------------------------
 1 file changed, 40 insertions(+), 48 deletions(-)

diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
index 36f3dfc..a795a23 100644
--- a/drivers/net/virtio_net.c
+++ b/drivers/net/virtio_net.c
@@ -123,6 +123,9 @@ struct virtnet_info {
 	/* Host can handle any s/g split between our header and packet data */
 	bool any_header_sg;
 
+	/* Packet virtio header size */
+	u8 hdr_len;
+
 	/* Active statistics */
 	struct virtnet_stats __percpu *stats;
 
@@ -139,21 +142,14 @@ struct virtnet_info {
 	struct notifier_block nb;
 };
 
-struct skb_vnet_hdr {
-	union {
-		struct virtio_net_hdr hdr;
-		struct virtio_net_hdr_mrg_rxbuf mhdr;
-	};
-};
-
 struct padded_vnet_hdr {
-	struct virtio_net_hdr hdr;
+	struct virtio_net_hdr_mrg_rxbuf hdr;
 	/*
-	 * virtio_net_hdr should be in a separated sg buffer because of a
-	 * QEMU bug, and data sg buffer shares same page with this header sg.
-	 * This padding makes next sg 16 byte aligned after virtio_net_hdr.
+	 * hdr is in a separate sg buffer, and data sg buffer shares same page
+	 * with this header sg. This padding makes next sg 16 byte aligned
+	 * after the header.
 	 */
-	char padding[6];
+	char padding[4];
 };
 
 /* Converting between virtqueue no. and kernel tx/rx queue no.
@@ -179,9 +175,9 @@ static int rxq2vq(int rxq)
 	return rxq * 2;
 }
 
-static inline struct skb_vnet_hdr *skb_vnet_hdr(struct sk_buff *skb)
+static inline struct virtio_net_hdr_mrg_rxbuf *skb_vnet_hdr(struct sk_buff
*skb)
 {
-	return (struct skb_vnet_hdr *)skb->cb;
+	return (struct virtio_net_hdr_mrg_rxbuf *)skb->cb;
 }
 
 /*
@@ -247,7 +243,7 @@ static struct sk_buff *page_to_skb(struct virtnet_info *vi,
 				   unsigned int len, unsigned int truesize)
 {
 	struct sk_buff *skb;
-	struct skb_vnet_hdr *hdr;
+	struct virtio_net_hdr_mrg_rxbuf *hdr;
 	unsigned int copy, hdr_len, hdr_padded_len;
 	char *p;
 
@@ -260,13 +256,11 @@ static struct sk_buff *page_to_skb(struct virtnet_info
*vi,
 
 	hdr = skb_vnet_hdr(skb);
 
-	if (vi->mergeable_rx_bufs) {
-		hdr_len = sizeof hdr->mhdr;
-		hdr_padded_len = sizeof hdr->mhdr;
-	} else {
-		hdr_len = sizeof hdr->hdr;
+	hdr_len = vi->hdr_len;
+	if (vi->mergeable_rx_bufs)
+		hdr_padded_len = sizeof *hdr;
+	else
 		hdr_padded_len = sizeof(struct padded_vnet_hdr);
-	}
 
 	memcpy(hdr, p, hdr_len);
 
@@ -317,11 +311,11 @@ static struct sk_buff *page_to_skb(struct virtnet_info
*vi,
 	return skb;
 }
 
-static struct sk_buff *receive_small(void *buf, unsigned int len)
+static struct sk_buff *receive_small(struct virtnet_info *vi, void *buf,
unsigned int len)
 {
 	struct sk_buff * skb = buf;
 
-	len -= sizeof(struct virtio_net_hdr);
+	len -= vi->hdr_len;
 	skb_trim(skb, len);
 
 	return skb;
@@ -354,8 +348,8 @@ static struct sk_buff *receive_mergeable(struct net_device
*dev,
 					 unsigned int len)
 {
 	void *buf = mergeable_ctx_to_buf_address(ctx);
-	struct skb_vnet_hdr *hdr = buf;
-	u16 num_buf = virtio16_to_cpu(rq->vq->vdev, hdr->mhdr.num_buffers);
+	struct virtio_net_hdr_mrg_rxbuf *hdr = buf;
+	u16 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 truesize = max(len, mergeable_ctx_to_buf_truesize(ctx));
@@ -373,8 +367,8 @@ static struct sk_buff *receive_mergeable(struct net_device
*dev,
 		if (unlikely(!ctx)) {
 			pr_debug("%s: rx error: %d buffers out of %d missing\n",
 				 dev->name, num_buf,
-				 virtio16_to_cpu(rq->vq->vdev,
-						 hdr->mhdr.num_buffers));
+				 virtio16_to_cpu(vi->vdev,
+						 hdr->num_buffers));
 			dev->stats.rx_length_errors++;
 			goto err_buf;
 		}
@@ -441,7 +435,7 @@ static void receive_buf(struct virtnet_info *vi, struct
receive_queue *rq,
 	struct net_device *dev = vi->dev;
 	struct virtnet_stats *stats = this_cpu_ptr(vi->stats);
 	struct sk_buff *skb;
-	struct skb_vnet_hdr *hdr;
+	struct virtio_net_hdr_mrg_rxbuf *hdr;
 
 	if (unlikely(len < sizeof(struct virtio_net_hdr) + ETH_HLEN)) {
 		pr_debug("%s: short packet %i\n", dev->name, len);
@@ -536,7 +530,7 @@ static int add_recvbuf_small(struct virtnet_info *vi, struct
receive_queue *rq,
 			     gfp_t gfp)
 {
 	struct sk_buff *skb;
-	struct skb_vnet_hdr *hdr;
+	struct virtio_net_hdr_mrg_rxbuf *hdr;
 	int err;
 
 	skb = __netdev_alloc_skb_ip_align(vi->dev, GOOD_PACKET_LEN, gfp);
@@ -547,7 +541,7 @@ static int add_recvbuf_small(struct virtnet_info *vi, struct
receive_queue *rq,
 
 	hdr = skb_vnet_hdr(skb);
 	sg_init_table(rq->sg, MAX_SKB_FRAGS + 2);
-	sg_set_buf(rq->sg, &hdr->hdr, sizeof hdr->hdr);
+	sg_set_buf(rq->sg, hdr, vi->hdr_len);
 	skb_to_sgvec(skb, rq->sg + 1, 0, skb->len);
 
 	err = virtqueue_add_inbuf(rq->vq, rq->sg, 2, skb, gfp);
@@ -557,7 +551,8 @@ static int add_recvbuf_small(struct virtnet_info *vi, struct
receive_queue *rq,
 	return err;
 }
 
-static int add_recvbuf_big(struct receive_queue *rq, gfp_t gfp)
+static int add_recvbuf_big(struct virtnet_info *vi, struct receive_queue *rq,
+			   gfp_t gfp)
 {
 	struct page *first, *list = NULL;
 	char *p;
@@ -588,8 +583,8 @@ static int add_recvbuf_big(struct receive_queue *rq, gfp_t
gfp)
 	p = page_address(first);
 
 	/* rq->sg[0], rq->sg[1] share the same page */
-	/* a separated rq->sg[0] for virtio_net_hdr only due to QEMU bug */
-	sg_set_buf(&rq->sg[0], p, sizeof(struct virtio_net_hdr));
+	/* a separated rq->sg[0] for header - required in case !any_header_sg */
+	sg_set_buf(&rq->sg[0], p, vi->hdr_len);
 
 	/* rq->sg[1] for data packet, from offset */
 	offset = sizeof(struct padded_vnet_hdr);
@@ -668,7 +663,7 @@ static bool try_fill_recv(struct virtnet_info *vi, struct
receive_queue *rq,
 		if (vi->mergeable_rx_bufs)
 			err = add_recvbuf_mergeable(rq, gfp);
 		else if (vi->big_packets)
-			err = add_recvbuf_big(rq, gfp);
+			err = add_recvbuf_big(vi, rq, gfp);
 		else
 			err = add_recvbuf_small(vi, rq, gfp);
 
@@ -848,18 +843,14 @@ static void free_old_xmit_skbs(struct send_queue *sq)
 
 static int xmit_skb(struct send_queue *sq, struct sk_buff *skb)
 {
-	struct skb_vnet_hdr *hdr;
+	struct virtio_net_hdr_mrg_rxbuf *hdr;
 	const unsigned char *dest = ((struct ethhdr *)skb->data)->h_dest;
 	struct virtnet_info *vi = sq->vq->vdev->priv;
 	unsigned num_sg;
-	unsigned hdr_len;
+	unsigned hdr_len = vi->hdr_len;
 	bool can_push;
 
 	pr_debug("%s: xmit %p %pM\n", vi->dev->name, skb, dest);
-	if (vi->mergeable_rx_bufs)
-		hdr_len = sizeof hdr->mhdr;
-	else
-		hdr_len = sizeof hdr->hdr;
 
 	can_push = vi->any_header_sg &&
 		!((unsigned long)skb->data & (__alignof__(*hdr) - 1)) &&
@@ -867,7 +858,7 @@ static int xmit_skb(struct send_queue *sq, struct sk_buff
*skb)
 	/* Even if we can, don't push here yet as this would skew
 	 * csum_start offset below. */
 	if (can_push)
-		hdr = (struct skb_vnet_hdr *)(skb->data - hdr_len);
+		hdr = (struct virtio_net_hdr_mrg_rxbuf *)(skb->data - hdr_len);
 	else
 		hdr = skb_vnet_hdr(skb);
 
@@ -902,7 +893,7 @@ static int xmit_skb(struct send_queue *sq, struct sk_buff
*skb)
 	}
 
 	if (vi->mergeable_rx_bufs)
-		hdr->mhdr.num_buffers = 0;
+		hdr->num_buffers = 0;
 
 	sg_init_table(sq->sg, MAX_SKB_FRAGS + 2);
 	if (can_push) {
@@ -1773,18 +1764,19 @@ static int virtnet_probe(struct virtio_device *vdev)
 	if (virtio_has_feature(vdev, VIRTIO_NET_F_MRG_RXBUF))
 		vi->mergeable_rx_bufs = true;
 
+	if (virtio_has_feature(vdev, VIRTIO_NET_F_MRG_RXBUF))
+		vi->hdr_len = sizeof(struct virtio_net_hdr_mrg_rxbuf);
+	else
+		vi->hdr_len = sizeof(struct virtio_net_hdr);
+
 	if (virtio_has_feature(vdev, VIRTIO_F_ANY_LAYOUT))
 		vi->any_header_sg = true;
 
 	if (virtio_has_feature(vdev, VIRTIO_NET_F_CTRL_VQ))
 		vi->has_cvq = true;
 
-	if (vi->any_header_sg) {
-		if (vi->mergeable_rx_bufs)
-			dev->needed_headroom = sizeof(struct virtio_net_hdr_mrg_rxbuf);
-		else
-			dev->needed_headroom = sizeof(struct virtio_net_hdr);
-	}
+	if (vi->any_header_sg)
+		dev->needed_headroom = vi->hdr_len;
 
 	/* Use single tx/rx queue pair as default */
 	vi->curr_queue_pairs = 1;
-- 
MST

Our buffer length check is not strict enough for mergeable
buffers: buffer can still be shorter that header + address
by 2 bytes.

Fix that up.

Signed-off-by: Michael S. Tsirkin <mst at 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 a795a23..9c6d50f 100644
--- a/drivers/net/virtio_net.c
+++ b/drivers/net/virtio_net.c
@@ -437,7 +437,7 @@ static void receive_buf(struct virtnet_info *vi, struct
receive_queue *rq,
 	struct sk_buff *skb;
 	struct virtio_net_hdr_mrg_rxbuf *hdr;
 
-	if (unlikely(len < sizeof(struct virtio_net_hdr) + ETH_HLEN)) {
+	if (unlikely(len < vi->hdr_len + ETH_HLEN)) {
 		pr_debug("%s: short packet %i\n", dev->name, len);
 		dev->stats.rx_length_errors++;
 		if (vi->mergeable_rx_bufs) {
-- 
MST

From: Michael S. Tsirkin
 
> Too many places poke at [rs]q->vq->vdev->priv just to get
> the the vi structure.  Let's just pass the pointer around: seems
> cleaner, and might even be faster.
> 
> Signed-off-by: Michael S. Tsirkin <mst at redhat.com>
> ---
>  drivers/net/virtio_net.c | 36 +++++++++++++++++++-----------------
>  1 file changed, 19 insertions(+), 17 deletions(-)
> 
> diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
> index 57cbc7d..36f3dfc 100644
> --- a/drivers/net/virtio_net.c
> +++ b/drivers/net/virtio_net.c
...
>  static struct sk_buff *receive_big(struct net_device *dev,
> +				   struct virtnet_info *vi,
Do you need to pass 'dev' here?
Looks like it is obtainable from vi->dev (as below).

	David
>  				   struct receive_queue *rq,
>  				   void *buf,
>  				   unsigned int len)
>  {
>  	struct page *page = buf;
> -	struct sk_buff *skb = page_to_skb(rq, page, 0, len, PAGE_SIZE);
> +	struct sk_buff *skb = page_to_skb(vi, rq, page, 0, len, PAGE_SIZE);
...
> -static void receive_buf(struct receive_queue *rq, void *buf, unsigned int
len)
> +static void receive_buf(struct virtnet_info *vi, struct receive_queue *rq,
> +			void *buf, unsigned int len)
>  {
> -	struct virtnet_info *vi = rq->vq->vdev->priv;
>  	struct net_device *dev = vi->dev;
...

On Fri, Oct 24, 2014 at 10:02:15AM +0000, David Laight
wrote:
> From: Michael S. Tsirkin
>  
> > Too many places poke at [rs]q->vq->vdev->priv just to get
> > the the vi structure.  Let's just pass the pointer around: seems
> > cleaner, and might even be faster.
> > 
> > Signed-off-by: Michael S. Tsirkin <mst at redhat.com>
> > ---
> >  drivers/net/virtio_net.c | 36 +++++++++++++++++++-----------------
> >  1 file changed, 19 insertions(+), 17 deletions(-)
> > 
> > diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
> > index 57cbc7d..36f3dfc 100644
> > --- a/drivers/net/virtio_net.c
> > +++ b/drivers/net/virtio_net.c
> ...
> >  static struct sk_buff *receive_big(struct net_device *dev,
> > +				   struct virtnet_info *vi,
> 
> Do you need to pass 'dev' here?
> Looks like it is obtainable from vi->dev (as below).
> 
> 	David
> 
> >  				   struct receive_queue *rq,
> >  				   void *buf,
> >  				   unsigned int len)
> >  {
> >  	struct page *page = buf;
> > -	struct sk_buff *skb = page_to_skb(rq, page, 0, len, PAGE_SIZE);
> > +	struct sk_buff *skb = page_to_skb(vi, rq, page, 0, len, PAGE_SIZE);
> ...
> > -static void receive_buf(struct receive_queue *rq, void *buf, unsigned
int len)
> > +static void receive_buf(struct virtnet_info *vi, struct receive_queue
*rq,
> > +			void *buf, unsigned int len)
> >  {
> > -	struct virtnet_info *vi = rq->vq->vdev->priv;
> >  	struct net_device *dev = vi->dev;
> ...
It's a matter of style, isn't it?
We have dev to hand, it seems cleaner to just pass it around.

-- 
MST

"Michael S. Tsirkin" <mst at redhat.com>
writes:
> Too many places poke at [rs]q->vq->vdev->priv just to get
> the the vi structure.  Let's just pass the pointer around: seems
> cleaner, and might even be faster.
Agreed, it's neater.

Acked-by: Rusty Russell <rusty at rustcorp.com.au>

Thanks,
Rusty.
>
> Signed-off-by: Michael S. Tsirkin <mst at redhat.com>
> ---
>  drivers/net/virtio_net.c | 36 +++++++++++++++++++-----------------
>  1 file changed, 19 insertions(+), 17 deletions(-)
>
> diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
> index 57cbc7d..36f3dfc 100644
> --- a/drivers/net/virtio_net.c
> +++ b/drivers/net/virtio_net.c
> @@ -241,11 +241,11 @@ static unsigned long mergeable_buf_to_ctx(void *buf,
unsigned int truesize)
>  }
>  
>  /* Called from bottom half context */
> -static struct sk_buff *page_to_skb(struct receive_queue *rq,
> +static struct sk_buff *page_to_skb(struct virtnet_info *vi,
> +				   struct receive_queue *rq,
>  				   struct page *page, unsigned int offset,
>  				   unsigned int len, unsigned int truesize)
>  {
> -	struct virtnet_info *vi = rq->vq->vdev->priv;
>  	struct sk_buff *skb;
>  	struct skb_vnet_hdr *hdr;
>  	unsigned int copy, hdr_len, hdr_padded_len;
> @@ -328,12 +328,13 @@ static struct sk_buff *receive_small(void *buf,
unsigned int len)
>  }
>  
>  static struct sk_buff *receive_big(struct net_device *dev,
> +				   struct virtnet_info *vi,
>  				   struct receive_queue *rq,
>  				   void *buf,
>  				   unsigned int len)
>  {
>  	struct page *page = buf;
> -	struct sk_buff *skb = page_to_skb(rq, page, 0, len, PAGE_SIZE);
> +	struct sk_buff *skb = page_to_skb(vi, rq, page, 0, len, PAGE_SIZE);
>  
>  	if (unlikely(!skb))
>  		goto err;
> @@ -359,7 +360,8 @@ static struct sk_buff *receive_mergeable(struct
net_device *dev,
>  	int offset = buf - page_address(page);
>  	unsigned int truesize = max(len, mergeable_ctx_to_buf_truesize(ctx));
>  
> -	struct sk_buff *head_skb = page_to_skb(rq, page, offset, len, truesize);
> +	struct sk_buff *head_skb = page_to_skb(vi, rq, page, offset, len,
> +					       truesize);
>  	struct sk_buff *curr_skb = head_skb;
>  
>  	if (unlikely(!curr_skb))
> @@ -433,9 +435,9 @@ err_buf:
>  	return NULL;
>  }
>  
> -static void receive_buf(struct receive_queue *rq, void *buf, unsigned int
len)
> +static void receive_buf(struct virtnet_info *vi, struct receive_queue *rq,
> +			void *buf, unsigned int len)
>  {
> -	struct virtnet_info *vi = rq->vq->vdev->priv;
>  	struct net_device *dev = vi->dev;
>  	struct virtnet_stats *stats = this_cpu_ptr(vi->stats);
>  	struct sk_buff *skb;
> @@ -459,9 +461,9 @@ static void receive_buf(struct receive_queue *rq, void
*buf, unsigned int len)
>  	if (vi->mergeable_rx_bufs)
>  		skb = receive_mergeable(dev, vi, rq, (unsigned long)buf, len);
>  	else if (vi->big_packets)
> -		skb = receive_big(dev, rq, buf, len);
> +		skb = receive_big(dev, vi, rq, buf, len);
>  	else
> -		skb = receive_small(buf, len);
> +		skb = receive_small(vi, buf, len);
>  
>  	if (unlikely(!skb))
>  		return;
> @@ -530,9 +532,9 @@ frame_err:
>  	dev_kfree_skb(skb);
>  }
>  
> -static int add_recvbuf_small(struct receive_queue *rq, gfp_t gfp)
> +static int add_recvbuf_small(struct virtnet_info *vi, struct receive_queue
*rq,
> +			     gfp_t gfp)
>  {
> -	struct virtnet_info *vi = rq->vq->vdev->priv;
>  	struct sk_buff *skb;
>  	struct skb_vnet_hdr *hdr;
>  	int err;
> @@ -655,9 +657,9 @@ static int add_recvbuf_mergeable(struct receive_queue
*rq, gfp_t gfp)
>   * before we're receiving packets, or from refill_work which is
>   * careful to disable receiving (using napi_disable).
>   */
> -static bool try_fill_recv(struct receive_queue *rq, gfp_t gfp)
> +static bool try_fill_recv(struct virtnet_info *vi, struct receive_queue
*rq,
> +			  gfp_t gfp)
>  {
> -	struct virtnet_info *vi = rq->vq->vdev->priv;
>  	int err;
>  	bool oom;
>  
> @@ -668,7 +670,7 @@ static bool try_fill_recv(struct receive_queue *rq,
gfp_t gfp)
>  		else if (vi->big_packets)
>  			err = add_recvbuf_big(rq, gfp);
>  		else
> -			err = add_recvbuf_small(rq, gfp);
> +			err = add_recvbuf_small(vi, rq, gfp);
>  
>  		oom = err == -ENOMEM;
>  		if (err)
> @@ -717,7 +719,7 @@ static void refill_work(struct work_struct *work)
>  		struct receive_queue *rq = &vi->rq[i];
>  
>  		napi_disable(&rq->napi);
> -		still_empty = !try_fill_recv(rq, GFP_KERNEL);
> +		still_empty = !try_fill_recv(vi, rq, GFP_KERNEL);
>  		virtnet_napi_enable(rq);
>  
>  		/* In theory, this can happen: if we don't get any buffers in
> @@ -736,12 +738,12 @@ static int virtnet_receive(struct receive_queue *rq,
int budget)
>  
>  	while (received < budget &&
>  	       (buf = virtqueue_get_buf(rq->vq, &len)) != NULL) {
> -		receive_buf(rq, buf, len);
> +		receive_buf(vi, rq, buf, len);
>  		received++;
>  	}
>  
>  	if (rq->vq->num_free > virtqueue_get_vring_size(rq->vq) / 2)
{
> -		if (!try_fill_recv(rq, GFP_ATOMIC))
> +		if (!try_fill_recv(vi, rq, GFP_ATOMIC))
>  			schedule_delayed_work(&vi->refill, 0);
>  	}
>  
> @@ -817,7 +819,7 @@ static int virtnet_open(struct net_device *dev)
>  	for (i = 0; i < vi->max_queue_pairs; i++) {
>  		if (i < vi->curr_queue_pairs)
>  			/* Make sure we have some buffers: if oom use wq. */
> -			if (!try_fill_recv(&vi->rq[i], GFP_KERNEL))
> +			if (!try_fill_recv(vi, &vi->rq[i], GFP_KERNEL))
>  				schedule_delayed_work(&vi->refill, 0);
>  		virtnet_napi_enable(&vi->rq[i]);
>  	}
> -- 
> MST

"Michael S. Tsirkin" <mst at redhat.com>
writes:
> virtio 1.0 doesn't use virtio_net_hdr anymore, and in fact, it's
not
> really useful since virtio_net_hdr_mrg_rxbuf includes that as the first
> field anyway.
>
> Let's drop it, precalculate header len and store within vi instead.
>
> This way we can also remove struct skb_vnet_hdr.
Yes, this is definitely a win.

Acked-by: Rusty Russell <rusty at rustcorp.com.au>

Thanks,
Rusty.
>
> Signed-off-by: Michael S. Tsirkin <mst at redhat.com>
> ---
>  drivers/net/virtio_net.c | 88
++++++++++++++++++++++--------------------------
>  1 file changed, 40 insertions(+), 48 deletions(-)
>
> diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
> index 36f3dfc..a795a23 100644
> --- a/drivers/net/virtio_net.c
> +++ b/drivers/net/virtio_net.c
> @@ -123,6 +123,9 @@ struct virtnet_info {
>  	/* Host can handle any s/g split between our header and packet data */
>  	bool any_header_sg;
>  
> +	/* Packet virtio header size */
> +	u8 hdr_len;
> +
>  	/* Active statistics */
>  	struct virtnet_stats __percpu *stats;
>  
> @@ -139,21 +142,14 @@ struct virtnet_info {
>  	struct notifier_block nb;
>  };
>  
> -struct skb_vnet_hdr {
> -	union {
> -		struct virtio_net_hdr hdr;
> -		struct virtio_net_hdr_mrg_rxbuf mhdr;
> -	};
> -};
> -
>  struct padded_vnet_hdr {
> -	struct virtio_net_hdr hdr;
> +	struct virtio_net_hdr_mrg_rxbuf hdr;
>  	/*
> -	 * virtio_net_hdr should be in a separated sg buffer because of a
> -	 * QEMU bug, and data sg buffer shares same page with this header sg.
> -	 * This padding makes next sg 16 byte aligned after virtio_net_hdr.
> +	 * hdr is in a separate sg buffer, and data sg buffer shares same page
> +	 * with this header sg. This padding makes next sg 16 byte aligned
> +	 * after the header.
>  	 */
> -	char padding[6];
> +	char padding[4];
>  };
>  
>  /* Converting between virtqueue no. and kernel tx/rx queue no.
> @@ -179,9 +175,9 @@ static int rxq2vq(int rxq)
>  	return rxq * 2;
>  }
>  
> -static inline struct skb_vnet_hdr *skb_vnet_hdr(struct sk_buff *skb)
> +static inline struct virtio_net_hdr_mrg_rxbuf *skb_vnet_hdr(struct sk_buff
*skb)
>  {
> -	return (struct skb_vnet_hdr *)skb->cb;
> +	return (struct virtio_net_hdr_mrg_rxbuf *)skb->cb;
>  }
>  
>  /*
> @@ -247,7 +243,7 @@ static struct sk_buff *page_to_skb(struct virtnet_info
*vi,
>  				   unsigned int len, unsigned int truesize)
>  {
>  	struct sk_buff *skb;
> -	struct skb_vnet_hdr *hdr;
> +	struct virtio_net_hdr_mrg_rxbuf *hdr;
>  	unsigned int copy, hdr_len, hdr_padded_len;
>  	char *p;
>  
> @@ -260,13 +256,11 @@ static struct sk_buff *page_to_skb(struct
virtnet_info *vi,
>  
>  	hdr = skb_vnet_hdr(skb);
>  
> -	if (vi->mergeable_rx_bufs) {
> -		hdr_len = sizeof hdr->mhdr;
> -		hdr_padded_len = sizeof hdr->mhdr;
> -	} else {
> -		hdr_len = sizeof hdr->hdr;
> +	hdr_len = vi->hdr_len;
> +	if (vi->mergeable_rx_bufs)
> +		hdr_padded_len = sizeof *hdr;
> +	else
>  		hdr_padded_len = sizeof(struct padded_vnet_hdr);
> -	}
>  
>  	memcpy(hdr, p, hdr_len);
>  
> @@ -317,11 +311,11 @@ static struct sk_buff *page_to_skb(struct
virtnet_info *vi,
>  	return skb;
>  }
>  
> -static struct sk_buff *receive_small(void *buf, unsigned int len)
> +static struct sk_buff *receive_small(struct virtnet_info *vi, void *buf,
unsigned int len)
>  {
>  	struct sk_buff * skb = buf;
>  
> -	len -= sizeof(struct virtio_net_hdr);
> +	len -= vi->hdr_len;
>  	skb_trim(skb, len);
>  
>  	return skb;
> @@ -354,8 +348,8 @@ static struct sk_buff *receive_mergeable(struct
net_device *dev,
>  					 unsigned int len)
>  {
>  	void *buf = mergeable_ctx_to_buf_address(ctx);
> -	struct skb_vnet_hdr *hdr = buf;
> -	u16 num_buf = virtio16_to_cpu(rq->vq->vdev,
hdr->mhdr.num_buffers);
> +	struct virtio_net_hdr_mrg_rxbuf *hdr = buf;
> +	u16 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 truesize = max(len, mergeable_ctx_to_buf_truesize(ctx));
> @@ -373,8 +367,8 @@ static struct sk_buff *receive_mergeable(struct
net_device *dev,
>  		if (unlikely(!ctx)) {
>  			pr_debug("%s: rx error: %d buffers out of %d missing\n",
>  				 dev->name, num_buf,
> -				 virtio16_to_cpu(rq->vq->vdev,
> -						 hdr->mhdr.num_buffers));
> +				 virtio16_to_cpu(vi->vdev,
> +						 hdr->num_buffers));
>  			dev->stats.rx_length_errors++;
>  			goto err_buf;
>  		}
> @@ -441,7 +435,7 @@ static void receive_buf(struct virtnet_info *vi, struct
receive_queue *rq,
>  	struct net_device *dev = vi->dev;
>  	struct virtnet_stats *stats = this_cpu_ptr(vi->stats);
>  	struct sk_buff *skb;
> -	struct skb_vnet_hdr *hdr;
> +	struct virtio_net_hdr_mrg_rxbuf *hdr;
>  
>  	if (unlikely(len < sizeof(struct virtio_net_hdr) + ETH_HLEN)) {
>  		pr_debug("%s: short packet %i\n", dev->name, len);
> @@ -536,7 +530,7 @@ static int add_recvbuf_small(struct virtnet_info *vi,
struct receive_queue *rq,
>  			     gfp_t gfp)
>  {
>  	struct sk_buff *skb;
> -	struct skb_vnet_hdr *hdr;
> +	struct virtio_net_hdr_mrg_rxbuf *hdr;
>  	int err;
>  
>  	skb = __netdev_alloc_skb_ip_align(vi->dev, GOOD_PACKET_LEN, gfp);
> @@ -547,7 +541,7 @@ static int add_recvbuf_small(struct virtnet_info *vi,
struct receive_queue *rq,
>  
>  	hdr = skb_vnet_hdr(skb);
>  	sg_init_table(rq->sg, MAX_SKB_FRAGS + 2);
> -	sg_set_buf(rq->sg, &hdr->hdr, sizeof hdr->hdr);
> +	sg_set_buf(rq->sg, hdr, vi->hdr_len);
>  	skb_to_sgvec(skb, rq->sg + 1, 0, skb->len);
>  
>  	err = virtqueue_add_inbuf(rq->vq, rq->sg, 2, skb, gfp);
> @@ -557,7 +551,8 @@ static int add_recvbuf_small(struct virtnet_info *vi,
struct receive_queue *rq,
>  	return err;
>  }
>  
> -static int add_recvbuf_big(struct receive_queue *rq, gfp_t gfp)
> +static int add_recvbuf_big(struct virtnet_info *vi, struct receive_queue
*rq,
> +			   gfp_t gfp)
>  {
>  	struct page *first, *list = NULL;
>  	char *p;
> @@ -588,8 +583,8 @@ static int add_recvbuf_big(struct receive_queue *rq,
gfp_t gfp)
>  	p = page_address(first);
>  
>  	/* rq->sg[0], rq->sg[1] share the same page */
> -	/* a separated rq->sg[0] for virtio_net_hdr only due to QEMU bug */
> -	sg_set_buf(&rq->sg[0], p, sizeof(struct virtio_net_hdr));
> +	/* a separated rq->sg[0] for header - required in case !any_header_sg
*/
> +	sg_set_buf(&rq->sg[0], p, vi->hdr_len);
>  
>  	/* rq->sg[1] for data packet, from offset */
>  	offset = sizeof(struct padded_vnet_hdr);
> @@ -668,7 +663,7 @@ static bool try_fill_recv(struct virtnet_info *vi,
struct receive_queue *rq,
>  		if (vi->mergeable_rx_bufs)
>  			err = add_recvbuf_mergeable(rq, gfp);
>  		else if (vi->big_packets)
> -			err = add_recvbuf_big(rq, gfp);
> +			err = add_recvbuf_big(vi, rq, gfp);
>  		else
>  			err = add_recvbuf_small(vi, rq, gfp);
>  
> @@ -848,18 +843,14 @@ static void free_old_xmit_skbs(struct send_queue *sq)
>  
>  static int xmit_skb(struct send_queue *sq, struct sk_buff *skb)
>  {
> -	struct skb_vnet_hdr *hdr;
> +	struct virtio_net_hdr_mrg_rxbuf *hdr;
>  	const unsigned char *dest = ((struct ethhdr *)skb->data)->h_dest;
>  	struct virtnet_info *vi = sq->vq->vdev->priv;
>  	unsigned num_sg;
> -	unsigned hdr_len;
> +	unsigned hdr_len = vi->hdr_len;
>  	bool can_push;
>  
>  	pr_debug("%s: xmit %p %pM\n", vi->dev->name, skb, dest);
> -	if (vi->mergeable_rx_bufs)
> -		hdr_len = sizeof hdr->mhdr;
> -	else
> -		hdr_len = sizeof hdr->hdr;
>  
>  	can_push = vi->any_header_sg &&
>  		!((unsigned long)skb->data & (__alignof__(*hdr) - 1)) &&
> @@ -867,7 +858,7 @@ static int xmit_skb(struct send_queue *sq, struct
sk_buff *skb)
>  	/* Even if we can, don't push here yet as this would skew
>  	 * csum_start offset below. */
>  	if (can_push)
> -		hdr = (struct skb_vnet_hdr *)(skb->data - hdr_len);
> +		hdr = (struct virtio_net_hdr_mrg_rxbuf *)(skb->data - hdr_len);
>  	else
>  		hdr = skb_vnet_hdr(skb);
>  
> @@ -902,7 +893,7 @@ static int xmit_skb(struct send_queue *sq, struct
sk_buff *skb)
>  	}
>  
>  	if (vi->mergeable_rx_bufs)
> -		hdr->mhdr.num_buffers = 0;
> +		hdr->num_buffers = 0;
>  
>  	sg_init_table(sq->sg, MAX_SKB_FRAGS + 2);
>  	if (can_push) {
> @@ -1773,18 +1764,19 @@ static int virtnet_probe(struct virtio_device
*vdev)
>  	if (virtio_has_feature(vdev, VIRTIO_NET_F_MRG_RXBUF))
>  		vi->mergeable_rx_bufs = true;
>  
> +	if (virtio_has_feature(vdev, VIRTIO_NET_F_MRG_RXBUF))
> +		vi->hdr_len = sizeof(struct virtio_net_hdr_mrg_rxbuf);
> +	else
> +		vi->hdr_len = sizeof(struct virtio_net_hdr);
> +
>  	if (virtio_has_feature(vdev, VIRTIO_F_ANY_LAYOUT))
>  		vi->any_header_sg = true;
>  
>  	if (virtio_has_feature(vdev, VIRTIO_NET_F_CTRL_VQ))
>  		vi->has_cvq = true;
>  
> -	if (vi->any_header_sg) {
> -		if (vi->mergeable_rx_bufs)
> -			dev->needed_headroom = sizeof(struct virtio_net_hdr_mrg_rxbuf);
> -		else
> -			dev->needed_headroom = sizeof(struct virtio_net_hdr);
> -	}
> +	if (vi->any_header_sg)
> +		dev->needed_headroom = vi->hdr_len;
>  
>  	/* Use single tx/rx queue pair as default */
>  	vi->curr_queue_pairs = 1;
> -- 
> MST