[PATCH v2 net 1/2] tuntap: correctly linearize skb when zerocopy is used

[PATCH v2 net 1/2] tuntap: correctly linearize skb when zerocopy is used

[Jason Wang]

Userspace may produce vectors greater than MAX_SKB_FRAGS. When we try to
linearize parts of the skb to let the rest of iov to be fit in
the frags, we need count copylen into linear when calling tun_alloc_skb()
instead of partly counting it into data_len. Since this breaks
zerocopy_sg_from_iovec() since its inner counter assumes nr_frags should
be zero at beginning. This cause nr_frags to be increased wrongly without
setting the correct frags.

This bug were introduced from 0690899b4d4501b3505be069b9a687e68ccbe15b
(tun: experimental zero copy tx support)

Cc: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Jason Wang <jasowang@redhat.com>
---
- This patch is needed for stable.
- Changes from v1: introduce a local variable to track linear size
---
 drivers/net/tun.c |    9 ++++++---
 1 files changed, 6 insertions(+), 3 deletions(-)

diff --git a/drivers/net/tun.c b/drivers/net/tun.c
index 9c61f87..c3cb60b 100644
--- a/drivers/net/tun.c
+++ b/drivers/net/tun.c
@@ -1044,7 +1044,7 @@ static ssize_t tun_get_user(struct tun_struct *tun, struct tun_file *tfile,
 {
 	struct tun_pi pi = { 0, cpu_to_be16(ETH_P_IP) };
 	struct sk_buff *skb;
-	size_t len = total_len, align = NET_SKB_PAD;
+	size_t len = total_len, align = NET_SKB_PAD, linear;
 	struct virtio_net_hdr gso = { 0 };
 	int offset = 0;
 	int copylen;
@@ -1108,10 +1108,13 @@ static ssize_t tun_get_user(struct tun_struct *tun, struct tun_file *tfile,
 			copylen = gso.hdr_len;
 		if (!copylen)
 			copylen = GOODCOPY_LEN;
-	} else
+		linear = copylen;
+	} else {
 		copylen = len;
+		linear = gso.hdr_len;
+	}
 
-	skb = tun_alloc_skb(tfile, align, copylen, gso.hdr_len, noblock);
+	skb = tun_alloc_skb(tfile, align, copylen, linear, noblock);
 	if (IS_ERR(skb)) {
 		if (PTR_ERR(skb) != -EAGAIN)
 			tun->dev->stats.rx_dropped++;
-- 
1.7.1

[PATCH v2 net 2/2] macvtap: correctly linearize skb when zerocopy is used

[Jason Wang]

Userspace may produce vectors greater than MAX_SKB_FRAGS. When we try to
linearize parts of the skb to let the rest of iov to be fit in
the frags, we need count copylen into linear when calling macvtap_alloc_skb()
instead of partly counting it into data_len. Since this breaks
zerocopy_sg_from_iovec() since its inner counter assumes nr_frags should
be zero at beginning. This cause nr_frags to be increased wrongly without
setting the correct frags.

This bug were introduced from b92946e2919134ebe2a4083e4302236295ea2a73
(macvtap: zerocopy: validate vectors before building skb).

Cc: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Jason Wang <jasowang@redhat.com>
---
- This patch is needed for stable.
- Changes from v1: introduce a local variable to track linear size.
---
 drivers/net/macvtap.c |    8 ++++++--
 1 files changed, 6 insertions(+), 2 deletions(-)

diff --git a/drivers/net/macvtap.c b/drivers/net/macvtap.c
index b6dd6a7..502d948 100644
--- a/drivers/net/macvtap.c
+++ b/drivers/net/macvtap.c
@@ -647,6 +647,7 @@ static ssize_t macvtap_get_user(struct macvtap_queue *q, struct msghdr *m,
 	int vnet_hdr_len = 0;
 	int copylen = 0;
 	bool zerocopy = false;
+	size_t linear;
 
 	if (q->flags & IFF_VNET_HDR) {
 		vnet_hdr_len = q->vnet_hdr_sz;
@@ -701,11 +702,14 @@ static ssize_t macvtap_get_user(struct macvtap_queue *q, struct msghdr *m,
 			copylen = vnet_hdr.hdr_len;
 		if (!copylen)
 			copylen = GOODCOPY_LEN;
-	} else
+		linear = copylen;
+	} else {
 		copylen = len;
+		linear = vnet_hdr.hdr_len;
+	}
 
 	skb = macvtap_alloc_skb(&q->sk, NET_IP_ALIGN, copylen,
-				vnet_hdr.hdr_len, noblock, &err);
+				linear, noblock, &err);
 	if (!skb)
 		goto err;
 
-- 
1.7.1

Re: [PATCH v2 net 1/2] tuntap: correctly linearize skb when zerocopy is used

[Michael S. Tsirkin]

On Wed, Jul 10, 2013 at 01:43:27PM +0800, Jason Wang wrote:
> Userspace may produce vectors greater than MAX_SKB_FRAGS. When we try to
> linearize parts of the skb to let the rest of iov to be fit in
> the frags, we need count copylen into linear when calling tun_alloc_skb()
> instead of partly counting it into data_len. Since this breaks
> zerocopy_sg_from_iovec() since its inner counter assumes nr_frags should
> be zero at beginning. This cause nr_frags to be increased wrongly without
> setting the correct frags.
> 
> This bug were introduced from 0690899b4d4501b3505be069b9a687e68ccbe15b
> (tun: experimental zero copy tx support)
> 
> Cc: Michael S. Tsirkin <mst@redhat.com>
> Signed-off-by: Jason Wang <jasowang@redhat.com>


Acked-by: Michael S. Tsirkin <mst@redhat.com>

> ---
> - This patch is needed for stable.
> - Changes from v1: introduce a local variable to track linear size
> ---
>  drivers/net/tun.c |    9 ++++++---
>  1 files changed, 6 insertions(+), 3 deletions(-)
> 
> diff --git a/drivers/net/tun.c b/drivers/net/tun.c
> index 9c61f87..c3cb60b 100644
> --- a/drivers/net/tun.c
> +++ b/drivers/net/tun.c
> @@ -1044,7 +1044,7 @@ static ssize_t tun_get_user(struct tun_struct *tun, struct tun_file *tfile,
>  {
>  	struct tun_pi pi = { 0, cpu_to_be16(ETH_P_IP) };
>  	struct sk_buff *skb;
> -	size_t len = total_len, align = NET_SKB_PAD;
> +	size_t len = total_len, align = NET_SKB_PAD, linear;
>  	struct virtio_net_hdr gso = { 0 };
>  	int offset = 0;
>  	int copylen;
> @@ -1108,10 +1108,13 @@ static ssize_t tun_get_user(struct tun_struct *tun, struct tun_file *tfile,
>  			copylen = gso.hdr_len;
>  		if (!copylen)
>  			copylen = GOODCOPY_LEN;
> -	} else
> +		linear = copylen;
> +	} else {
>  		copylen = len;
> +		linear = gso.hdr_len;
> +	}
>  
> -	skb = tun_alloc_skb(tfile, align, copylen, gso.hdr_len, noblock);
> +	skb = tun_alloc_skb(tfile, align, copylen, linear, noblock);
>  	if (IS_ERR(skb)) {
>  		if (PTR_ERR(skb) != -EAGAIN)
>  			tun->dev->stats.rx_dropped++;
> -- 
> 1.7.1

Re: [PATCH v2 net 2/2] macvtap: correctly linearize skb when zerocopy is used

[Michael S. Tsirkin]

On Wed, Jul 10, 2013 at 01:43:28PM +0800, Jason Wang wrote:
> Userspace may produce vectors greater than MAX_SKB_FRAGS. When we try to
> linearize parts of the skb to let the rest of iov to be fit in
> the frags, we need count copylen into linear when calling macvtap_alloc_skb()
> instead of partly counting it into data_len. Since this breaks
> zerocopy_sg_from_iovec() since its inner counter assumes nr_frags should
> be zero at beginning. This cause nr_frags to be increased wrongly without
> setting the correct frags.
> 
> This bug were introduced from b92946e2919134ebe2a4083e4302236295ea2a73
> (macvtap: zerocopy: validate vectors before building skb).
> 
> Cc: Michael S. Tsirkin <mst@redhat.com>
> Signed-off-by: Jason Wang <jasowang@redhat.com>


Acked-by: Michael S. Tsirkin <mst@redhat.com>

> ---
> - This patch is needed for stable.
> - Changes from v1: introduce a local variable to track linear size.
> ---
>  drivers/net/macvtap.c |    8 ++++++--
>  1 files changed, 6 insertions(+), 2 deletions(-)
> 
> diff --git a/drivers/net/macvtap.c b/drivers/net/macvtap.c
> index b6dd6a7..502d948 100644
> --- a/drivers/net/macvtap.c
> +++ b/drivers/net/macvtap.c
> @@ -647,6 +647,7 @@ static ssize_t macvtap_get_user(struct macvtap_queue *q, struct msghdr *m,
>  	int vnet_hdr_len = 0;
>  	int copylen = 0;
>  	bool zerocopy = false;
> +	size_t linear;
>  
>  	if (q->flags & IFF_VNET_HDR) {
>  		vnet_hdr_len = q->vnet_hdr_sz;
> @@ -701,11 +702,14 @@ static ssize_t macvtap_get_user(struct macvtap_queue *q, struct msghdr *m,
>  			copylen = vnet_hdr.hdr_len;
>  		if (!copylen)
>  			copylen = GOODCOPY_LEN;
> -	} else
> +		linear = copylen;
> +	} else {
>  		copylen = len;
> +		linear = vnet_hdr.hdr_len;
> +	}
>  
>  	skb = macvtap_alloc_skb(&q->sk, NET_IP_ALIGN, copylen,
> -				vnet_hdr.hdr_len, noblock, &err);
> +				linear, noblock, &err);
>  	if (!skb)
>  		goto err;
>  
> -- 
> 1.7.1