Xen devel - Nov 2013 - [PATCH net-next] xen-netback: Fix pull size in checksum_setup_ip*

Before checksum setup we need to make sure we have enough data in linear
buffer, so we do a pull if not. The calculation of the required data counts
with skb->network_header, which is the headroom size actually. This is not
necessary, and can cause unreasonable pulling. I''ve also removed
MAX_IPOPTLEN, as we only need the base header at that point.

Signed-off-by: Zoltan Kiss <zoltan.kiss@citrix.com>

---
 drivers/net/xen-netback/netback.c |   24 ++++++++----------------
 1 file changed, 8 insertions(+), 16 deletions(-)

diff --git a/drivers/net/xen-netback/netback.c
b/drivers/net/xen-netback/netback.c
index 919b650..adadf76 100644
--- a/drivers/net/xen-netback/netback.c
+++ b/drivers/net/xen-netback/netback.c
@@ -1169,8 +1169,7 @@ static int checksum_setup_ip(struct xenvif *vif, struct
sk_buff *skb,
 
 	off = sizeof(struct iphdr);
 
-	header_size = skb->network_header + off + MAX_IPOPTLEN;
-	maybe_pull_tail(skb, header_size);
+	maybe_pull_tail(skb, off);
 
 	off = iph->ihl * 4;
 
@@ -1183,8 +1182,7 @@ static int checksum_setup_ip(struct xenvif *vif, struct
sk_buff *skb,
 		if (recalculate_partial_csum) {
 			struct tcphdr *tcph = tcp_hdr(skb);
 
-			header_size = skb->network_header +
-				off +
+			header_size = off +
 				sizeof(struct tcphdr);
 			maybe_pull_tail(skb, header_size);
 
@@ -1201,8 +1199,7 @@ static int checksum_setup_ip(struct xenvif *vif, struct
sk_buff *skb,
 		if (recalculate_partial_csum) {
 			struct udphdr *udph = udp_hdr(skb);
 
-			header_size = skb->network_header +
-				off +
+			header_size = off +
 				sizeof(struct udphdr);
 			maybe_pull_tail(skb, header_size);
 
@@ -1241,8 +1238,7 @@ static int checksum_setup_ipv6(struct xenvif *vif, struct
sk_buff *skb,
 
 	off = sizeof(struct ipv6hdr);
 
-	header_size = skb->network_header + off;
-	maybe_pull_tail(skb, header_size);
+	maybe_pull_tail(skb, off);
 
 	nexthdr = ipv6h->nexthdr;
 
@@ -1254,8 +1250,7 @@ static int checksum_setup_ipv6(struct xenvif *vif, struct
sk_buff *skb,
 		case IPPROTO_ROUTING: {
 			struct ipv6_opt_hdr *hp = (void *)(skb->data + off);
 
-			header_size = skb->network_header +
-				off +
+			header_size = off +
 				sizeof(struct ipv6_opt_hdr);
 			maybe_pull_tail(skb, header_size);
 
@@ -1266,8 +1261,7 @@ static int checksum_setup_ipv6(struct xenvif *vif, struct
sk_buff *skb,
 		case IPPROTO_AH: {
 			struct ip_auth_hdr *hp = (void *)(skb->data + off);
 
-			header_size = skb->network_header +
-				off +
+			header_size = off +
 				sizeof(struct ip_auth_hdr);
 			maybe_pull_tail(skb, header_size);
 
@@ -1305,8 +1299,7 @@ static int checksum_setup_ipv6(struct xenvif *vif, struct
sk_buff *skb,
 		if (recalculate_partial_csum) {
 			struct tcphdr *tcph = tcp_hdr(skb);
 
-			header_size = skb->network_header +
-				off +
+			header_size = off +
 				sizeof(struct tcphdr);
 			maybe_pull_tail(skb, header_size);
 
@@ -1324,8 +1317,7 @@ static int checksum_setup_ipv6(struct xenvif *vif, struct
sk_buff *skb,
 		if (recalculate_partial_csum) {
 			struct udphdr *udph = udp_hdr(skb);
 
-			header_size = skb->network_header +
-				off +
+			header_size = off +
 				sizeof(struct udphdr);
 			maybe_pull_tail(skb, header_size);

Or would it make sense to drop header_size and use off + something as 
function parameters?

Zoli

On 29/11/13 18:56, Zoltan Kiss wrote:
> Before checksum setup we need to make sure we have enough data in linear
> buffer, so we do a pull if not. The calculation of the required data counts
> with skb->network_header, which is the headroom size actually. This is
not
> necessary, and can cause unreasonable pulling. I''ve also removed
> MAX_IPOPTLEN, as we only need the base header at that point.
>
> Signed-off-by: Zoltan Kiss <zoltan.kiss@citrix.com>
>
> ---
>   drivers/net/xen-netback/netback.c |   24 ++++++++----------------
>   1 file changed, 8 insertions(+), 16 deletions(-)
>
> diff --git a/drivers/net/xen-netback/netback.c
b/drivers/net/xen-netback/netback.c
> index 919b650..adadf76 100644
> --- a/drivers/net/xen-netback/netback.c
> +++ b/drivers/net/xen-netback/netback.c
> @@ -1169,8 +1169,7 @@ static int checksum_setup_ip(struct xenvif *vif,
struct sk_buff *skb,
>
>   	off = sizeof(struct iphdr);
>
> -	header_size = skb->network_header + off + MAX_IPOPTLEN;
> -	maybe_pull_tail(skb, header_size);
> +	maybe_pull_tail(skb, off);
>
>   	off = iph->ihl * 4;
>
> @@ -1183,8 +1182,7 @@ static int checksum_setup_ip(struct xenvif *vif,
struct sk_buff *skb,
>   		if (recalculate_partial_csum) {
>   			struct tcphdr *tcph = tcp_hdr(skb);
>
> -			header_size = skb->network_header +
> -				off +
> +			header_size = off +
>   				sizeof(struct tcphdr);
>   			maybe_pull_tail(skb, header_size);
>
> @@ -1201,8 +1199,7 @@ static int checksum_setup_ip(struct xenvif *vif,
struct sk_buff *skb,
>   		if (recalculate_partial_csum) {
>   			struct udphdr *udph = udp_hdr(skb);
>
> -			header_size = skb->network_header +
> -				off +
> +			header_size = off +
>   				sizeof(struct udphdr);
>   			maybe_pull_tail(skb, header_size);
>
> @@ -1241,8 +1238,7 @@ static int checksum_setup_ipv6(struct xenvif *vif,
struct sk_buff *skb,
>
>   	off = sizeof(struct ipv6hdr);
>
> -	header_size = skb->network_header + off;
> -	maybe_pull_tail(skb, header_size);
> +	maybe_pull_tail(skb, off);
>
>   	nexthdr = ipv6h->nexthdr;
>
> @@ -1254,8 +1250,7 @@ static int checksum_setup_ipv6(struct xenvif *vif,
struct sk_buff *skb,
>   		case IPPROTO_ROUTING: {
>   			struct ipv6_opt_hdr *hp = (void *)(skb->data + off);
>
> -			header_size = skb->network_header +
> -				off +
> +			header_size = off +
>   				sizeof(struct ipv6_opt_hdr);
>   			maybe_pull_tail(skb, header_size);
>
> @@ -1266,8 +1261,7 @@ static int checksum_setup_ipv6(struct xenvif *vif,
struct sk_buff *skb,
>   		case IPPROTO_AH: {
>   			struct ip_auth_hdr *hp = (void *)(skb->data + off);
>
> -			header_size = skb->network_header +
> -				off +
> +			header_size = off +
>   				sizeof(struct ip_auth_hdr);
>   			maybe_pull_tail(skb, header_size);
>
> @@ -1305,8 +1299,7 @@ static int checksum_setup_ipv6(struct xenvif *vif,
struct sk_buff *skb,
>   		if (recalculate_partial_csum) {
>   			struct tcphdr *tcph = tcp_hdr(skb);
>
> -			header_size = skb->network_header +
> -				off +
> +			header_size = off +
>   				sizeof(struct tcphdr);
>   			maybe_pull_tail(skb, header_size);
>
> @@ -1324,8 +1317,7 @@ static int checksum_setup_ipv6(struct xenvif *vif,
struct sk_buff *skb,
>   		if (recalculate_partial_csum) {
>   			struct udphdr *udph = udp_hdr(skb);
>
> -			header_size = skb->network_header +
> -				off +
> +			header_size = off +
>   				sizeof(struct udphdr);
>   			maybe_pull_tail(skb, header_size);
>
>

From: Zoltan Kiss <zoltan.kiss@citrix.com>
Date: Fri, 29 Nov 2013 18:56:47 +0000
> Before checksum setup we need to make sure we have enough data in linear
> buffer, so we do a pull if not. The calculation of the required data counts
> with skb->network_header, which is the headroom size actually. This is
not
> necessary, and can cause unreasonable pulling. I''ve also removed
> MAX_IPOPTLEN, as we only need the base header at that point.
> 
> Signed-off-by: Zoltan Kiss <zoltan.kiss@citrix.com>
The net-next tree is no open, therefore it is not appropriate to submit
patches targetting to that tree at this time.

I will make an explicit announcement on netdev when I have openned
the net-next tree, so it will be clear exactly when you can submit
patches targetting that tree again.

Thank you.

> -----Original Message-----
> From: Zoltan Kiss
> Sent: 29 November 2013 20:26
> To: Ian Campbell; Wei Liu; xen-devel@lists.xenproject.org;
> netdev@vger.kernel.org; linux-kernel@vger.kernel.org; Jonathan Davies;
> Paul Durrant
> Cc: Zoltan Kiss
> Subject: Re: [PATCH net-next] xen-netback: Fix pull size in
> checksum_setup_ip*
> 
> Or would it make sense to drop header_size and use off + something as
> function parameters?
> 
Possibly best not to top-post to your own thread ;-)

Agreed; now that we''ve gone from mostly 3 variable additions to 2, I
don''t think we need header_size to neaten things up any more.

  Paul
> Zoli
> 
> On 29/11/13 18:56, Zoltan Kiss wrote:
> > Before checksum setup we need to make sure we have enough data in
> linear
> > buffer, so we do a pull if not. The calculation of the required data
counts
> > with skb->network_header, which is the headroom size actually. This
is not
> > necessary, and can cause unreasonable pulling. I''ve also
removed
> > MAX_IPOPTLEN, as we only need the base header at that point.
> >
> > Signed-off-by: Zoltan Kiss <zoltan.kiss@citrix.com>
> >
> > ---
> >   drivers/net/xen-netback/netback.c |   24 ++++++++----------------
> >   1 file changed, 8 insertions(+), 16 deletions(-)
> >
> > diff --git a/drivers/net/xen-netback/netback.c b/drivers/net/xen-
> netback/netback.c
> > index 919b650..adadf76 100644
> > --- a/drivers/net/xen-netback/netback.c
> > +++ b/drivers/net/xen-netback/netback.c
> > @@ -1169,8 +1169,7 @@ static int checksum_setup_ip(struct xenvif *vif,
> struct sk_buff *skb,
> >
> >   	off = sizeof(struct iphdr);
> >
> > -	header_size = skb->network_header + off + MAX_IPOPTLEN;
> > -	maybe_pull_tail(skb, header_size);
> > +	maybe_pull_tail(skb, off);
> >
> >   	off = iph->ihl * 4;
> >
> > @@ -1183,8 +1182,7 @@ static int checksum_setup_ip(struct xenvif *vif,
> struct sk_buff *skb,
> >   		if (recalculate_partial_csum) {
> >   			struct tcphdr *tcph = tcp_hdr(skb);
> >
> > -			header_size = skb->network_header +
> > -				off +
> > +			header_size = off +
> >   				sizeof(struct tcphdr);
> >   			maybe_pull_tail(skb, header_size);
> >
> > @@ -1201,8 +1199,7 @@ static int checksum_setup_ip(struct xenvif *vif,
> struct sk_buff *skb,
> >   		if (recalculate_partial_csum) {
> >   			struct udphdr *udph = udp_hdr(skb);
> >
> > -			header_size = skb->network_header +
> > -				off +
> > +			header_size = off +
> >   				sizeof(struct udphdr);
> >   			maybe_pull_tail(skb, header_size);
> >
> > @@ -1241,8 +1238,7 @@ static int checksum_setup_ipv6(struct xenvif
*vif,
> struct sk_buff *skb,
> >
> >   	off = sizeof(struct ipv6hdr);
> >
> > -	header_size = skb->network_header + off;
> > -	maybe_pull_tail(skb, header_size);
> > +	maybe_pull_tail(skb, off);
> >
> >   	nexthdr = ipv6h->nexthdr;
> >
> > @@ -1254,8 +1250,7 @@ static int checksum_setup_ipv6(struct xenvif
*vif,
> struct sk_buff *skb,
> >   		case IPPROTO_ROUTING: {
> >   			struct ipv6_opt_hdr *hp = (void *)(skb->data + off);
> >
> > -			header_size = skb->network_header +
> > -				off +
> > +			header_size = off +
> >   				sizeof(struct ipv6_opt_hdr);
> >   			maybe_pull_tail(skb, header_size);
> >
> > @@ -1266,8 +1261,7 @@ static int checksum_setup_ipv6(struct xenvif
*vif,
> struct sk_buff *skb,
> >   		case IPPROTO_AH: {
> >   			struct ip_auth_hdr *hp = (void *)(skb->data + off);
> >
> > -			header_size = skb->network_header +
> > -				off +
> > +			header_size = off +
> >   				sizeof(struct ip_auth_hdr);
> >   			maybe_pull_tail(skb, header_size);
> >
> > @@ -1305,8 +1299,7 @@ static int checksum_setup_ipv6(struct xenvif
*vif,
> struct sk_buff *skb,
> >   		if (recalculate_partial_csum) {
> >   			struct tcphdr *tcph = tcp_hdr(skb);
> >
> > -			header_size = skb->network_header +
> > -				off +
> > +			header_size = off +
> >   				sizeof(struct tcphdr);
> >   			maybe_pull_tail(skb, header_size);
> >
> > @@ -1324,8 +1317,7 @@ static int checksum_setup_ipv6(struct xenvif
*vif,
> struct sk_buff *skb,
> >   		if (recalculate_partial_csum) {
> >   			struct udphdr *udph = udp_hdr(skb);
> >
> > -			header_size = skb->network_header +
> > -				off +
> > +			header_size = off +
> >   				sizeof(struct udphdr);
> >   			maybe_pull_tail(skb, header_size);
> >
> >

> -----Original Message-----
> From: David Miller [mailto:davem@davemloft.net]
> Sent: 30 November 2013 21:10
> To: Zoltan Kiss
> Cc: Ian Campbell; Wei Liu; xen-devel@lists.xenproject.org;
> netdev@vger.kernel.org; linux-kernel@vger.kernel.org; Jonathan Davies;
> Paul Durrant
> Subject: Re: [PATCH net-next] xen-netback: Fix pull size in
> checksum_setup_ip*
> 
> From: Zoltan Kiss <zoltan.kiss@citrix.com>
> Date: Fri, 29 Nov 2013 18:56:47 +0000
> 
> > Before checksum setup we need to make sure we have enough data in
> linear
> > buffer, so we do a pull if not. The calculation of the required data
counts
> > with skb->network_header, which is the headroom size actually. This
is not
> > necessary, and can cause unreasonable pulling. I''ve also
removed
> > MAX_IPOPTLEN, as we only need the base header at that point.
> >
> > Signed-off-by: Zoltan Kiss <zoltan.kiss@citrix.com>
> 
> The net-next tree is no open, therefore it is not appropriate to submit
> patches targetting to that tree at this time.
> 
> I will make an explicit announcement on netdev when I have openned
> the net-next tree, so it will be clear exactly when you can submit
> patches targetting that tree again.
> 
> Thank you.
I think this patch should be for net rather than net-next anyway. Since it
appears that maybe_pull_tail() has bigger problems I''ll incorporate
this fix as part of v4 of the fix for fragment detection.

  Paul