| Message ID | 1368773342-31882-1-git-send-email-horms@verge.net.au |
|---|---|
| State | Changes Requested, archived |
| Delegated to: | David Miller |
| Headers | show |
On Fri, 2013-05-17 at 15:49 +0900, Simon Horman wrote: > In the case where the ability to offload checksums changes > then it may be necessary for software checksumming of an skb to occur. > > An example of such a case is where a non-GRE packet is received but > is to be encapsulated and transmitted as GRE. > > Another example relates to my proposed support for for GSO packets > that are non-MPLS when received but MPLS when transmitted. > > The cost of this change is that csum may be checked when it previously > was not. In the case where csum is true this is pure overhead. In the > case where csum is false it leads to software checksumming, which > I believe also leads to correct checksums in transmitted packets for > the cases described above. > > This patch relies on the return value of can_checksum_protocol() > being correct and in turn the return value of skb_network_protocol(), > used to provide the protocol parameter of can_checksum_protocol(), > being correct. It also relies on the features passed to skb_segment() > and in turn to can_checksum_protocol() being correct. > > I believe that this problem does has not been observed for VLANs because it > appears that all drivers except xgbe set vlan_features such that that the > checksum offload support for VLAN packets is greater than or equal to that > of non-VLAN packets. > > I wonder if the code in xgbe may be an oversight and the hardware does > support checksumming of VLAN packets. If so it may be worth updating the > vlan_features of the driver as this patch will force such checksums to be > performed in software rather than hardware. > > Signed-off-by: Simon Horman <horms@verge.net.au> > > --- > > Tested using the bnx2 driver. > > v2 > Rewrite changlog after further analysis > --- > net/core/skbuff.c | 3 ++- > 1 file changed, 2 insertions(+), 1 deletion(-) > > diff --git a/net/core/skbuff.c b/net/core/skbuff.c > index af9185d..3a43be4 100644 > --- a/net/core/skbuff.c > +++ b/net/core/skbuff.c > @@ -2853,7 +2853,7 @@ struct sk_buff *skb_segment(struct sk_buff *skb, netdev_features_t features) > doffset + tnl_hlen); > > if (fskb != skb_shinfo(skb)->frag_list) > - continue; > + goto csum; > > if (!sg) { > nskb->ip_summed = CHECKSUM_NONE; > @@ -2917,6 +2917,7 @@ skip_fraglist: > nskb->len += nskb->data_len; > nskb->truesize += nskb->data_len; > > +csum: > if (!csum) { > nskb->csum = skb_checksum(nskb, doffset, > nskb->len - doffset, 0); I had hard time to understand this patch, but I think I now get it. Changelog seems to be very specific, while the change looks like a more generic problem about skbs having a frag_list ? Oh well... skb_segment() is probably too complex for my understanding. Could you please change the label name ? perform_csum_check for example, and maybe it would be easier :) -- To unsubscribe from this list: send the line "unsubscribe netdev" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On Fri, May 17, 2013 at 10:02:03AM -0700, Eric Dumazet wrote: > On Fri, 2013-05-17 at 15:49 +0900, Simon Horman wrote: > > In the case where the ability to offload checksums changes > > then it may be necessary for software checksumming of an skb to occur. > > > > An example of such a case is where a non-GRE packet is received but > > is to be encapsulated and transmitted as GRE. > > > > Another example relates to my proposed support for for GSO packets > > that are non-MPLS when received but MPLS when transmitted. > > > > The cost of this change is that csum may be checked when it previously > > was not. In the case where csum is true this is pure overhead. In the > > case where csum is false it leads to software checksumming, which > > I believe also leads to correct checksums in transmitted packets for > > the cases described above. > > > > This patch relies on the return value of can_checksum_protocol() > > being correct and in turn the return value of skb_network_protocol(), > > used to provide the protocol parameter of can_checksum_protocol(), > > being correct. It also relies on the features passed to skb_segment() > > and in turn to can_checksum_protocol() being correct. > > > > I believe that this problem does has not been observed for VLANs because it > > appears that all drivers except xgbe set vlan_features such that that the > > checksum offload support for VLAN packets is greater than or equal to that > > of non-VLAN packets. > > > > I wonder if the code in xgbe may be an oversight and the hardware does > > support checksumming of VLAN packets. If so it may be worth updating the > > vlan_features of the driver as this patch will force such checksums to be > > performed in software rather than hardware. > > > > Signed-off-by: Simon Horman <horms@verge.net.au> > > > > --- > > > > Tested using the bnx2 driver. > > > > v2 > > Rewrite changlog after further analysis > > --- > > net/core/skbuff.c | 3 ++- > > 1 file changed, 2 insertions(+), 1 deletion(-) > > > > diff --git a/net/core/skbuff.c b/net/core/skbuff.c > > index af9185d..3a43be4 100644 > > --- a/net/core/skbuff.c > > +++ b/net/core/skbuff.c > > @@ -2853,7 +2853,7 @@ struct sk_buff *skb_segment(struct sk_buff *skb, netdev_features_t features) > > doffset + tnl_hlen); > > > > if (fskb != skb_shinfo(skb)->frag_list) > > - continue; > > + goto csum; > > > > if (!sg) { > > nskb->ip_summed = CHECKSUM_NONE; > > @@ -2917,6 +2917,7 @@ skip_fraglist: > > nskb->len += nskb->data_len; > > nskb->truesize += nskb->data_len; > > > > +csum: > > if (!csum) { > > nskb->csum = skb_checksum(nskb, doffset, > > nskb->len - doffset, 0); > > > I had hard time to understand this patch, but I think I now get it. > > Changelog seems to be very specific, while the change looks like a more > generic problem about skbs having a frag_list ? Oh well... > > skb_segment() is probably too complex for my understanding. > > Could you please change the label name ? > > perform_csum_check for example, and maybe it would be easier :) Thanks, I have updated and reposted the patch accordingly. I also update the change log to try and make things clearer. -- To unsubscribe from this list: send the line "unsubscribe netdev" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
diff --git a/net/core/skbuff.c b/net/core/skbuff.c index af9185d..3a43be4 100644 --- a/net/core/skbuff.c +++ b/net/core/skbuff.c @@ -2853,7 +2853,7 @@ struct sk_buff *skb_segment(struct sk_buff *skb, netdev_features_t features) doffset + tnl_hlen); if (fskb != skb_shinfo(skb)->frag_list) - continue; + goto csum; if (!sg) { nskb->ip_summed = CHECKSUM_NONE; @@ -2917,6 +2917,7 @@ skip_fraglist: nskb->len += nskb->data_len; nskb->truesize += nskb->data_len; +csum: if (!csum) { nskb->csum = skb_checksum(nskb, doffset, nskb->len - doffset, 0);
In the case where the ability to offload checksums changes then it may be necessary for software checksumming of an skb to occur. An example of such a case is where a non-GRE packet is received but is to be encapsulated and transmitted as GRE. Another example relates to my proposed support for for GSO packets that are non-MPLS when received but MPLS when transmitted. The cost of this change is that csum may be checked when it previously was not. In the case where csum is true this is pure overhead. In the case where csum is false it leads to software checksumming, which I believe also leads to correct checksums in transmitted packets for the cases described above. This patch relies on the return value of can_checksum_protocol() being correct and in turn the return value of skb_network_protocol(), used to provide the protocol parameter of can_checksum_protocol(), being correct. It also relies on the features passed to skb_segment() and in turn to can_checksum_protocol() being correct. I believe that this problem does has not been observed for VLANs because it appears that all drivers except xgbe set vlan_features such that that the checksum offload support for VLAN packets is greater than or equal to that of non-VLAN packets. I wonder if the code in xgbe may be an oversight and the hardware does support checksumming of VLAN packets. If so it may be worth updating the vlan_features of the driver as this patch will force such checksums to be performed in software rather than hardware. Signed-off-by: Simon Horman <horms@verge.net.au> --- Tested using the bnx2 driver. v2 Rewrite changlog after further analysis --- net/core/skbuff.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)