On Wed, 2018-07-25 at 15:52 +0200, Jiri Pirko wrote:
> Tue, Jul 24, 2018 at 10:06:43PM CEST, pab...@redhat.com wrote:
> > When mirred is invoked from the ingress path, and it wants to redirect
> > the processed packet, it can now use the TC_ACT_REINJECT action,
> > filling the tcf_result accordingly, and avoiding a per packet
> > skb_clone().
> >
> > Overall this gives a ~10% improvement in forwarding performance for the
> > TC S/W data path and TC S/W performances are now comparable to the
> > kernel openvswitch datapath.
> >
> > v1 -> v2: use ACT_MIRRED instead of ACT_REDIRECT
> > v2 -> v3: updated after action rename, fixed a typo into the commit
> > message
> >
> > Signed-off-by: Paolo Abeni <pab...@redhat.com>
> > ---
> > net/sched/act_mirred.c | 34 ++++++++++++++++++++++++----------
> > 1 file changed, 24 insertions(+), 10 deletions(-)
> >
> > diff --git a/net/sched/act_mirred.c b/net/sched/act_mirred.c
> > index eeb335f03102..368187312136 100644
> > --- a/net/sched/act_mirred.c
> > +++ b/net/sched/act_mirred.c
> > @@ -25,6 +25,7 @@
> > #include <net/net_namespace.h>
> > #include <net/netlink.h>
> > #include <net/pkt_sched.h>
> > +#include <net/pkt_cls.h>
> > #include <linux/tc_act/tc_mirred.h>
> > #include <net/tc_act/tc_mirred.h>
> >
> > @@ -171,10 +172,12 @@ static int tcf_mirred(struct sk_buff *skb, const
> > struct tc_action *a,
> > struct tcf_result *res)
> > {
> > struct tcf_mirred *m = to_mirred(a);
> > + struct sk_buff *skb2 = skb;
> > bool m_mac_header_xmit;
> > struct net_device *dev;
> > - struct sk_buff *skb2;
> > int retval, err = 0;
> > + bool want_ingress;
> > + bool is_redirect;
> > int m_eaction;
> > int mac_len;
> >
> > @@ -196,16 +199,19 @@ static int tcf_mirred(struct sk_buff *skb, const
> > struct tc_action *a,
> > goto out;
> > }
> >
> > - skb2 = skb_clone(skb, GFP_ATOMIC);
> > - if (!skb2)
> > - goto out;
> > + is_redirect = tcf_mirred_is_act_redirect(m_eaction);
> > + if (!skb_at_tc_ingress(skb) || !is_redirect) {
> > + skb2 = skb_clone(skb, GFP_ATOMIC);
> > + if (!skb2)
> > + goto out;
> > + }
> >
> > /* If action's target direction differs than filter's direction,
> > * and devices expect a mac header on xmit, then mac push/pull is
> > * needed.
> > */
> > - if (skb_at_tc_ingress(skb) != tcf_mirred_act_wants_ingress(m_eaction) &&
> > - m_mac_header_xmit) {
> > + want_ingress = tcf_mirred_act_wants_ingress(m_eaction);
> > + if (skb_at_tc_ingress(skb) != want_ingress && m_mac_header_xmit) {
> > if (!skb_at_tc_ingress(skb)) {
> > /* caught at egress, act ingress: pull mac */
> > mac_len = skb_network_header(skb) - skb_mac_header(skb);
> > @@ -216,15 +222,23 @@ static int tcf_mirred(struct sk_buff *skb, const
> > struct tc_action *a,
> > }
> > }
> >
> > + skb2->skb_iif = skb->dev->ifindex;
> > + skb2->dev = dev;
> > +
> > /* mirror is always swallowed */
> > - if (tcf_mirred_is_act_redirect(m_eaction)) {
> > + if (is_redirect) {
> > skb2->tc_redirected = 1;
> > skb2->tc_from_ingress = skb2->tc_at_ingress;
> > +
> > + /* let's the caller reinject the packet, if possible */
> > + if (skb_at_tc_ingress(skb)) {
>
> I probably missed something. Why only on ingress?
To keep the implementation as simple as possible: if I read correctly,
it is impossible for a filter detect if called by the clsact or the dev
root qdisc, and I think we could safely avoid the skb clone with a not
invasive patch, only if called from the clsact.
[please let me know if the above is somewhat clear ;)]
Also this covers nicely the relevant use case (TC S/W datapath).
Thanks,
Paolo
