Message ID | 1583676662-15180-6-git-send-email-paulb@mellanox.com |
---|---|
State | Changes Requested |
Delegated to: | David Miller |
Headers | show |
Series | Introduce connection tracking offload | expand |
On 08/03/2020 14:10, Paul Blakey wrote: > Pass the zone's flow table instance on the flow action to the drivers. > Thus, allowing drivers to register FT add/del/stats callbacks. > > Finally, enable hardware offload on the flow table instance. > > Signed-off-by: Paul Blakey <paulb@mellanox.com> > Reviewed-by: Jiri Pirko <jiri@mellanox.com> > --- > <snip> > diff --git a/net/sched/act_ct.c b/net/sched/act_ct.c > index 84d5abf..d52185d 100644 > --- a/net/sched/act_ct.c > +++ b/net/sched/act_ct.c > @@ -292,6 +292,7 @@ static int tcf_ct_flow_table_get(struct tcf_ct_params *params) > goto err_insert; > > ct_ft->nf_ft.type = &flowtable_ct; > + ct_ft->nf_ft.flags |= NF_FLOWTABLE_HW_OFFLOAD; > err = nf_flow_table_init(&ct_ft->nf_ft); > if (err) > goto err_init; > @@ -299,6 +300,7 @@ static int tcf_ct_flow_table_get(struct tcf_ct_params *params) > __module_get(THIS_MODULE); > take_ref: > params->ct_ft = ct_ft; > + params->nf_ft = &ct_ft->nf_ft; > ct_ft->ref++; > spin_unlock_bh(&zones_lock); This doesn't seem to apply to net-next (34a568a244be); the label after the __module_get() is 'out_unlock', not 'take_ref'. Is there a missing prerequisite patch? Or am I just failing to drive 'git am' correctly? -ed
On Mon, Mar 09, 2020 at 09:25:49PM +0000, Edward Cree wrote: > On 08/03/2020 14:10, Paul Blakey wrote: > > Pass the zone's flow table instance on the flow action to the drivers. > > Thus, allowing drivers to register FT add/del/stats callbacks. > > > > Finally, enable hardware offload on the flow table instance. > > > > Signed-off-by: Paul Blakey <paulb@mellanox.com> > > Reviewed-by: Jiri Pirko <jiri@mellanox.com> > > --- > > <snip> > > diff --git a/net/sched/act_ct.c b/net/sched/act_ct.c > > index 84d5abf..d52185d 100644 > > --- a/net/sched/act_ct.c > > +++ b/net/sched/act_ct.c > > @@ -292,6 +292,7 @@ static int tcf_ct_flow_table_get(struct tcf_ct_params *params) > > goto err_insert; > > > > ct_ft->nf_ft.type = &flowtable_ct; > > + ct_ft->nf_ft.flags |= NF_FLOWTABLE_HW_OFFLOAD; > > err = nf_flow_table_init(&ct_ft->nf_ft); > > if (err) > > goto err_init; > > @@ -299,6 +300,7 @@ static int tcf_ct_flow_table_get(struct tcf_ct_params *params) > > __module_get(THIS_MODULE); > > take_ref: > > params->ct_ft = ct_ft; > > + params->nf_ft = &ct_ft->nf_ft; > > ct_ft->ref++; > > spin_unlock_bh(&zones_lock); > This doesn't seem to apply to net-next (34a568a244be); the label after > the __module_get() is 'out_unlock', not 'take_ref'. Is there a missing > prerequisite patch? Or am I just failing to drive 'git am' correctly? That's a mid-air collision with Eric's [PATCH net-next] net/sched: act_ct: fix lockdep splat in tcf_ct_flow_table_get That went in in between v1 and v2 here. Marcelo
On 09/03/2020 22:19, Marcelo Ricardo Leitner wrote: > On Mon, Mar 09, 2020 at 09:25:49PM +0000, Edward Cree wrote: >> This doesn't seem to apply to net-next (34a568a244be); the label after >> the __module_get() is 'out_unlock', not 'take_ref'. Is there a missing >> prerequisite patch? Or am I just failing to drive 'git am' correctly? > That's a mid-air collision with Eric's > [PATCH net-next] net/sched: act_ct: fix lockdep splat in tcf_ct_flow_table_get > That went in in between v1 and v2 here. > > Marcelo Thanks. Unfortunately, although going back to that commit's parentmakes this patch apply, #6 still doesn't, and I don't particularly feel like playing whack-a-mole. Paul, could you either specify the base from which you generated this series, or stick the branch somewhere that can be pulled from, please? I'd like to build it so that I can start experimenting with driver code to interface with it. TiA, -ed
On 3/9/2020 11:25 PM, Edward Cree wrote: > On 08/03/2020 14:10, Paul Blakey wrote: >> Pass the zone's flow table instance on the flow action to the drivers. >> Thus, allowing drivers to register FT add/del/stats callbacks. >> >> Finally, enable hardware offload on the flow table instance. >> >> Signed-off-by: Paul Blakey <paulb@mellanox.com> >> Reviewed-by: Jiri Pirko <jiri@mellanox.com> >> --- >> <snip> >> diff --git a/net/sched/act_ct.c b/net/sched/act_ct.c >> index 84d5abf..d52185d 100644 >> --- a/net/sched/act_ct.c >> +++ b/net/sched/act_ct.c >> @@ -292,6 +292,7 @@ static int tcf_ct_flow_table_get(struct tcf_ct_params *params) >> goto err_insert; >> >> ct_ft->nf_ft.type = &flowtable_ct; >> + ct_ft->nf_ft.flags |= NF_FLOWTABLE_HW_OFFLOAD; >> err = nf_flow_table_init(&ct_ft->nf_ft); >> if (err) >> goto err_init; >> @@ -299,6 +300,7 @@ static int tcf_ct_flow_table_get(struct tcf_ct_params *params) >> __module_get(THIS_MODULE); >> take_ref: >> params->ct_ft = ct_ft; >> + params->nf_ft = &ct_ft->nf_ft; >> ct_ft->ref++; >> spin_unlock_bh(&zones_lock); > This doesn't seem to apply to net-next (34a568a244be); the label after > the __module_get() is 'out_unlock', not 'take_ref'. Is there a missing > prerequisite patch? Or am I just failing to drive 'git am' correctly? It's was rebased only on top of saeeds ct-offload, will rebase it on net next as well. > > -ed
diff --git a/include/net/flow_offload.h b/include/net/flow_offload.h index fc79c8c..af317c4 100644 --- a/include/net/flow_offload.h +++ b/include/net/flow_offload.h @@ -197,6 +197,7 @@ struct flow_action_entry { struct { /* FLOW_ACTION_CT */ int action; u16 zone; + struct nf_flowtable *flow_table; } ct; struct { unsigned long cookie; diff --git a/include/net/tc_act/tc_ct.h b/include/net/tc_act/tc_ct.h index 735da59..79654bc 100644 --- a/include/net/tc_act/tc_ct.h +++ b/include/net/tc_act/tc_ct.h @@ -27,6 +27,7 @@ struct tcf_ct_params { struct rcu_head rcu; struct tcf_ct_flow_table *ct_ft; + struct nf_flowtable *nf_ft; }; struct tcf_ct { @@ -50,9 +51,18 @@ static inline int tcf_ct_action(const struct tc_action *a) return to_ct_params(a)->ct_action; } +static inline struct nf_flowtable *tcf_ct_ft(const struct tc_action *a) +{ + return to_ct_params(a)->nf_ft; +} + #else static inline uint16_t tcf_ct_zone(const struct tc_action *a) { return 0; } static inline int tcf_ct_action(const struct tc_action *a) { return 0; } +static inline struct nf_flowtable *tcf_ct_ft(const struct tc_action *a) +{ + return NULL; +} #endif /* CONFIG_NF_CONNTRACK */ #if IS_ENABLED(CONFIG_NET_ACT_CT) diff --git a/net/sched/act_ct.c b/net/sched/act_ct.c index 84d5abf..d52185d 100644 --- a/net/sched/act_ct.c +++ b/net/sched/act_ct.c @@ -292,6 +292,7 @@ static int tcf_ct_flow_table_get(struct tcf_ct_params *params) goto err_insert; ct_ft->nf_ft.type = &flowtable_ct; + ct_ft->nf_ft.flags |= NF_FLOWTABLE_HW_OFFLOAD; err = nf_flow_table_init(&ct_ft->nf_ft); if (err) goto err_init; @@ -299,6 +300,7 @@ static int tcf_ct_flow_table_get(struct tcf_ct_params *params) __module_get(THIS_MODULE); take_ref: params->ct_ft = ct_ft; + params->nf_ft = &ct_ft->nf_ft; ct_ft->ref++; spin_unlock_bh(&zones_lock); diff --git a/net/sched/cls_api.c b/net/sched/cls_api.c index e604ebe..67221fc 100644 --- a/net/sched/cls_api.c +++ b/net/sched/cls_api.c @@ -3600,6 +3600,7 @@ int tc_setup_flow_action(struct flow_action *flow_action, entry->id = FLOW_ACTION_CT; entry->ct.action = tcf_ct_action(act); entry->ct.zone = tcf_ct_zone(act); + entry->ct.flow_table = tcf_ct_ft(act); } else if (is_tcf_mpls(act)) { switch (tcf_mpls_action(act)) { case TCA_MPLS_ACT_PUSH: