Message ID | 1457016960-27832-2-git-send-email-amir@vadai.me |
---|---|
State | Changes Requested, archived |
Delegated to: | David Miller |
Headers | show |
Thu, Mar 03, 2016 at 03:55:51PM CET, amir@vadai.me wrote: >This patch is based on a patch made by John Fastabend. >It adds support for offloading cls_flower. >when NETIF_F_HW_TC is on: > flags = 0 => Rule will be processed twice - by hardware, and if > still relevant, by software. > flags = SKIP_HW => Rull will be processed by software only > >If hardare fail/not capabale to apply the rule, operation will fail. > >Suggested-by: John Fastabend <john.r.fastabend@intel.com> >Signed-off-by: Amir Vadai <amir@vadai.me> Acked-by: Jiri Pirko <jiri@mellanox.com>
From: Amir Vadai <amir@vadai.me> Date: Thu, 3 Mar 2016 16:55:51 +0200 > @@ -454,11 +501,13 @@ static int fl_change(struct net *net, struct sk_buff *in_skb, > u32 handle, struct nlattr **tca, > unsigned long *arg, bool ovr) > { > + struct net_device *dev = tp->q->dev_queue->dev; > struct cls_fl_head *head = rtnl_dereference(tp->root); This variable is not used. And the compiler warns about this, and because of this I am pretty sure you aren't looking at the compiler output while testing your builds which is a big no-no.
On Thu, Mar 03, 2016 at 12:30:33PM -0500, David Miller wrote: > From: Amir Vadai <amir@vadai.me> > Date: Thu, 3 Mar 2016 16:55:51 +0200 > > > @@ -454,11 +501,13 @@ static int fl_change(struct net *net, struct sk_buff *in_skb, > > u32 handle, struct nlattr **tca, > > unsigned long *arg, bool ovr) > > { > > + struct net_device *dev = tp->q->dev_queue->dev; > > struct cls_fl_head *head = rtnl_dereference(tp->root); > > This variable is not used. > > And the compiler warns about this, and because of this I am pretty sure you > aren't looking at the compiler output while testing your builds which is a > big no-no. My bad. I did a last minute change that left this variable and somehow missed the warning (though I did compile and test it). Will fix for v3
On 16-03-03 06:55 AM, Amir Vadai wrote: > This patch is based on a patch made by John Fastabend. > It adds support for offloading cls_flower. > when NETIF_F_HW_TC is on: > flags = 0 => Rule will be processed twice - by hardware, and if > still relevant, by software. > flags = SKIP_HW => Rull will be processed by software only > > If hardare fail/not capabale to apply the rule, operation will fail. > > Suggested-by: John Fastabend <john.r.fastabend@intel.com> > Signed-off-by: Amir Vadai <amir@vadai.me> > --- [...] > static bool fl_destroy(struct tcf_proto *tp, bool force) > { > struct cls_fl_head *head = rtnl_dereference(tp->root); > @@ -174,6 +220,7 @@ static bool fl_destroy(struct tcf_proto *tp, bool force) > return false; > > list_for_each_entry_safe(f, next, &head->filters, list) { > + fl_hw_destroy_filter(tp, (u64)f); > list_del_rcu(&f->list); > call_rcu(&f->rcu, fl_destroy_filter); > } > @@ -454,11 +501,13 @@ static int fl_change(struct net *net, struct sk_buff *in_skb, > u32 handle, struct nlattr **tca, > unsigned long *arg, bool ovr) > { > + struct net_device *dev = tp->q->dev_queue->dev; > struct cls_fl_head *head = rtnl_dereference(tp->root); > struct cls_fl_filter *fold = (struct cls_fl_filter *) *arg; > struct cls_fl_filter *fnew; > struct nlattr *tb[TCA_FLOWER_MAX + 1]; > struct fl_flow_mask mask = {}; > + u32 flags = 0; > int err; > > if (!tca[TCA_OPTIONS]) > @@ -486,6 +535,9 @@ static int fl_change(struct net *net, struct sk_buff *in_skb, > } > fnew->handle = handle; > > + if (tb[TCA_FLOWER_FLAGS]) > + flags = nla_get_u32(tb[TCA_FLOWER_FLAGS]); > + > err = fl_set_parms(net, tp, fnew, &mask, base, tb, tca[TCA_RATE], ovr); > if (err) > goto errout; > @@ -498,9 +550,22 @@ static int fl_change(struct net *net, struct sk_buff *in_skb, > head->ht_params); > if (err) > goto errout; > - if (fold) > + > + err = fl_hw_replace_filter(tp, > + &head->dissector, > + &mask.key, > + &fnew->key, > + &fnew->exts, > + (u64)fnew, > + flags); > + if (err) > + goto err_hash_remove; > + This behaviour is different than how I did u32 in the u32 case I just let the software case get loaded and do not throw any errors. The intent was if we required a HW entry we would explicitly state that with the SKIP_SW (to be implemented) flag. This error path seems to block the software filter when the hardware fails. I think it would be best to do the same as u32 here and use the error path only if SKIP_SW is set. Or if you really want an error path on SW/HW loads then use another bit in the flag to specify STRICT or something along those lines. > + if (fold) { > rhashtable_remove_fast(&head->ht, &fold->ht_node, > head->ht_params); > + fl_hw_destroy_filter(tp, (u64)fold); > + } > > *arg = (unsigned long) fnew; > > @@ -514,6 +579,9 @@ static int fl_change(struct net *net, struct sk_buff *in_skb, > > return 0; > > +err_hash_remove: > + rhashtable_remove_fast(&head->ht, &fnew->ht_node, head->ht_params); > + > errout: > kfree(fnew); > return err; > @@ -527,6 +595,7 @@ static int fl_delete(struct tcf_proto *tp, unsigned long arg) > rhashtable_remove_fast(&head->ht, &f->ht_node, > head->ht_params); > list_del_rcu(&f->list); > + fl_hw_destroy_filter(tp, (u64)f); > tcf_unbind_filter(tp, &f->res); > call_rcu(&f->rcu, fl_destroy_filter); > return 0; >
On Fri, Mar 04, 2016 at 09:01:39AM -0800, John Fastabend wrote: > On 16-03-03 06:55 AM, Amir Vadai wrote: > > This patch is based on a patch made by John Fastabend. > > It adds support for offloading cls_flower. > > when NETIF_F_HW_TC is on: > > flags = 0 => Rule will be processed twice - by hardware, and if > > still relevant, by software. > > flags = SKIP_HW => Rull will be processed by software only > > > > If hardare fail/not capabale to apply the rule, operation will fail. > > > > Suggested-by: John Fastabend <john.r.fastabend@intel.com> > > Signed-off-by: Amir Vadai <amir@vadai.me> > > --- > > [...] > > > static bool fl_destroy(struct tcf_proto *tp, bool force) > > { > > struct cls_fl_head *head = rtnl_dereference(tp->root); > > @@ -174,6 +220,7 @@ static bool fl_destroy(struct tcf_proto *tp, bool force) > > return false; > > > > list_for_each_entry_safe(f, next, &head->filters, list) { > > + fl_hw_destroy_filter(tp, (u64)f); > > list_del_rcu(&f->list); > > call_rcu(&f->rcu, fl_destroy_filter); > > } > > @@ -454,11 +501,13 @@ static int fl_change(struct net *net, struct sk_buff *in_skb, > > u32 handle, struct nlattr **tca, > > unsigned long *arg, bool ovr) > > { > > + struct net_device *dev = tp->q->dev_queue->dev; > > struct cls_fl_head *head = rtnl_dereference(tp->root); > > struct cls_fl_filter *fold = (struct cls_fl_filter *) *arg; > > struct cls_fl_filter *fnew; > > struct nlattr *tb[TCA_FLOWER_MAX + 1]; > > struct fl_flow_mask mask = {}; > > + u32 flags = 0; > > int err; > > > > if (!tca[TCA_OPTIONS]) > > @@ -486,6 +535,9 @@ static int fl_change(struct net *net, struct sk_buff *in_skb, > > } > > fnew->handle = handle; > > > > + if (tb[TCA_FLOWER_FLAGS]) > > + flags = nla_get_u32(tb[TCA_FLOWER_FLAGS]); > > + > > err = fl_set_parms(net, tp, fnew, &mask, base, tb, tca[TCA_RATE], ovr); > > if (err) > > goto errout; > > @@ -498,9 +550,22 @@ static int fl_change(struct net *net, struct sk_buff *in_skb, > > head->ht_params); > > if (err) > > goto errout; > > - if (fold) > > + > > + err = fl_hw_replace_filter(tp, > > + &head->dissector, > > + &mask.key, > > + &fnew->key, > > + &fnew->exts, > > + (u64)fnew, > > + flags); > > + if (err) > > + goto err_hash_remove; > > + > > This behaviour is different than how I did u32 in the u32 case I just > let the software case get loaded and do not throw any errors. The > intent was if we required a HW entry we would explicitly state that > with the SKIP_SW (to be implemented) flag. This error path seems > to block the software filter when the hardware fails. Makes sense. > > I think it would be best to do the same as u32 here and use the error > path only if SKIP_SW is set. Or if you really want an error path on > SW/HW loads then use another bit in the flag to specify STRICT or > something along those lines. I will do the same as u32. I won't add this STRICT flag, because I don't have any use case for this mode in which processing is done in both SW and HW. > > > > + if (fold) { > > rhashtable_remove_fast(&head->ht, &fold->ht_node, > > head->ht_params); > > + fl_hw_destroy_filter(tp, (u64)fold); > > + } > > > > *arg = (unsigned long) fnew; > > > > @@ -514,6 +579,9 @@ static int fl_change(struct net *net, struct sk_buff *in_skb, > > > > return 0; > > > > +err_hash_remove: > > + rhashtable_remove_fast(&head->ht, &fnew->ht_node, head->ht_params); > > + > > errout: > > kfree(fnew); > > return err; > > @@ -527,6 +595,7 @@ static int fl_delete(struct tcf_proto *tp, unsigned long arg) > > rhashtable_remove_fast(&head->ht, &f->ht_node, > > head->ht_params); > > list_del_rcu(&f->list); > > + fl_hw_destroy_filter(tp, (u64)f); > > tcf_unbind_filter(tp, &f->res); > > call_rcu(&f->rcu, fl_destroy_filter); > > return 0; > > >
diff --git a/include/linux/netdevice.h b/include/linux/netdevice.h index efe7cec..12db9d6 100644 --- a/include/linux/netdevice.h +++ b/include/linux/netdevice.h @@ -785,6 +785,7 @@ typedef u16 (*select_queue_fallback_t)(struct net_device *dev, enum { TC_SETUP_MQPRIO, TC_SETUP_CLSU32, + TC_SETUP_CLSFLOWER, }; struct tc_cls_u32_offload; @@ -794,6 +795,7 @@ struct tc_to_netdev { union { u8 tc; struct tc_cls_u32_offload *cls_u32; + struct tc_cls_flower_offload *cls_flower; }; }; diff --git a/include/net/pkt_cls.h b/include/net/pkt_cls.h index bea14ee..5b4e8f0 100644 --- a/include/net/pkt_cls.h +++ b/include/net/pkt_cls.h @@ -409,4 +409,18 @@ static inline bool tc_should_offload(struct net_device *dev, u32 flags) return true; } +enum tc_fl_command { + TC_CLSFLOWER_REPLACE, + TC_CLSFLOWER_DESTROY, +}; + +struct tc_cls_flower_offload { + enum tc_fl_command command; + u64 cookie; + struct flow_dissector *dissector; + struct fl_flow_key *mask; + struct fl_flow_key *key; + struct tcf_exts *exts; +}; + #endif diff --git a/include/uapi/linux/pkt_cls.h b/include/uapi/linux/pkt_cls.h index 9874f568..c43c5f7 100644 --- a/include/uapi/linux/pkt_cls.h +++ b/include/uapi/linux/pkt_cls.h @@ -417,6 +417,8 @@ enum { TCA_FLOWER_KEY_TCP_DST, /* be16 */ TCA_FLOWER_KEY_UDP_SRC, /* be16 */ TCA_FLOWER_KEY_UDP_DST, /* be16 */ + + TCA_FLOWER_FLAGS, __TCA_FLOWER_MAX, }; diff --git a/net/sched/cls_flower.c b/net/sched/cls_flower.c index 95b0212..ed3cd5a 100644 --- a/net/sched/cls_flower.c +++ b/net/sched/cls_flower.c @@ -165,6 +165,52 @@ static void fl_destroy_filter(struct rcu_head *head) kfree(f); } +static void fl_hw_destroy_filter(struct tcf_proto *tp, u64 cookie) +{ + struct net_device *dev = tp->q->dev_queue->dev; + struct tc_cls_flower_offload offload = {0}; + struct tc_to_netdev tc; + + if (!tc_should_offload(dev, 0)) + return; + + offload.command = TC_CLSFLOWER_DESTROY; + offload.cookie = cookie; + + tc.type = TC_SETUP_CLSFLOWER; + tc.cls_flower = &offload; + + dev->netdev_ops->ndo_setup_tc(dev, tp->q->handle, tp->protocol, &tc); +} + +static int fl_hw_replace_filter(struct tcf_proto *tp, + struct flow_dissector *dissector, + struct fl_flow_key *mask, + struct fl_flow_key *key, + struct tcf_exts *actions, + u64 cookie, u32 flags) +{ + struct net_device *dev = tp->q->dev_queue->dev; + struct tc_cls_flower_offload offload = {0}; + struct tc_to_netdev tc; + + if (!tc_should_offload(dev, flags)) + return 0; + + offload.command = TC_CLSFLOWER_REPLACE; + offload.cookie = cookie; + offload.dissector = dissector; + offload.mask = mask; + offload.key = key; + offload.exts = actions; + + tc.type = TC_SETUP_CLSFLOWER; + tc.cls_flower = &offload; + + return dev->netdev_ops->ndo_setup_tc(dev, tp->q->handle, tp->protocol, + &tc); +} + static bool fl_destroy(struct tcf_proto *tp, bool force) { struct cls_fl_head *head = rtnl_dereference(tp->root); @@ -174,6 +220,7 @@ static bool fl_destroy(struct tcf_proto *tp, bool force) return false; list_for_each_entry_safe(f, next, &head->filters, list) { + fl_hw_destroy_filter(tp, (u64)f); list_del_rcu(&f->list); call_rcu(&f->rcu, fl_destroy_filter); } @@ -454,11 +501,13 @@ static int fl_change(struct net *net, struct sk_buff *in_skb, u32 handle, struct nlattr **tca, unsigned long *arg, bool ovr) { + struct net_device *dev = tp->q->dev_queue->dev; struct cls_fl_head *head = rtnl_dereference(tp->root); struct cls_fl_filter *fold = (struct cls_fl_filter *) *arg; struct cls_fl_filter *fnew; struct nlattr *tb[TCA_FLOWER_MAX + 1]; struct fl_flow_mask mask = {}; + u32 flags = 0; int err; if (!tca[TCA_OPTIONS]) @@ -486,6 +535,9 @@ static int fl_change(struct net *net, struct sk_buff *in_skb, } fnew->handle = handle; + if (tb[TCA_FLOWER_FLAGS]) + flags = nla_get_u32(tb[TCA_FLOWER_FLAGS]); + err = fl_set_parms(net, tp, fnew, &mask, base, tb, tca[TCA_RATE], ovr); if (err) goto errout; @@ -498,9 +550,22 @@ static int fl_change(struct net *net, struct sk_buff *in_skb, head->ht_params); if (err) goto errout; - if (fold) + + err = fl_hw_replace_filter(tp, + &head->dissector, + &mask.key, + &fnew->key, + &fnew->exts, + (u64)fnew, + flags); + if (err) + goto err_hash_remove; + + if (fold) { rhashtable_remove_fast(&head->ht, &fold->ht_node, head->ht_params); + fl_hw_destroy_filter(tp, (u64)fold); + } *arg = (unsigned long) fnew; @@ -514,6 +579,9 @@ static int fl_change(struct net *net, struct sk_buff *in_skb, return 0; +err_hash_remove: + rhashtable_remove_fast(&head->ht, &fnew->ht_node, head->ht_params); + errout: kfree(fnew); return err; @@ -527,6 +595,7 @@ static int fl_delete(struct tcf_proto *tp, unsigned long arg) rhashtable_remove_fast(&head->ht, &f->ht_node, head->ht_params); list_del_rcu(&f->list); + fl_hw_destroy_filter(tp, (u64)f); tcf_unbind_filter(tp, &f->res); call_rcu(&f->rcu, fl_destroy_filter); return 0;
This patch is based on a patch made by John Fastabend. It adds support for offloading cls_flower. when NETIF_F_HW_TC is on: flags = 0 => Rule will be processed twice - by hardware, and if still relevant, by software. flags = SKIP_HW => Rull will be processed by software only If hardare fail/not capabale to apply the rule, operation will fail. Suggested-by: John Fastabend <john.r.fastabend@intel.com> Signed-off-by: Amir Vadai <amir@vadai.me> --- include/linux/netdevice.h | 2 ++ include/net/pkt_cls.h | 14 +++++++++ include/uapi/linux/pkt_cls.h | 2 ++ net/sched/cls_flower.c | 71 +++++++++++++++++++++++++++++++++++++++++++- 4 files changed, 88 insertions(+), 1 deletion(-)