From patchwork Tue Mar 3 13:07:49 2020 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Paul Blakey X-Patchwork-Id: 1248360 X-Patchwork-Delegate: davem@davemloft.net Return-Path: X-Original-To: patchwork-incoming-netdev@ozlabs.org Delivered-To: patchwork-incoming-netdev@ozlabs.org Authentication-Results: ozlabs.org; spf=none (no SPF record) smtp.mailfrom=vger.kernel.org (client-ip=209.132.180.67; helo=vger.kernel.org; envelope-from=netdev-owner@vger.kernel.org; receiver=) Authentication-Results: ozlabs.org; dmarc=fail (p=none dis=none) header.from=mellanox.com Received: from vger.kernel.org (vger.kernel.org [209.132.180.67]) by ozlabs.org (Postfix) with ESMTP id 48Wy406M9Nz9sPg for ; Wed, 4 Mar 2020 00:08:00 +1100 (AEDT) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1729358AbgCCNIA (ORCPT ); Tue, 3 Mar 2020 08:08:00 -0500 Received: from mail-il-dmz.mellanox.com ([193.47.165.129]:34621 "EHLO mellanox.co.il" rhost-flags-OK-OK-OK-FAIL) by vger.kernel.org with ESMTP id S1729327AbgCCNH5 (ORCPT ); Tue, 3 Mar 2020 08:07:57 -0500 Received: from Internal Mail-Server by MTLPINE2 (envelope-from paulb@mellanox.com) with ESMTPS (AES256-SHA encrypted); 3 Mar 2020 15:07:54 +0200 Received: from reg-r-vrt-019-120.mtr.labs.mlnx (reg-r-vrt-019-120.mtr.labs.mlnx [10.213.19.120]) by labmailer.mlnx (8.13.8/8.13.8) with ESMTP id 023D7rV4032430; Tue, 3 Mar 2020 15:07:53 +0200 From: Paul Blakey To: Paul Blakey , Saeed Mahameed , Oz Shlomo , Jakub Kicinski , Vlad Buslov , David Miller , "netdev@vger.kernel.org" , Jiri Pirko , Roi Dayan Subject: [PATCH net-next v4 1/3] net/sched: act_ct: Create nf flow table per zone Date: Tue, 3 Mar 2020 15:07:49 +0200 Message-Id: <1583240871-27320-2-git-send-email-paulb@mellanox.com> X-Mailer: git-send-email 1.8.4.3 In-Reply-To: <1583240871-27320-1-git-send-email-paulb@mellanox.com> References: <1583240871-27320-1-git-send-email-paulb@mellanox.com> Sender: netdev-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: netdev@vger.kernel.org Use the NF flow tables infrastructure for CT offload. Create a nf flow table per zone. Next patches will add FT entries to this table, and do the software offload. Signed-off-by: Paul Blakey Reviewed-by: Jiri Pirko --- Changelog: v3->v4: Alloc GFP_ATOMIC v2->v3: Ditch re-locking to alloc, and use atomic allocation v1->v2: Use spin_lock_bh instead of spin_lock, and unlock for alloc (as it can sleep) Free ft on last tc act instance instead of last instance + last offloaded tuple, this removes cleanup cb and netfilter patches, and is simpler Removed accidental mlx5/core/en_tc.c change Removed reviewed by Jiri - patch changed include/net/tc_act/tc_ct.h | 2 + net/sched/Kconfig | 2 +- net/sched/act_ct.c | 134 ++++++++++++++++++++++++++++++++++++++++++++- 3 files changed, 136 insertions(+), 2 deletions(-) diff --git a/include/net/tc_act/tc_ct.h b/include/net/tc_act/tc_ct.h index a8b1564..cf3492e 100644 --- a/include/net/tc_act/tc_ct.h +++ b/include/net/tc_act/tc_ct.h @@ -25,6 +25,8 @@ struct tcf_ct_params { u16 ct_action; struct rcu_head rcu; + + struct tcf_ct_flow_table *ct_ft; }; struct tcf_ct { diff --git a/net/sched/Kconfig b/net/sched/Kconfig index edde0e5..bfbefb7 100644 --- a/net/sched/Kconfig +++ b/net/sched/Kconfig @@ -972,7 +972,7 @@ config NET_ACT_TUNNEL_KEY config NET_ACT_CT tristate "connection tracking tc action" - depends on NET_CLS_ACT && NF_CONNTRACK && NF_NAT + depends on NET_CLS_ACT && NF_CONNTRACK && NF_NAT && NF_FLOW_TABLE help Say Y here to allow sending the packets to conntrack module. diff --git a/net/sched/act_ct.c b/net/sched/act_ct.c index f685c0d..3321087 100644 --- a/net/sched/act_ct.c +++ b/net/sched/act_ct.c @@ -15,6 +15,7 @@ #include #include #include +#include #include #include #include @@ -24,6 +25,7 @@ #include #include +#include #include #include #include @@ -31,6 +33,108 @@ #include #include +static struct workqueue_struct *act_ct_wq; +static struct rhashtable zones_ht; +static DEFINE_SPINLOCK(zones_lock); + +struct tcf_ct_flow_table { + struct rhash_head node; /* In zones tables */ + + struct rcu_work rwork; + struct nf_flowtable nf_ft; + u16 zone; + u32 ref; + + bool dying; +}; + +static const struct rhashtable_params zones_params = { + .head_offset = offsetof(struct tcf_ct_flow_table, node), + .key_offset = offsetof(struct tcf_ct_flow_table, zone), + .key_len = sizeof_field(struct tcf_ct_flow_table, zone), + .automatic_shrinking = true, +}; + +static struct nf_flowtable_type flowtable_ct = { + .owner = THIS_MODULE, +}; + +static int tcf_ct_flow_table_get(struct tcf_ct_params *params) +{ + struct tcf_ct_flow_table *ct_ft; + int err = -ENOMEM; + + spin_lock_bh(&zones_lock); + ct_ft = rhashtable_lookup_fast(&zones_ht, ¶ms->zone, zones_params); + if (ct_ft) + goto take_ref; + + ct_ft = kzalloc(sizeof(*ct_ft), GFP_ATOMIC); + if (!ct_ft) + goto err_alloc; + + ct_ft->zone = params->zone; + err = rhashtable_insert_fast(&zones_ht, &ct_ft->node, zones_params); + if (err) + goto err_insert; + + ct_ft->nf_ft.type = &flowtable_ct; + err = nf_flow_table_init(&ct_ft->nf_ft); + if (err) + goto err_init; + + __module_get(THIS_MODULE); +take_ref: + params->ct_ft = ct_ft; + ct_ft->ref++; + spin_unlock_bh(&zones_lock); + + return 0; + +err_init: + rhashtable_remove_fast(&zones_ht, &ct_ft->node, zones_params); +err_insert: + kfree(ct_ft); +err_alloc: + spin_unlock_bh(&zones_lock); + return err; +} + +static void tcf_ct_flow_table_cleanup_work(struct work_struct *work) +{ + struct tcf_ct_flow_table *ct_ft; + + ct_ft = container_of(to_rcu_work(work), struct tcf_ct_flow_table, + rwork); + nf_flow_table_free(&ct_ft->nf_ft); + kfree(ct_ft); + + module_put(THIS_MODULE); +} + +static void tcf_ct_flow_table_put(struct tcf_ct_params *params) +{ + struct tcf_ct_flow_table *ct_ft = params->ct_ft; + + spin_lock_bh(&zones_lock); + if (--params->ct_ft->ref == 0) { + rhashtable_remove_fast(&zones_ht, &ct_ft->node, zones_params); + INIT_RCU_WORK(&ct_ft->rwork, tcf_ct_flow_table_cleanup_work); + queue_rcu_work(act_ct_wq, &ct_ft->rwork); + } + spin_unlock_bh(&zones_lock); +} + +static int tcf_ct_flow_tables_init(void) +{ + return rhashtable_init(&zones_ht, &zones_params); +} + +static void tcf_ct_flow_tables_uninit(void) +{ + rhashtable_destroy(&zones_ht); +} + static struct tc_action_ops act_ct_ops; static unsigned int ct_net_id; @@ -207,6 +311,8 @@ static void tcf_ct_params_free(struct rcu_head *head) struct tcf_ct_params *params = container_of(head, struct tcf_ct_params, rcu); + tcf_ct_flow_table_put(params); + if (params->tmpl) nf_conntrack_put(¶ms->tmpl->ct_general); kfree(params); @@ -730,6 +836,10 @@ static int tcf_ct_init(struct net *net, struct nlattr *nla, if (err) goto cleanup; + err = tcf_ct_flow_table_get(params); + if (err) + goto cleanup; + spin_lock_bh(&c->tcf_lock); goto_ch = tcf_action_set_ctrlact(*a, parm->action, goto_ch); params = rcu_replace_pointer(c->params, params, @@ -974,12 +1084,34 @@ static void __net_exit ct_exit_net(struct list_head *net_list) static int __init ct_init_module(void) { - return tcf_register_action(&act_ct_ops, &ct_net_ops); + int err; + + act_ct_wq = alloc_ordered_workqueue("act_ct_workqueue", 0); + if (!act_ct_wq) + return -ENOMEM; + + err = tcf_ct_flow_tables_init(); + if (err) + goto err_tbl_init; + + err = tcf_register_action(&act_ct_ops, &ct_net_ops); + if (err) + goto err_register; + + return 0; + +err_tbl_init: + destroy_workqueue(act_ct_wq); +err_register: + tcf_ct_flow_tables_uninit(); + return err; } static void __exit ct_cleanup_module(void) { tcf_unregister_action(&act_ct_ops, &ct_net_ops); + tcf_ct_flow_tables_uninit(); + destroy_workqueue(act_ct_wq); } module_init(ct_init_module); From patchwork Tue Mar 3 13:07:51 2020 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Paul Blakey X-Patchwork-Id: 1248359 X-Patchwork-Delegate: davem@davemloft.net Return-Path: X-Original-To: patchwork-incoming-netdev@ozlabs.org Delivered-To: patchwork-incoming-netdev@ozlabs.org Authentication-Results: ozlabs.org; spf=none (no SPF record) smtp.mailfrom=vger.kernel.org (client-ip=209.132.180.67; helo=vger.kernel.org; envelope-from=netdev-owner@vger.kernel.org; receiver=) Authentication-Results: ozlabs.org; dmarc=fail (p=none dis=none) header.from=mellanox.com Received: from vger.kernel.org (vger.kernel.org [209.132.180.67]) by ozlabs.org (Postfix) with ESMTP id 48Wy3z0F4Cz9sRY for ; Wed, 4 Mar 2020 00:07:59 +1100 (AEDT) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1729354AbgCCNH6 (ORCPT ); Tue, 3 Mar 2020 08:07:58 -0500 Received: from mail-il-dmz.mellanox.com ([193.47.165.129]:34623 "EHLO mellanox.co.il" rhost-flags-OK-OK-OK-FAIL) by vger.kernel.org with ESMTP id S1729328AbgCCNH5 (ORCPT ); Tue, 3 Mar 2020 08:07:57 -0500 Received: from Internal Mail-Server by MTLPINE2 (envelope-from paulb@mellanox.com) with ESMTPS (AES256-SHA encrypted); 3 Mar 2020 15:07:54 +0200 Received: from reg-r-vrt-019-120.mtr.labs.mlnx (reg-r-vrt-019-120.mtr.labs.mlnx [10.213.19.120]) by labmailer.mlnx (8.13.8/8.13.8) with ESMTP id 023D7rV6032430; Tue, 3 Mar 2020 15:07:54 +0200 From: Paul Blakey To: Paul Blakey , Saeed Mahameed , Oz Shlomo , Jakub Kicinski , Vlad Buslov , David Miller , "netdev@vger.kernel.org" , Jiri Pirko , Roi Dayan Subject: [PATCH net-next v4 3/3] net/sched: act_ct: Software offload of established flows Date: Tue, 3 Mar 2020 15:07:51 +0200 Message-Id: <1583240871-27320-4-git-send-email-paulb@mellanox.com> X-Mailer: git-send-email 1.8.4.3 In-Reply-To: <1583240871-27320-1-git-send-email-paulb@mellanox.com> References: <1583240871-27320-1-git-send-email-paulb@mellanox.com> Sender: netdev-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: netdev@vger.kernel.org Offload nf conntrack processing by looking up the 5-tuple in the zone's flow table. The nf conntrack module will process the packets until a connection is in established state. Once in established state, the ct state pointer (nf_conn) will be restored on the skb from a successful ft lookup. Signed-off-by: Paul Blakey Acked-by: Jiri Pirko --- Changelog: v1->v2: Add !skip_add curly braces Removed extra setting thoff again Check tcp proto outside of tcf_ct_flow_table_check_tcp net/sched/act_ct.c | 160 ++++++++++++++++++++++++++++++++++++++++++++++++++++- 1 file changed, 158 insertions(+), 2 deletions(-) diff --git a/net/sched/act_ct.c b/net/sched/act_ct.c index 2ab38431..a2d5582 100644 --- a/net/sched/act_ct.c +++ b/net/sched/act_ct.c @@ -186,6 +186,155 @@ static void tcf_ct_flow_table_process_conn(struct tcf_ct_flow_table *ct_ft, tcf_ct_flow_table_add(ct_ft, ct, tcp); } +static bool +tcf_ct_flow_table_fill_tuple_ipv4(struct sk_buff *skb, + struct flow_offload_tuple *tuple) +{ + struct flow_ports *ports; + unsigned int thoff; + struct iphdr *iph; + + if (!pskb_may_pull(skb, sizeof(*iph))) + return false; + + iph = ip_hdr(skb); + thoff = iph->ihl * 4; + + if (ip_is_fragment(iph) || + unlikely(thoff != sizeof(struct iphdr))) + return false; + + if (iph->protocol != IPPROTO_TCP && + iph->protocol != IPPROTO_UDP) + return false; + + if (iph->ttl <= 1) + return false; + + if (!pskb_may_pull(skb, thoff + sizeof(*ports))) + return false; + + ports = (struct flow_ports *)(skb_network_header(skb) + thoff); + + tuple->src_v4.s_addr = iph->saddr; + tuple->dst_v4.s_addr = iph->daddr; + tuple->src_port = ports->source; + tuple->dst_port = ports->dest; + tuple->l3proto = AF_INET; + tuple->l4proto = iph->protocol; + + return true; +} + +static bool +tcf_ct_flow_table_fill_tuple_ipv6(struct sk_buff *skb, + struct flow_offload_tuple *tuple) +{ + struct flow_ports *ports; + struct ipv6hdr *ip6h; + unsigned int thoff; + + if (!pskb_may_pull(skb, sizeof(*ip6h))) + return false; + + ip6h = ipv6_hdr(skb); + + if (ip6h->nexthdr != IPPROTO_TCP && + ip6h->nexthdr != IPPROTO_UDP) + return false; + + if (ip6h->hop_limit <= 1) + return false; + + thoff = sizeof(*ip6h); + if (!pskb_may_pull(skb, thoff + sizeof(*ports))) + return false; + + ports = (struct flow_ports *)(skb_network_header(skb) + thoff); + + tuple->src_v6 = ip6h->saddr; + tuple->dst_v6 = ip6h->daddr; + tuple->src_port = ports->source; + tuple->dst_port = ports->dest; + tuple->l3proto = AF_INET6; + tuple->l4proto = ip6h->nexthdr; + + return true; +} + +static bool tcf_ct_flow_table_check_tcp(struct flow_offload *flow, + struct sk_buff *skb, + unsigned int thoff) +{ + struct tcphdr *tcph; + + if (!pskb_may_pull(skb, thoff + sizeof(*tcph))) + return false; + + tcph = (void *)(skb_network_header(skb) + thoff); + if (unlikely(tcph->fin || tcph->rst)) { + flow_offload_teardown(flow); + return false; + } + + return true; +} + +static bool tcf_ct_flow_table_lookup(struct tcf_ct_params *p, + struct sk_buff *skb, + u8 family) +{ + struct nf_flowtable *nf_ft = &p->ct_ft->nf_ft; + struct flow_offload_tuple_rhash *tuplehash; + struct flow_offload_tuple tuple = {}; + enum ip_conntrack_info ctinfo; + struct flow_offload *flow; + struct nf_conn *ct; + unsigned int thoff; + int ip_proto; + u8 dir; + + /* Previously seen or loopback */ + ct = nf_ct_get(skb, &ctinfo); + if ((ct && !nf_ct_is_template(ct)) || ctinfo == IP_CT_UNTRACKED) + return false; + + switch (family) { + case NFPROTO_IPV4: + if (!tcf_ct_flow_table_fill_tuple_ipv4(skb, &tuple)) + return false; + break; + case NFPROTO_IPV6: + if (!tcf_ct_flow_table_fill_tuple_ipv6(skb, &tuple)) + return false; + break; + default: + return false; + } + + tuplehash = flow_offload_lookup(nf_ft, &tuple); + if (!tuplehash) + return false; + + dir = tuplehash->tuple.dir; + flow = container_of(tuplehash, struct flow_offload, tuplehash[dir]); + ct = flow->ct; + + ctinfo = dir == FLOW_OFFLOAD_DIR_ORIGINAL ? IP_CT_ESTABLISHED : + IP_CT_ESTABLISHED_REPLY; + + thoff = ip_hdr(skb)->ihl * 4; + ip_proto = ip_hdr(skb)->protocol; + if (ip_proto == IPPROTO_TCP && + !tcf_ct_flow_table_check_tcp(flow, skb, thoff)) + return false; + + nf_conntrack_get(&ct->ct_general); + nf_ct_set(skb, ct, ctinfo); + + return true; +} + static int tcf_ct_flow_tables_init(void) { return rhashtable_init(&zones_ht, &zones_params); @@ -554,6 +703,7 @@ static int tcf_ct_act(struct sk_buff *skb, const struct tc_action *a, struct nf_hook_state state; int nh_ofs, err, retval; struct tcf_ct_params *p; + bool skip_add = false; struct nf_conn *ct; u8 family; @@ -603,6 +753,11 @@ static int tcf_ct_act(struct sk_buff *skb, const struct tc_action *a, */ cached = tcf_ct_skb_nfct_cached(net, skb, p->zone, force); if (!cached) { + if (!commit && tcf_ct_flow_table_lookup(p, skb, family)) { + skip_add = true; + goto do_nat; + } + /* Associate skb with specified zone. */ if (tmpl) { ct = nf_ct_get(skb, &ctinfo); @@ -620,6 +775,7 @@ static int tcf_ct_act(struct sk_buff *skb, const struct tc_action *a, goto out_push; } +do_nat: ct = nf_ct_get(skb, &ctinfo); if (!ct) goto out_push; @@ -637,10 +793,10 @@ static int tcf_ct_act(struct sk_buff *skb, const struct tc_action *a, * even if the connection is already confirmed. */ nf_conntrack_confirm(skb); + } else if (!skip_add) { + tcf_ct_flow_table_process_conn(p->ct_ft, ct, ctinfo); } - tcf_ct_flow_table_process_conn(p->ct_ft, ct, ctinfo); - out_push: skb_push_rcsum(skb, nh_ofs);