net/sched: cls_api: Support hardware miss to tc action

For drivers to support partial offload of a filter's action list,
add support for action miss to specify an action instance to
continue from in sw.

CT action in particular can't be fully offloaded, as new connections
need to be handled in software. This imposes other limitations on
the actions that can be offloaded together with the CT action, such
as packet modifications.

Assign each action on a filter's action list a unique miss_cookie
which drivers can then use to fill action_miss part of the tc skb
extension. On getting back this miss_cookie, find the action
instance with relevant cookie and continue classifying from there.

Signed-off-by: Paul Blakey <paulb@nvidia.com>
Reviewed-by: Jiri Pirko <jiri@nvidia.com>
Reviewed-by: Simon Horman <simon.horman@corigine.com>
Reviewed-by: Marcelo Ricardo Leitner <marcelo.leitner@gmail.com>
Acked-by: Jamal Hadi Salim <jhs@mojatatu.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

2 files changed, 206 insertions, 11 deletions

diff --git a/net/sched/act_api.c b/net/sched/act_api.c
index e67ebc939901..fce522886099 100644
--- a/net/sched/act_api.c
+++ b/net/sched/act_api.c
@@ -268,7 +268,7 @@ static int tcf_action_offload_add_ex(struct tc_action *action,
 	if (err)
 		goto fl_err;
 
-	err = tc_setup_action(&fl_action->action, actions, extack);
+	err = tc_setup_action(&fl_action->action, actions, 0, extack);
 	if (err) {
 		NL_SET_ERR_MSG_MOD(extack,
 				   "Failed to setup tc actions for offload");
diff --git a/net/sched/cls_api.c b/net/sched/cls_api.c
index 656049ead8bb..3569e2c3660c 100644
--- a/net/sched/cls_api.c
+++ b/net/sched/cls_api.c
@@ -22,6 +22,7 @@
 #include <linux/idr.h>
 #include <linux/jhash.h>
 #include <linux/rculist.h>
+#include <linux/rhashtable.h>
 #include <net/net_namespace.h>
 #include <net/sock.h>
 #include <net/netlink.h>
@@ -50,6 +51,109 @@ static LIST_HEAD(tcf_proto_base);
 /* Protects list of registered TC modules. It is pure SMP lock. */
 static DEFINE_RWLOCK(cls_mod_lock);
 
+static struct xarray tcf_exts_miss_cookies_xa;
+struct tcf_exts_miss_cookie_node {
+	const struct tcf_chain *chain;
+	const struct tcf_proto *tp;
+	const struct tcf_exts *exts;
+	u32 chain_index;
+	u32 tp_prio;
+	u32 handle;
+	u32 miss_cookie_base;
+	struct rcu_head rcu;
+};
+
+/* Each tc action entry cookie will be comprised of 32bit miss_cookie_base +
+ * action index in the exts tc actions array.
+ */
+union tcf_exts_miss_cookie {
+	struct {
+		u32 miss_cookie_base;
+		u32 act_index;
+	};
+	u64 miss_cookie;
+};
+
+#if IS_ENABLED(CONFIG_NET_TC_SKB_EXT)
+static int
+tcf_exts_miss_cookie_base_alloc(struct tcf_exts *exts, struct tcf_proto *tp,
+				u32 handle)
+{
+	struct tcf_exts_miss_cookie_node *n;
+	static u32 next;
+	int err;
+
+	if (WARN_ON(!handle || !tp->ops->get_exts))
+		return -EINVAL;
+
+	n = kzalloc(sizeof(*n), GFP_KERNEL);
+	if (!n)
+		return -ENOMEM;
+
+	n->chain_index = tp->chain->index;
+	n->chain = tp->chain;
+	n->tp_prio = tp->prio;
+	n->tp = tp;
+	n->exts = exts;
+	n->handle = handle;
+
+	err = xa_alloc_cyclic(&tcf_exts_miss_cookies_xa, &n->miss_cookie_base,
+			      n, xa_limit_32b, &next, GFP_KERNEL);
+	if (err)
+		goto err_xa_alloc;
+
+	exts->miss_cookie_node = n;
+	return 0;
+
+err_xa_alloc:
+	kfree(n);
+	return err;
+}
+
+static void tcf_exts_miss_cookie_base_destroy(struct tcf_exts *exts)
+{
+	struct tcf_exts_miss_cookie_node *n;
+
+	if (!exts->miss_cookie_node)
+		return;
+
+	n = exts->miss_cookie_node;
+	xa_erase(&tcf_exts_miss_cookies_xa, n->miss_cookie_base);
+	kfree_rcu(n, rcu);
+}
+
+static struct tcf_exts_miss_cookie_node *
+tcf_exts_miss_cookie_lookup(u64 miss_cookie, int *act_index)
+{
+	union tcf_exts_miss_cookie mc = { .miss_cookie = miss_cookie, };
+
+	*act_index = mc.act_index;
+	return xa_load(&tcf_exts_miss_cookies_xa, mc.miss_cookie_base);
+}
+#else /* IS_ENABLED(CONFIG_NET_TC_SKB_EXT) */
+static int
+tcf_exts_miss_cookie_base_alloc(struct tcf_exts *exts, struct tcf_proto *tp,
+				u32 handle)
+{
+	return 0;
+}
+
+static void tcf_exts_miss_cookie_base_destroy(struct tcf_exts *exts)
+{
+}
+#endif /* IS_ENABLED(CONFIG_NET_TC_SKB_EXT) */
+
+static u64 tcf_exts_miss_cookie_get(u32 miss_cookie_base, int act_index)
+{
+	union tcf_exts_miss_cookie mc = { .act_index = act_index, };
+
+	if (!miss_cookie_base)
+		return 0;
+
+	mc.miss_cookie_base = miss_cookie_base;
+	return mc.miss_cookie;
+}
+
 #ifdef CONFIG_NET_CLS_ACT
 DEFINE_STATIC_KEY_FALSE(tc_skb_ext_tc);
 EXPORT_SYMBOL(tc_skb_ext_tc);
@@ -1549,6 +1653,8 @@ static inline int __tcf_classify(struct sk_buff *skb,
 				 const struct tcf_proto *orig_tp,
 				 struct tcf_result *res,
 				 bool compat_mode,
+				 struct tcf_exts_miss_cookie_node *n,
+				 int act_index,
 				 u32 *last_executed_chain)
 {
 #ifdef CONFIG_NET_CLS_ACT
@@ -1560,13 +1666,36 @@ reclassify:
 #endif
 	for (; tp; tp = rcu_dereference_bh(tp->next)) {
 		__be16 protocol = skb_protocol(skb, false);
-		int err;
+		int err = 0;
 
-		if (tp->protocol != protocol &&
-		    tp->protocol != htons(ETH_P_ALL))
-			continue;
+		if (n) {
+			struct tcf_exts *exts;
+
+			if (n->tp_prio != tp->prio)
+				continue;
+
+			/* We re-lookup the tp and chain based on index instead
+			 * of having hard refs and locks to them, so do a sanity
+			 * check if any of tp,chain,exts was replaced by the
+			 * time we got here with a cookie from hardware.
+			 */
+			if (unlikely(n->tp != tp || n->tp->chain != n->chain ||
+				     !tp->ops->get_exts))
+				return TC_ACT_SHOT;
+
+			exts = tp->ops->get_exts(tp, n->handle);
+			if (unlikely(!exts || n->exts != exts))
+				return TC_ACT_SHOT;
 
-		err = tc_classify(skb, tp, res);
+			n = NULL;
+			err = tcf_exts_exec_ex(skb, exts, act_index, res);
+		} else {
+			if (tp->protocol != protocol &&
+			    tp->protocol != htons(ETH_P_ALL))
+				continue;
+
+			err = tc_classify(skb, tp, res);
+		}
 #ifdef CONFIG_NET_CLS_ACT
 		if (unlikely(err == TC_ACT_RECLASSIFY && !compat_mode)) {
 			first_tp = orig_tp;
@@ -1582,6 +1711,9 @@ reclassify:
 			return err;
 	}
 
+	if (unlikely(n))
+		return TC_ACT_SHOT;
+
 	return TC_ACT_UNSPEC; /* signal: continue lookup */
 #ifdef CONFIG_NET_CLS_ACT
 reset:
@@ -1606,21 +1738,35 @@ int tcf_classify(struct sk_buff *skb,
 #if !IS_ENABLED(CONFIG_NET_TC_SKB_EXT)
 	u32 last_executed_chain = 0;
 
-	return __tcf_classify(skb, tp, tp, res, compat_mode,
+	return __tcf_classify(skb, tp, tp, res, compat_mode, NULL, 0,
 			      &last_executed_chain);
 #else
 	u32 last_executed_chain = tp ? tp->chain->index : 0;
+	struct tcf_exts_miss_cookie_node *n = NULL;
 	const struct tcf_proto *orig_tp = tp;
 	struct tc_skb_ext *ext;
+	int act_index = 0;
 	int ret;
 
 	if (block) {
 		ext = skb_ext_find(skb, TC_SKB_EXT);
 
-		if (ext && ext->chain) {
+		if (ext && (ext->chain || ext->act_miss)) {
 			struct tcf_chain *fchain;
+			u32 chain;
+
+			if (ext->act_miss) {
+				n = tcf_exts_miss_cookie_lookup(ext->act_miss_cookie,
+								&act_index);
+				if (!n)
+					return TC_ACT_SHOT;
 
-			fchain = tcf_chain_lookup_rcu(block, ext->chain);
+				chain = n->chain_index;
+			} else {
+				chain = ext->chain;
+			}
+
+			fchain = tcf_chain_lookup_rcu(block, chain);
 			if (!fchain)
 				return TC_ACT_SHOT;
 
@@ -1632,7 +1778,7 @@ int tcf_classify(struct sk_buff *skb,
 		}
 	}
 
-	ret = __tcf_classify(skb, tp, orig_tp, res, compat_mode,
+	ret = __tcf_classify(skb, tp, orig_tp, res, compat_mode, n, act_index,
 			     &last_executed_chain);
 
 	if (tc_skb_ext_tc_enabled()) {
@@ -3056,9 +3202,48 @@ out:
 	return skb->len;
 }
 
+int tcf_exts_init_ex(struct tcf_exts *exts, struct net *net, int action,
+		     int police, struct tcf_proto *tp, u32 handle,
+		     bool use_action_miss)
+{
+	int err = 0;
+
+#ifdef CONFIG_NET_CLS_ACT
+	exts->type = 0;
+	exts->nr_actions = 0;
+	/* Note: we do not own yet a reference on net.
+	 * This reference might be taken later from tcf_exts_get_net().
+	 */
+	exts->net = net;
+	exts->actions = kcalloc(TCA_ACT_MAX_PRIO, sizeof(struct tc_action *),
+				GFP_KERNEL);
+	if (!exts->actions)
+		return -ENOMEM;
+#endif
+
+	exts->action = action;
+	exts->police = police;
+
+	if (!use_action_miss)
+		return 0;
+
+	err = tcf_exts_miss_cookie_base_alloc(exts, tp, handle);
+	if (err)
+		goto err_miss_alloc;
+
+	return 0;
+
+err_miss_alloc:
+	tcf_exts_destroy(exts);
+	return err;
+}
+EXPORT_SYMBOL(tcf_exts_init_ex);
+
 void tcf_exts_destroy(struct tcf_exts *exts)
 {
 #ifdef CONFIG_NET_CLS_ACT
+	tcf_exts_miss_cookie_base_destroy(exts);
+
 	if (exts->actions) {
 		tcf_action_destroy(exts->actions, TCA_ACT_UNBIND);
 		kfree(exts->actions);
@@ -3547,6 +3732,7 @@ static int tc_setup_offload_act(struct tc_action *act,
 
 int tc_setup_action(struct flow_action *flow_action,
 		    struct tc_action *actions[],
+		    u32 miss_cookie_base,
 		    struct netlink_ext_ack *extack)
 {
 	int i, j, k, index, err = 0;
@@ -3578,6 +3764,8 @@ int tc_setup_action(struct flow_action *flow_action,
 			entry[k].hw_stats = tc_act_hw_stats(act->hw_stats);
 			entry[k].hw_index = act->tcfa_index;
 			entry[k].cookie = (unsigned long)act;
+			entry[k].miss_cookie =
+				tcf_exts_miss_cookie_get(miss_cookie_base, i);
 		}
 
 		j += index;
@@ -3600,10 +3788,15 @@ int tc_setup_offload_action(struct flow_action *flow_action,
 			    struct netlink_ext_ack *extack)
 {
 #ifdef CONFIG_NET_CLS_ACT
+	u32 miss_cookie_base;
+
 	if (!exts)
 		return 0;
 
-	return tc_setup_action(flow_action, exts->actions, extack);
+	miss_cookie_base = exts->miss_cookie_node ?
+			   exts->miss_cookie_node->miss_cookie_base : 0;
+	return tc_setup_action(flow_action, exts->actions, miss_cookie_base,
+			       extack);
 #else
 	return 0;
 #endif
@@ -3771,6 +3964,8 @@ static int __init tc_filter_init(void)
 	if (err)
 		goto err_register_pernet_subsys;
 
+	xa_init_flags(&tcf_exts_miss_cookies_xa, XA_FLAGS_ALLOC1);
+
 	rtnl_register(PF_UNSPEC, RTM_NEWTFILTER, tc_new_tfilter, NULL,
 		      RTNL_FLAG_DOIT_UNLOCKED);
 	rtnl_register(PF_UNSPEC, RTM_DELTFILTER, tc_del_tfilter, NULL,