nfp: flower: refactor match functions to take flow_rule as input
authorLouis Peens <louis.peens@corigine.com>
Thu, 22 Jul 2021 07:58:01 +0000 (09:58 +0200)
committerDavid S. Miller <davem@davemloft.net>
Thu, 22 Jul 2021 09:22:22 +0000 (02:22 -0700)
This is a small cleanup to pass in flow->rule to some of the compile
functions instead of extracting it every time. This is will also be
useful for conntrack patches later.

Signed-off-by: Louis Peens <louis.peens@corigine.com>
Signed-off-by: Yinjun Zhang <yinjun.zhang@corigine.com>
Signed-off-by: Simon Horman <simon.horman@corigine.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
drivers/net/ethernet/netronome/nfp/flower/main.h
drivers/net/ethernet/netronome/nfp/flower/match.c
drivers/net/ethernet/netronome/nfp/flower/metadata.c
drivers/net/ethernet/netronome/nfp/flower/offload.c

index beb19de..0046655 100644 (file)
@@ -468,7 +468,7 @@ nfp_flower_compile_ipv6_gre_tun(struct nfp_flower_ipv6_gre_tun *ext,
                                struct nfp_flower_ipv6_gre_tun *msk,
                                struct flow_rule *rule);
 int nfp_flower_compile_flow_match(struct nfp_app *app,
-                                 struct flow_cls_offload *flow,
+                                 struct flow_rule *rule,
                                  struct nfp_fl_key_ls *key_ls,
                                  struct net_device *netdev,
                                  struct nfp_fl_payload *nfp_flow,
@@ -479,8 +479,7 @@ int nfp_flower_compile_action(struct nfp_app *app,
                              struct net_device *netdev,
                              struct nfp_fl_payload *nfp_flow,
                              struct netlink_ext_ack *extack);
-int nfp_compile_flow_metadata(struct nfp_app *app,
-                             struct flow_cls_offload *flow,
+int nfp_compile_flow_metadata(struct nfp_app *app, u32 cookie,
                              struct nfp_fl_payload *nfp_flow,
                              struct net_device *netdev,
                              struct netlink_ext_ack *extack);
index 9af1bd9..9d86eea 100644 (file)
@@ -511,14 +511,13 @@ nfp_flower_compile_ipv6_gre_tun(struct nfp_flower_ipv6_gre_tun *ext,
 }
 
 int nfp_flower_compile_flow_match(struct nfp_app *app,
-                                 struct flow_cls_offload *flow,
+                                 struct flow_rule *rule,
                                  struct nfp_fl_key_ls *key_ls,
                                  struct net_device *netdev,
                                  struct nfp_fl_payload *nfp_flow,
                                  enum nfp_flower_tun_type tun_type,
                                  struct netlink_ext_ack *extack)
 {
-       struct flow_rule *rule = flow_cls_offload_flow_rule(flow);
        struct nfp_flower_priv *priv = app->priv;
        bool qinq_sup;
        u32 port_id;
index 6211136..2af9fae 100644 (file)
@@ -290,8 +290,7 @@ nfp_check_mask_remove(struct nfp_app *app, char *mask_data, u32 mask_len,
        return true;
 }
 
-int nfp_compile_flow_metadata(struct nfp_app *app,
-                             struct flow_cls_offload *flow,
+int nfp_compile_flow_metadata(struct nfp_app *app, u32 cookie,
                              struct nfp_fl_payload *nfp_flow,
                              struct net_device *netdev,
                              struct netlink_ext_ack *extack)
@@ -310,7 +309,7 @@ int nfp_compile_flow_metadata(struct nfp_app *app,
        }
 
        nfp_flow->meta.host_ctx_id = cpu_to_be32(stats_cxt);
-       nfp_flow->meta.host_cookie = cpu_to_be64(flow->cookie);
+       nfp_flow->meta.host_cookie = cpu_to_be64(cookie);
        nfp_flow->ingress_dev = netdev;
 
        ctx_entry = kzalloc(sizeof(*ctx_entry), GFP_KERNEL);
@@ -357,7 +356,7 @@ int nfp_compile_flow_metadata(struct nfp_app *app,
        priv->stats[stats_cxt].bytes = 0;
        priv->stats[stats_cxt].used = jiffies;
 
-       check_entry = nfp_flower_search_fl_table(app, flow->cookie, netdev);
+       check_entry = nfp_flower_search_fl_table(app, cookie, netdev);
        if (check_entry) {
                NL_SET_ERR_MSG_MOD(extack, "invalid entry: cannot offload duplicate flow entry");
                if (nfp_release_stats_entry(app, stats_cxt)) {
index 2406d33..46bd5da 100644 (file)
@@ -134,20 +134,16 @@ nfp_flower_xmit_flow(struct nfp_app *app, struct nfp_fl_payload *nfp_flow,
        return 0;
 }
 
-static bool nfp_flower_check_higher_than_mac(struct flow_cls_offload *f)
+static bool nfp_flower_check_higher_than_mac(struct flow_rule *rule)
 {
-       struct flow_rule *rule = flow_cls_offload_flow_rule(f);
-
        return flow_rule_match_key(rule, FLOW_DISSECTOR_KEY_IPV4_ADDRS) ||
               flow_rule_match_key(rule, FLOW_DISSECTOR_KEY_IPV6_ADDRS) ||
               flow_rule_match_key(rule, FLOW_DISSECTOR_KEY_PORTS) ||
               flow_rule_match_key(rule, FLOW_DISSECTOR_KEY_ICMP);
 }
 
-static bool nfp_flower_check_higher_than_l3(struct flow_cls_offload *f)
+static bool nfp_flower_check_higher_than_l3(struct flow_rule *rule)
 {
-       struct flow_rule *rule = flow_cls_offload_flow_rule(f);
-
        return flow_rule_match_key(rule, FLOW_DISSECTOR_KEY_PORTS) ||
               flow_rule_match_key(rule, FLOW_DISSECTOR_KEY_ICMP);
 }
@@ -240,11 +236,10 @@ static int
 nfp_flower_calculate_key_layers(struct nfp_app *app,
                                struct net_device *netdev,
                                struct nfp_fl_key_ls *ret_key_ls,
-                               struct flow_cls_offload *flow,
+                               struct flow_rule *rule,
                                enum nfp_flower_tun_type *tun_type,
                                struct netlink_ext_ack *extack)
 {
-       struct flow_rule *rule = flow_cls_offload_flow_rule(flow);
        struct flow_dissector *dissector = rule->match.dissector;
        struct flow_match_basic basic = { NULL, NULL};
        struct nfp_flower_priv *priv = app->priv;
@@ -452,7 +447,7 @@ nfp_flower_calculate_key_layers(struct nfp_app *app,
                        NL_SET_ERR_MSG_MOD(extack, "unsupported offload: match on given EtherType is not supported");
                        return -EOPNOTSUPP;
                }
-       } else if (nfp_flower_check_higher_than_mac(flow)) {
+       } else if (nfp_flower_check_higher_than_mac(rule)) {
                NL_SET_ERR_MSG_MOD(extack, "unsupported offload: cannot match above L2 without specified EtherType");
                return -EOPNOTSUPP;
        }
@@ -471,7 +466,7 @@ nfp_flower_calculate_key_layers(struct nfp_app *app,
        }
 
        if (!(key_layer & NFP_FLOWER_LAYER_TP) &&
-           nfp_flower_check_higher_than_l3(flow)) {
+           nfp_flower_check_higher_than_l3(rule)) {
                NL_SET_ERR_MSG_MOD(extack, "unsupported offload: cannot match on L4 information without specified IP protocol type");
                return -EOPNOTSUPP;
        }
@@ -1005,9 +1000,7 @@ int nfp_flower_merge_offloaded_flows(struct nfp_app *app,
                                     struct nfp_fl_payload *sub_flow1,
                                     struct nfp_fl_payload *sub_flow2)
 {
-       struct flow_cls_offload merge_tc_off;
        struct nfp_flower_priv *priv = app->priv;
-       struct netlink_ext_ack *extack = NULL;
        struct nfp_fl_payload *merge_flow;
        struct nfp_fl_key_ls merge_key_ls;
        struct nfp_merge_info *merge_info;
@@ -1016,7 +1009,6 @@ int nfp_flower_merge_offloaded_flows(struct nfp_app *app,
 
        ASSERT_RTNL();
 
-       extack = merge_tc_off.common.extack;
        if (sub_flow1 == sub_flow2 ||
            nfp_flower_is_merge_flow(sub_flow1) ||
            nfp_flower_is_merge_flow(sub_flow2))
@@ -1061,9 +1053,8 @@ int nfp_flower_merge_offloaded_flows(struct nfp_app *app,
        if (err)
                goto err_unlink_sub_flow1;
 
-       merge_tc_off.cookie = merge_flow->tc_flower_cookie;
-       err = nfp_compile_flow_metadata(app, &merge_tc_off, merge_flow,
-                                       merge_flow->ingress_dev, extack);
+       err = nfp_compile_flow_metadata(app, merge_flow->tc_flower_cookie, merge_flow,
+                                       merge_flow->ingress_dev, NULL);
        if (err)
                goto err_unlink_sub_flow2;
 
@@ -1305,6 +1296,7 @@ static int
 nfp_flower_add_offload(struct nfp_app *app, struct net_device *netdev,
                       struct flow_cls_offload *flow)
 {
+       struct flow_rule *rule = flow_cls_offload_flow_rule(flow);
        enum nfp_flower_tun_type tun_type = NFP_FL_TUNNEL_NONE;
        struct nfp_flower_priv *priv = app->priv;
        struct netlink_ext_ack *extack = NULL;
@@ -1330,7 +1322,7 @@ nfp_flower_add_offload(struct nfp_app *app, struct net_device *netdev,
        if (!key_layer)
                return -ENOMEM;
 
-       err = nfp_flower_calculate_key_layers(app, netdev, key_layer, flow,
+       err = nfp_flower_calculate_key_layers(app, netdev, key_layer, rule,
                                              &tun_type, extack);
        if (err)
                goto err_free_key_ls;
@@ -1341,7 +1333,7 @@ nfp_flower_add_offload(struct nfp_app *app, struct net_device *netdev,
                goto err_free_key_ls;
        }
 
-       err = nfp_flower_compile_flow_match(app, flow, key_layer, netdev,
+       err = nfp_flower_compile_flow_match(app, rule, key_layer, netdev,
                                            flow_pay, tun_type, extack);
        if (err)
                goto err_destroy_flow;
@@ -1356,7 +1348,7 @@ nfp_flower_add_offload(struct nfp_app *app, struct net_device *netdev,
                        goto err_destroy_flow;
        }
 
-       err = nfp_compile_flow_metadata(app, flow, flow_pay, netdev, extack);
+       err = nfp_compile_flow_metadata(app, flow->cookie, flow_pay, netdev, extack);
        if (err)
                goto err_destroy_flow;