OSDN Git Service

net_sched: refactor notification code for RTM_DELTFILTER
authorWANG Cong <xiyou.wangcong@gmail.com>
Sat, 5 Aug 2017 04:31:42 +0000 (21:31 -0700)
committerDavid S. Miller <davem@davemloft.net>
Mon, 7 Aug 2017 21:12:17 +0000 (14:12 -0700)
It is confusing to use 'unsigned long fh' as both a handle
and a pointer, especially commit 9ee7837449b3
("net sched filters: fix notification of filter delete with proper handle").

This patch introduces tfilter_del_notify() so that we can
pass it as a pointer as before, and we don't need to check
RTM_DELTFILTER in tcf_fill_node() any more.

This prepares for the next patch.

Cc: Jamal Hadi Salim <jhs@mojatatu.com>
Cc: Jiri Pirko <jiri@resnulli.us>
Signed-off-by: Cong Wang <xiyou.wangcong@gmail.com>
Acked-by: Jamal Hadi Salim <jhs@mojatatu.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
net/sched/cls_api.c

index e655221..afd0997 100644 (file)
@@ -104,6 +104,10 @@ static int tfilter_notify(struct net *net, struct sk_buff *oskb,
                          struct nlmsghdr *n, struct tcf_proto *tp,
                          unsigned long fh, int event, bool unicast);
 
+static int tfilter_del_notify(struct net *net, struct sk_buff *oskb,
+                             struct nlmsghdr *n, struct tcf_proto *tp,
+                             unsigned long fh, bool unicast, bool *last);
+
 static void tfilter_notify_chain(struct net *net, struct sk_buff *oskb,
                                 struct nlmsghdr *n,
                                 struct tcf_chain *chain, int event)
@@ -595,11 +599,10 @@ replay:
                        }
                        break;
                case RTM_DELTFILTER:
-                       err = tp->ops->delete(tp, fh, &last);
+                       err = tfilter_del_notify(net, skb, n, tp, fh, false,
+                                                &last);
                        if (err)
                                goto errout;
-                       tfilter_notify(net, skb, n, tp, t->tcm_handle,
-                                      RTM_DELTFILTER, false);
                        if (last) {
                                tcf_chain_tp_remove(chain, &chain_info, tp);
                                tcf_proto_destroy(tp);
@@ -659,9 +662,9 @@ static int tcf_fill_node(struct net *net, struct sk_buff *skb,
                goto nla_put_failure;
        if (nla_put_u32(skb, TCA_CHAIN, tp->chain->index))
                goto nla_put_failure;
-       tcm->tcm_handle = fh;
-       if (RTM_DELTFILTER != event) {
+       if (!fh) {
                tcm->tcm_handle = 0;
+       } else {
                if (tp->ops->dump && tp->ops->dump(net, tp, fh, skb, tcm) < 0)
                        goto nla_put_failure;
        }
@@ -698,6 +701,37 @@ static int tfilter_notify(struct net *net, struct sk_buff *oskb,
                              n->nlmsg_flags & NLM_F_ECHO);
 }
 
+static int tfilter_del_notify(struct net *net, struct sk_buff *oskb,
+                             struct nlmsghdr *n, struct tcf_proto *tp,
+                             unsigned long fh, bool unicast, bool *last)
+{
+       struct sk_buff *skb;
+       u32 portid = oskb ? NETLINK_CB(oskb).portid : 0;
+       int err;
+
+       skb = alloc_skb(NLMSG_GOODSIZE, GFP_KERNEL);
+       if (!skb)
+               return -ENOBUFS;
+
+       if (tcf_fill_node(net, skb, tp, fh, portid, n->nlmsg_seq,
+                         n->nlmsg_flags, RTM_DELTFILTER) <= 0) {
+               kfree_skb(skb);
+               return -EINVAL;
+       }
+
+       err = tp->ops->delete(tp, fh, last);
+       if (err) {
+               kfree_skb(skb);
+               return err;
+       }
+
+       if (unicast)
+               return netlink_unicast(net->rtnl, skb, portid, MSG_DONTWAIT);
+
+       return rtnetlink_send(skb, net, portid, RTNLGRP_TC,
+                             n->nlmsg_flags & NLM_F_ECHO);
+}
+
 struct tcf_dump_args {
        struct tcf_walker w;
        struct sk_buff *skb;