Thread (16 messages) 16 messages, 3 authors, 2023-06-29

Re: [PATCH bpf-next 4/7] netfilter: bpf: Support BPF_F_NETFILTER_IP_DEFRAG in netfilter link

From: Florian Westphal <fw@strlen.de>
Date: 2023-06-27 11:13:06
Also in: bpf, lkml, netfilter-devel

Daniel Xu [off-list ref] wrote:
+static int bpf_nf_enable_defrag(struct bpf_nf_link *link)
+{
+	int err;
+
+	switch (link->hook_ops.pf) {
+#if IS_ENABLED(CONFIG_NF_DEFRAG_IPV4)
+	case NFPROTO_IPV4:
+		const struct nf_defrag_v4_hook *v4_hook;
+
+		err = request_module("nf_defrag_ipv4");
+		if (err)
+			return err;
+
+		rcu_read_lock();
+		v4_hook = rcu_dereference(nf_defrag_v4_hook);
+		err = v4_hook->enable(link->net);
+		rcu_read_unlock();
I'd reverse this, first try rcu_dereference(), then modprobe
if thats returned NULL.
+static void bpf_nf_disable_defrag(struct bpf_nf_link *link)
+{
+	switch (link->hook_ops.pf) {
+#if IS_ENABLED(CONFIG_NF_DEFRAG_IPV4)
+	case NFPROTO_IPV4:
+		const struct nf_defrag_v4_hook *v4_hook;
+
+		rcu_read_lock();
+		v4_hook = rcu_dereference(nf_defrag_v4_hook);
+		v4_hook->disable(link->net);
+		rcu_read_unlock();
if (v4_hook)
	v4_hook->disable()

Else we get trouble on manual 'rmmod'.
+	/* make sure conntrack confirm is always last */
+	prio = attr->link_create.netfilter.priority;
+	if (prio == NF_IP_PRI_FIRST)
+		return -ERANGE;  /* sabotage_in and other warts */
+	else if (prio == NF_IP_PRI_LAST)
+		return -ERANGE;  /* e.g. conntrack confirm */
+	else if ((attr->link_create.netfilter.flags & BPF_F_NETFILTER_IP_DEFRAG) &&
+		 (prio > NF_IP_PRI_FIRST && prio <= NF_IP_PRI_CONNTRACK_DEFRAG))
+		return -ERANGE;  /* cannot use defrag if prog runs before nf_defrag */
You could elide the (prio > NF_IP_PRI_FIRST, its already handled by
first conditional.  Otherwise this looks good to me.
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help