[RFC nf-next v1 0/2] netfilter: bpf: support prog update

[RFC nf-next v1 0/2] netfilter: bpf: support prog update

[D. Wythe]

From: "D. Wythe" <alibuda@linux.alibaba.com>

This patches attempt to implements updating of progs within
bpf netfilter link, allowing user update their ebpf netfilter
prog in hot update manner.

Besides, a corresponding test case has been added to verify
whether the update works.

--
v1:
1. remove unnecessary context, access the prog directly via rcu.
2. remove synchronize_rcu(), dealloc the nf_link via kfree_rcu.
3. check the dead flag during the update.

D. Wythe (2):
  netfilter: bpf: support prog update
  selftests/bpf: Add netfilter link prog update test

 net/netfilter/nf_bpf_link.c                        | 68 ++++++++++++++----
 .../bpf/prog_tests/netfilter_link_update_prog.c    | 83 ++++++++++++++++++++++
 .../bpf/progs/test_netfilter_link_update_prog.c    | 24 +++++++
 3 files changed, 160 insertions(+), 15 deletions(-)
 create mode 100644 tools/testing/selftests/bpf/prog_tests/netfilter_link_update_prog.c
 create mode 100644 tools/testing/selftests/bpf/progs/test_netfilter_link_update_prog.c

-- 
1.8.3.1

[RFC nf-next v1 1/2] netfilter: bpf: support prog update

[D. Wythe]

From: "D. Wythe" <alibuda@linux.alibaba.com>

To support the prog update, we need to ensure that the prog seen
within the hook is always valid. Considering that hooks are always
protected by rcu_read_lock(), which provide us the ability to
access the prog under rcu.

Signed-off-by: D. Wythe <alibuda@linux.alibaba.com>
---
 net/netfilter/nf_bpf_link.c | 68 +++++++++++++++++++++++++++++++++++----------
 1 file changed, 53 insertions(+), 15 deletions(-)

diff --git a/net/netfilter/nf_bpf_link.c b/net/netfilter/nf_bpf_link.c
index e502ec0..e17f256 100644
--- a/net/netfilter/nf_bpf_link.c
+++ b/net/netfilter/nf_bpf_link.c
@@ -8,17 +8,8 @@
 #include <net/netfilter/nf_bpf_link.h>
 #include <uapi/linux/netfilter_ipv4.h>
 
-static unsigned int nf_hook_run_bpf(void *bpf_prog, struct sk_buff *skb,
-				    const struct nf_hook_state *s)
-{
-	const struct bpf_prog *prog = bpf_prog;
-	struct bpf_nf_ctx ctx = {
-		.state = s,
-		.skb = skb,
-	};
-
-	return bpf_prog_run(prog, &ctx);
-}
+/* protect link update in parallel */
+static DEFINE_MUTEX(bpf_nf_mutex);
 
 struct bpf_nf_link {
 	struct bpf_link link;
@@ -26,8 +17,22 @@ struct bpf_nf_link {
 	struct net *net;
 	u32 dead;
 	const struct nf_defrag_hook *defrag_hook;
+	const struct bpf_prog __rcu *nf_prog;
+	struct rcu_head head;
 };
 
+static unsigned int nf_hook_run_bpf(void *bpf_link, struct sk_buff *skb,
+				    const struct nf_hook_state *s)
+{
+	const struct bpf_nf_link *link = bpf_link;
+	struct bpf_nf_ctx ctx = {
+		.state = s,
+		.skb = skb,
+	};
+
+	return bpf_prog_run(rcu_dereference(link->nf_prog), &ctx);
+}
+
 #if IS_ENABLED(CONFIG_NF_DEFRAG_IPV4) || IS_ENABLED(CONFIG_NF_DEFRAG_IPV6)
 static const struct nf_defrag_hook *
 get_proto_defrag_hook(struct bpf_nf_link *link,
@@ -126,8 +131,7 @@ static void bpf_nf_link_release(struct bpf_link *link)
 static void bpf_nf_link_dealloc(struct bpf_link *link)
 {
 	struct bpf_nf_link *nf_link = container_of(link, struct bpf_nf_link, link);
-
-	kfree(nf_link);
+	kfree_rcu(nf_link, head);
 }
 
 static int bpf_nf_link_detach(struct bpf_link *link)
@@ -162,7 +166,35 @@ static int bpf_nf_link_fill_link_info(const struct bpf_link *link,
 static int bpf_nf_link_update(struct bpf_link *link, struct bpf_prog *new_prog,
 			      struct bpf_prog *old_prog)
 {
-	return -EOPNOTSUPP;
+	struct bpf_nf_link *nf_link = container_of(link, struct bpf_nf_link, link);
+	int err = 0;
+
+	mutex_lock(&bpf_nf_mutex);
+
+	if (nf_link->dead) {
+		err = -EPERM;
+		goto out;
+	}
+
+	/* target old_prog mismatch */
+	if (old_prog && link->prog != old_prog) {
+		err = -EPERM;
+		goto out;
+	}
+
+	old_prog = link->prog;
+	if (old_prog == new_prog) {
+		/* don't need update */
+		bpf_prog_put(new_prog);
+		goto out;
+	}
+
+	rcu_assign_pointer(nf_link->nf_prog, new_prog);
+	old_prog = xchg(&link->prog, new_prog);
+	bpf_prog_put(old_prog);
+out:
+	mutex_unlock(&bpf_nf_mutex);
+	return err;
 }
 
 static const struct bpf_link_ops bpf_nf_link_lops = {
@@ -226,7 +258,13 @@ int bpf_nf_link_attach(const union bpf_attr *attr, struct bpf_prog *prog)
 
 	link->hook_ops.hook = nf_hook_run_bpf;
 	link->hook_ops.hook_ops_type = NF_HOOK_OP_BPF;
-	link->hook_ops.priv = prog;
+
+	rcu_assign_pointer(link->nf_prog, prog);
+
+	/* bpf_nf_link_release & bpf_nf_link_dealloc() can ensures that link remains
+	 * valid at all times within nf_hook_run_bpf().
+	 */
+	link->hook_ops.priv = link;
 
 	link->hook_ops.pf = attr->link_create.netfilter.pf;
 	link->hook_ops.priority = attr->link_create.netfilter.priority;
-- 
1.8.3.1

Re: [RFC nf-next v1 1/2] netfilter: bpf: support prog update

[Florian Westphal]

D. Wythe <alibuda@linux.alibaba.com> wrote:
>  	const struct nf_defrag_hook *defrag_hook;
> +	const struct bpf_prog __rcu *nf_prog;

Hmm, why do we need this pointer?
Can't you just re-use bpf_nf_link->link.prog?

> +	rcu_assign_pointer(nf_link->nf_prog, new_prog);
> +	old_prog = xchg(&link->prog, new_prog);

This looks redundant, I think you can remove the nf_prog
pointer again.

Rest LGTM.

Re: [RFC nf-next v1 1/2] netfilter: bpf: support prog update

[D. Wythe]

On 12/15/23 10:17 PM, Florian Westphal wrote:
> D. Wythe <alibuda@linux.alibaba.com> wrote:
>>   	const struct nf_defrag_hook *defrag_hook;
>> +	const struct bpf_prog __rcu *nf_prog;
> Hmm, why do we need this pointer?
> Can't you just re-use bpf_nf_link->link.prog?
Accessing nf_link->link.prog directly is a bit strange because it is not 
marked as __rcu, which will generate a compilation warning,
thus we need to perform a type conversion.

But I do not intend to insist on it. I will remove it in the next version.

Best wishes,
D. Wythe
>
>> +	rcu_assign_pointer(nf_link->nf_prog, new_prog);
>> +	old_prog = xchg(&link->prog, new_prog);
> This looks redundant, I think you can remove the nf_prog
> pointer again.
>
> Rest LGTM.

[RFC nf-next v1 2/2] selftests/bpf: Add netfilter link prog update test

[D. Wythe]

From: "D. Wythe" <alibuda@linux.alibaba.com>

Update prog for active links and verify whether
the prog has been successfully replaced.

Expected output:

./test_progs -t netfilter_link_update_prog
Summary: 1/0 PASSED, 0 SKIPPED, 0 FAILED

Signed-off-by: D. Wythe <alibuda@linux.alibaba.com>
---
 .../bpf/prog_tests/netfilter_link_update_prog.c    | 83 ++++++++++++++++++++++
 .../bpf/progs/test_netfilter_link_update_prog.c    | 24 +++++++
 2 files changed, 107 insertions(+)
 create mode 100644 tools/testing/selftests/bpf/prog_tests/netfilter_link_update_prog.c
 create mode 100644 tools/testing/selftests/bpf/progs/test_netfilter_link_update_prog.c

diff --git a/tools/testing/selftests/bpf/prog_tests/netfilter_link_update_prog.c b/tools/testing/selftests/bpf/prog_tests/netfilter_link_update_prog.c
new file mode 100644
index 00000000..d23b544
--- /dev/null
+++ b/tools/testing/selftests/bpf/prog_tests/netfilter_link_update_prog.c
@@ -0,0 +1,83 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#include <test_progs.h>
+#include <linux/netfilter.h>
+#include <network_helpers.h>
+#include "test_netfilter_link_update_prog.skel.h"
+
+#define SERVER_ADDR "127.0.0.1"
+#define SERVER_PORT 12345
+
+static const char dummy_message[] = "A dummy message";
+
+static int send_dummy(int client_fd)
+{
+	struct sockaddr_storage saddr;
+	struct sockaddr *saddr_p;
+	socklen_t saddr_len;
+	int err;
+
+	saddr_p = (struct sockaddr *)&saddr;
+	err = make_sockaddr(AF_INET, SERVER_ADDR, SERVER_PORT, &saddr, &saddr_len);
+	if (!ASSERT_OK(err, "make_sockaddr"))
+		return -1;
+
+	err = sendto(client_fd, dummy_message, sizeof(dummy_message) - 1, 0, saddr_p, saddr_len);
+	if (!ASSERT_GE(err, 0, "sendto"))
+		return -1;
+
+	return 0;
+}
+
+void test_netfilter_link_update_prog(void)
+{
+	LIBBPF_OPTS(bpf_netfilter_opts, opts,
+		.pf = NFPROTO_IPV4,
+		.hooknum = NF_INET_LOCAL_OUT,
+		.priority = 100);
+	struct test_netfilter_link_update_prog *skel;
+	struct bpf_program *prog;
+	int server_fd, client_fd;
+	int err;
+
+	skel = test_netfilter_link_update_prog__open_and_load();
+	if (!ASSERT_OK_PTR(skel, "test_netfilter_link_update_prog__open_and_load"))
+		goto out;
+
+	prog = skel->progs.nf_link_prog;
+
+	if (!ASSERT_OK_PTR(prog, "load program"))
+		goto out;
+
+	skel->links.nf_link_prog = bpf_program__attach_netfilter(prog, &opts);
+	if (!ASSERT_OK_PTR(skel->links.nf_link_prog, "attach netfilter program"))
+		goto out;
+
+	server_fd = start_server(AF_INET, SOCK_DGRAM, SERVER_ADDR, SERVER_PORT, 0);
+	if (!ASSERT_GE(server_fd, 0, "start_server"))
+		goto out;
+
+	client_fd = connect_to_fd(server_fd, 0);
+	if (!ASSERT_GE(client_fd, 0, "connect_to_fd"))
+		goto out;
+
+	send_dummy(client_fd);
+
+	ASSERT_EQ(skel->bss->counter, 0, "counter should be zero");
+
+	err = bpf_link__update_program(skel->links.nf_link_prog, skel->progs.nf_link_prog_new);
+	if (!ASSERT_OK(err, "bpf_link__update_program"))
+		goto out;
+
+	send_dummy(client_fd);
+	ASSERT_GE(skel->bss->counter, 0, "counter should be greater than zero");
+out:
+	if (client_fd > 0)
+		close(client_fd);
+	if (server_fd > 0)
+		close(server_fd);
+
+	test_netfilter_link_update_prog__destroy(skel);
+}
+
+
diff --git a/tools/testing/selftests/bpf/progs/test_netfilter_link_update_prog.c b/tools/testing/selftests/bpf/progs/test_netfilter_link_update_prog.c
new file mode 100644
index 00000000..42ae332
--- /dev/null
+++ b/tools/testing/selftests/bpf/progs/test_netfilter_link_update_prog.c
@@ -0,0 +1,24 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#include "vmlinux.h"
+#include <bpf/bpf_helpers.h>
+
+#define NF_ACCEPT 1
+
+SEC("netfilter")
+int nf_link_prog(struct bpf_nf_ctx *ctx)
+{
+	return NF_ACCEPT;
+}
+
+u64 counter = 0;
+
+SEC("netfilter")
+int nf_link_prog_new(struct bpf_nf_ctx *ctx)
+{
+	counter++;
+	return NF_ACCEPT;
+}
+
+char _license[] SEC("license") = "GPL";
+
-- 
1.8.3.1