Skip to contentSkip to conversation

net: netfilter: Make ct zone opts configurable for bpf ct helpers

PATCH Superseded Brad Cowie 2 participants 3 messages May 8, 2024 – May 20, 2024

Subsystems (6)

v3 of 4 Superseded Latest is v4 →

lore
Share & tools
lore

Brad Cowie

author
#0 lore raw

PATCH v3 1/2 net: netfilter: Make ct zone opts configurable for bpf ct helpers

Add ct zone id and direction to bpf_ct_opts so that arbitrary ct zones can be used for xdp/tc bpf ct helper functions bpf_{xdp,skb}_ct_alloc and bpf_{xdp,skb}_ct_lookup.

---
v2 -> v3:
- Remove whitespace changes
- Add reserved padding options
- If ct_zone_id is set when opts__sz isn't 16 return -EINVAL
- Remove ct_zone_flags, not used by nf_conntrack_alloc or nf_conntrack_find_get
- Update comments to reflect opts__sz can be 16 or 12

v1 -> v2:
- Make ct zone flags/dir configurable

---
 net/netfilter/nf_conntrack_bpf.c | 68 ++++++++++++++++++++++++++------
 1 file changed, 55 insertions(+), 13 deletions(-)
net/netfilter/nf_conntrack_bpf.c +55 -13
index d2492d050fe6..4a136fc3a9c0 100644
@@ -32,7 +32,9 @@
  *		   -EINVAL - Passed NULL for bpf_tuple pointer
  *		   -EINVAL - opts->reserved is not 0
  *		   -EINVAL - netns_id is less than -1
- *		   -EINVAL - opts__sz isn't NF_BPF_CT_OPTS_SZ (12)
+ *		   -EINVAL - opts__sz isn't NF_BPF_CT_OPTS_SZ (16) or 12
+ *		   -EINVAL - opts->ct_zone_id set when
+			     opts__sz isn't NF_BPF_CT_OPTS_SZ (16)
  *		   -EPROTO - l4proto isn't one of IPPROTO_TCP or IPPROTO_UDP
  *		   -ENONET - No network namespace found for netns_id
  *		   -ENOENT - Conntrack lookup could not find entry for tuple
@@ -42,6 +44,8 @@
  *		 Values:
  *		   IPPROTO_TCP, IPPROTO_UDP
  * @dir:       - connection tracking tuple direction.
+ * @ct_zone_id - connection tracking zone id.
+ * @ct_zone_dir - connection tracking zone direction.
  * @reserved   - Reserved member, will be reused for more options in future
  *		 Values:
  *		   0
@@ -51,11 +55,13 @@ struct bpf_ct_opts {
 	s32 error;
 	u8 l4proto;
 	u8 dir;
-	u8 reserved[2];
+	u16 ct_zone_id;
+	u8 ct_zone_dir;
+	u8 reserved[3];
 };
 
 enum {
-	NF_BPF_CT_OPTS_SZ = 12,
+	NF_BPF_CT_OPTS_SZ = 16,
 };
 
 static int bpf_nf_ct_tuple_parse(struct bpf_sock_tuple *bpf_tuple,
@@ -104,12 +110,21 @@ __bpf_nf_ct_alloc_entry(struct net *net, struct bpf_sock_tuple *bpf_tuple,
 			u32 timeout)
 {
 	struct nf_conntrack_tuple otuple, rtuple;
+	struct nf_conntrack_zone ct_zone;
 	struct nf_conn *ct;
 	int err;
 
-	if (!opts || !bpf_tuple || opts->reserved[0] || opts->reserved[1] ||
-	    opts_len != NF_BPF_CT_OPTS_SZ)
+	if (!opts || !bpf_tuple)
 		return ERR_PTR(-EINVAL);
+	if (!(opts_len == NF_BPF_CT_OPTS_SZ || opts_len == 12))
+		return ERR_PTR(-EINVAL);
+	if (opts_len == NF_BPF_CT_OPTS_SZ) {
+		if (opts->reserved[0] || opts->reserved[1] || opts->reserved[2])
+			return ERR_PTR(-EINVAL);
+	} else {
+		if (opts->ct_zone_id)
+			return ERR_PTR(-EINVAL);
+	}
 
 	if (unlikely(opts->netns_id < BPF_F_CURRENT_NETNS))
 		return ERR_PTR(-EINVAL);
@@ -130,7 +145,16 @@ __bpf_nf_ct_alloc_entry(struct net *net, struct bpf_sock_tuple *bpf_tuple,
 			return ERR_PTR(-ENONET);
 	}
 
-	ct = nf_conntrack_alloc(net, &nf_ct_zone_dflt, &otuple, &rtuple,
+	if (opts_len == NF_BPF_CT_OPTS_SZ) {
+		if (opts->ct_zone_dir == 0)
+			opts->ct_zone_dir = NF_CT_DEFAULT_ZONE_DIR;
+		nf_ct_zone_init(&ct_zone,
+				opts->ct_zone_id, opts->ct_zone_dir, 0);
+	} else {
+		ct_zone = nf_ct_zone_dflt;
+	}
+
+	ct = nf_conntrack_alloc(net, &ct_zone, &otuple, &rtuple,
 				GFP_ATOMIC);
 	if (IS_ERR(ct))
 		goto out;
@@ -152,12 +176,21 @@ static struct nf_conn *__bpf_nf_ct_lookup(struct net *net,
 {
 	struct nf_conntrack_tuple_hash *hash;
 	struct nf_conntrack_tuple tuple;
+	struct nf_conntrack_zone ct_zone;
 	struct nf_conn *ct;
 	int err;
 
-	if (!opts || !bpf_tuple || opts->reserved[0] || opts->reserved[1] ||
-	    opts_len != NF_BPF_CT_OPTS_SZ)
+	if (!opts || !bpf_tuple)
 		return ERR_PTR(-EINVAL);
+	if (!(opts_len == NF_BPF_CT_OPTS_SZ || opts_len == 12))
+		return ERR_PTR(-EINVAL);
+	if (opts_len == NF_BPF_CT_OPTS_SZ) {
+		if (opts->reserved[0] || opts->reserved[1] || opts->reserved[2])
+			return ERR_PTR(-EINVAL);
+	} else {
+		if (opts->ct_zone_id)
+			return ERR_PTR(-EINVAL);
+	}
 	if (unlikely(opts->l4proto != IPPROTO_TCP && opts->l4proto != IPPROTO_UDP))
 		return ERR_PTR(-EPROTO);
 	if (unlikely(opts->netns_id < BPF_F_CURRENT_NETNS))
@@ -174,7 +207,16 @@ static struct nf_conn *__bpf_nf_ct_lookup(struct net *net,
 			return ERR_PTR(-ENONET);
 	}
 
-	hash = nf_conntrack_find_get(net, &nf_ct_zone_dflt, &tuple);
+	if (opts_len == NF_BPF_CT_OPTS_SZ) {
+		if (opts->ct_zone_dir == 0)
+			opts->ct_zone_dir = NF_CT_DEFAULT_ZONE_DIR;
+		nf_ct_zone_init(&ct_zone,
+				opts->ct_zone_id, opts->ct_zone_dir, 0);
+	} else {
+		ct_zone = nf_ct_zone_dflt;
+	}
+
+	hash = nf_conntrack_find_get(net, &ct_zone, &tuple);
 	if (opts->netns_id >= 0)
 		put_net(net);
 	if (!hash)
@@ -245,7 +287,7 @@ __bpf_kfunc_start_defs();
  * @opts	- Additional options for allocation (documented above)
  *		    Cannot be NULL
  * @opts__sz	- Length of the bpf_ct_opts structure
- *		    Must be NF_BPF_CT_OPTS_SZ (12)
+ *		    Must be NF_BPF_CT_OPTS_SZ (16) or 12
  */
 __bpf_kfunc struct nf_conn___init *
 bpf_xdp_ct_alloc(struct xdp_md *xdp_ctx, struct bpf_sock_tuple *bpf_tuple,
@@ -279,7 +321,7 @@ bpf_xdp_ct_alloc(struct xdp_md *xdp_ctx, struct bpf_sock_tuple *bpf_tuple,
  * @opts	- Additional options for lookup (documented above)
  *		    Cannot be NULL
  * @opts__sz	- Length of the bpf_ct_opts structure
- *		    Must be NF_BPF_CT_OPTS_SZ (12)
+ *		    Must be NF_BPF_CT_OPTS_SZ (16) or 12
  */
 __bpf_kfunc struct nf_conn *
 bpf_xdp_ct_lookup(struct xdp_md *xdp_ctx, struct bpf_sock_tuple *bpf_tuple,
@@ -312,7 +354,7 @@ bpf_xdp_ct_lookup(struct xdp_md *xdp_ctx, struct bpf_sock_tuple *bpf_tuple,
  * @opts	- Additional options for allocation (documented above)
  *		    Cannot be NULL
  * @opts__sz	- Length of the bpf_ct_opts structure
- *		    Must be NF_BPF_CT_OPTS_SZ (12)
+ *		    Must be NF_BPF_CT_OPTS_SZ (16) or 12
  */
 __bpf_kfunc struct nf_conn___init *
 bpf_skb_ct_alloc(struct __sk_buff *skb_ctx, struct bpf_sock_tuple *bpf_tuple,
@@ -347,7 +389,7 @@ bpf_skb_ct_alloc(struct __sk_buff *skb_ctx, struct bpf_sock_tuple *bpf_tuple,
  * @opts	- Additional options for lookup (documented above)
  *		    Cannot be NULL
  * @opts__sz	- Length of the bpf_ct_opts structure
- *		    Must be NF_BPF_CT_OPTS_SZ (12)
+ *		    Must be NF_BPF_CT_OPTS_SZ (16) or 12
  */
 __bpf_kfunc struct nf_conn *
 bpf_skb_ct_lookup(struct __sk_buff *skb_ctx, struct bpf_sock_tuple *bpf_tuple,
Signature · 1 line
2.34.1
author
#1 lore raw

PATCH v3 2/2 selftests/bpf: Update tests for new ct zone opts for nf_conntrack kfuncs

Add test for allocating and looking up ct entry in a non-default ct zone with kfuncs bpf_{xdp,skb}_ct_alloc and bpf_{xdp,skb}_ct_lookup.

Add negative tests for looking up ct entry in a different ct zone to where it was allocated and with a different direction.

---
v2 -> v3:
- Test both old and new bpf_ct_opts struct definitions
- Restore test for reserved options
- Add test for ct_zone_dir

v1 -> v2:
- Separate test changes into different patch
- Add test for allocating/looking up entry in non-default ct zone

---
 tools/testing/selftests/bpf/config            |   1 +
 .../testing/selftests/bpf/prog_tests/bpf_nf.c |   6 +
 .../testing/selftests/bpf/progs/test_bpf_nf.c | 117 ++++++++++++++++--
 3 files changed, 114 insertions(+), 10 deletions(-)
tools/testing/selftests/bpf/config +1 -0
index eeabd798bc3a..2fb16da78dce 100644
@@ -80,6 +80,7 @@ CONFIG_NETFILTER_XT_TARGET_CT=y
 CONFIG_NETKIT=y
 CONFIG_NF_CONNTRACK=y
 CONFIG_NF_CONNTRACK_MARK=y
+CONFIG_NF_CONNTRACK_ZONES=y
 CONFIG_NF_DEFRAG_IPV4=y
 CONFIG_NF_DEFRAG_IPV6=y
 CONFIG_NF_NAT=y
tools/testing/selftests/bpf/prog_tests/bpf_nf.c +6 -0
index b30ff6b3b81a..b73401a71e4a 100644
@@ -122,6 +122,12 @@ static void test_bpf_nf_ct(int mode)
 	ASSERT_EQ(skel->bss->test_exist_lookup_mark, 43, "Test existing connection lookup ctmark");
 	ASSERT_EQ(skel->data->test_snat_addr, 0, "Test for source natting");
 	ASSERT_EQ(skel->data->test_dnat_addr, 0, "Test for destination natting");
+	ASSERT_EQ(skel->data->test_ct_zone_id_alloc_entry, 0, "Test for alloc new entry in specified ct zone");
+	ASSERT_EQ(skel->data->test_ct_zone_id_insert_entry, 0, "Test for insert new entry in specified ct zone");
+	ASSERT_EQ(skel->data->test_ct_zone_id_succ_lookup, 0, "Test for successful lookup in specified ct_zone");
+	ASSERT_EQ(skel->bss->test_ct_zone_dir_enoent_lookup, -ENOENT, "Test ENOENT for lookup with wrong ct zone dir");
+	ASSERT_EQ(skel->bss->test_ct_zone_id_enoent_lookup, -ENOENT, "Test ENOENT for lookup in wrong ct zone");
+
 end:
 	if (client_fd != -1)
 		close(client_fd);
tools/testing/selftests/bpf/progs/test_bpf_nf.c +107 -10

157 lines not on this page · raw message

Signature · 1 line
2.34.1
#2 lore raw

Re: [PATCH bpf-next v3 2/2] selftests/bpf: Update tests for new ct zone opts for nf_conntrack kfuncs

Quoted 6 lines · Brad Cowie
tools/testing/selftests/bpf/progs/test_bpf_nf.c +0 -2
@@ -84,16 +102,6 @@ nf_ct_test(struct nf_conn *(*lookup_fn)(void *, struct bpf_sock_tuple *, u32,
  	else
  		test_einval_bpf_tuple = opts_def.error;
 
-	opts_def.reserved[0] = 1;
-	ct = lookup_fn(ctx, &bpf_tuple, sizeof(bpf_tuple.ipv4), &opts_def,

tools/testing/selftests/bpf/progs/test_bpf_nf.c nf_ct_test(struct nf_conn *(*lookup_fn)(void *, struct bpf_sock_tuple *, u32, · line 106

Quoted 10 lines
tools/testing/selftests/bpf/progs/test_bpf_nf.c +0 -8
-		       sizeof(opts_def));
-	opts_def.reserved[0] = 0;
-	opts_def.l4proto = IPPROTO_TCP;
-	if (ct)
-		bpf_ct_release(ct);
-	else
-		test_einval_reserved = opts_def.error;
-
  	opts_def.netns_id = -2;
  	ct = lookup_fn(ctx, &bpf_tuple, sizeof(bpf_tuple.ipv4), &opts_def,

This non-zero reserved[0] test is still valid and useful. How about create a new test_einval_reserved_new for testing the new struct?

pw-bot: cr

[ Sorry for the delay. I have some backlog. ].