Skip to contentSkip to conversation

xfrm: interface: Add unstable helpers for XFRM metadata

PATCH Superseded Eyal Birger 2 participants 8 messages Dec 1, 2022 – Dec 2, 2022

Subsystems (6)

v3 of 6 Superseded Latest is v6 →

lore
Share & tools
lore
#0 lore raw

v3 cover xfrm: interface: Add unstable helpers for XFRM metadata

This patch series adds xfrm metadata helpers using the unstable kfunc call interface for the TC-BPF hooks.

This allows steering traffic towards different IPsec connections based on logic implemented in bpf programs.

The helpers are integrated into the xfrm_interface module. For this purpose the main functionality of this module is moved to xfrm_interface_core.c.

---

Series changes in v3:
- tag bpf-next tree instead of ipsec-next
- add IFLA_XFRM_COLLECT_METADATA sync patch

Eyal Birger (4):
xfrm: interface: rename xfrm_interface.c to xfrm_interface_core.c
xfrm: interface: Add unstable helpers for setting/getting XFRM
metadata from TC-BPF
tools: add IFLA_XFRM_COLLECT_METADATA to uapi/linux/if_link.h selftests/bpf: add xfrm_info tests

 include/net/dst_metadata.h                    |   1 +
 include/net/xfrm.h                            |  20 +
 net/core/dst.c                                |   8 +-
 net/xfrm/Makefile                             |   8 +
 net/xfrm/xfrm_interface_bpf.c                 |  99 +++++
 ...xfrm_interface.c => xfrm_interface_core.c} |  15 +
 tools/include/uapi/linux/if_link.h            |   1 +
 tools/testing/selftests/bpf/config            |   2 +
 .../selftests/bpf/prog_tests/xfrm_info.c      | 365 ++++++++++++++++++
 tools/testing/selftests/bpf/progs/xfrm_info.c |  40 ++
 10 files changed, 557 insertions(+), 2 deletions(-)
 create mode 100644 net/xfrm/xfrm_interface_bpf.c
 rename net/xfrm/{xfrm_interface.c => xfrm_interface_core.c} (98%)
 create mode 100644 tools/testing/selftests/bpf/prog_tests/xfrm_info.c
 create mode 100644 tools/testing/selftests/bpf/progs/xfrm_info.c
Signature · 1 line
2.34.1
#1 lore raw

PATCH v3 1/4 xfrm: interface: rename xfrm_interface.c to xfrm_interface_core.c

This change allows adding additional files to the xfrm_interface module.

---
 net/xfrm/Makefile                                    | 2 ++
 net/xfrm/{xfrm_interface.c => xfrm_interface_core.c} | 0
 2 files changed, 2 insertions(+)
 rename net/xfrm/{xfrm_interface.c => xfrm_interface_core.c} (100%)
net/xfrm/Makefile +2 -0
index 494aa744bfb9..08a2870fdd36 100644
@@ -3,6 +3,8 @@
 # Makefile for the XFRM subsystem.
 #
 
+xfrm_interface-$(CONFIG_XFRM_INTERFACE) += xfrm_interface_core.o
+
 obj-$(CONFIG_XFRM) := xfrm_policy.o xfrm_state.o xfrm_hash.o \
 		      xfrm_input.o xfrm_output.o \
 		      xfrm_sysctl.o xfrm_replay.o xfrm_device.o
net/xfrm/xfrm_interface_core.c renamed from net/xfrm/xfrm_interface.c +0 -0
similarity index 100%
rename from net/xfrm/xfrm_interface.c
rename to net/xfrm/xfrm_interface_core.c
Signature · 1 line
2.34.1
#2 lore raw

PATCH v3 2/4 xfrm: interface: Add unstable helpers for setting/getting XFRM metadata from TC-BPF

In reply to Eyal Birger

This change adds xfrm metadata helpers using the unstable kfunc call interface for the TC-BPF hooks. This allows steering traffic towards different IPsec connections based on logic implemented in bpf programs.

This object is built based on the availabilty of BTF debug info.

The metadata percpu dsts used on TX take ownership of the original skb dsts so that they may be used as part of the xfrm transmittion logic - e.g. for MTU calculations.

---

v3:
- remove redunant memset() as suggested by Martin KaFai Lau
- remove __exit annotation from cleanup() function as it's used from an __init function

v2: changed added following points raised by Martin KaFai Lau:
- make sure dst is refcounted prior to caching
- free dst_orig regardless of CONFIG_DST_CACHE
- call xfrm interface bpf cleanup in case of kfunc registration errors

---
 include/net/dst_metadata.h     |  1 +
 include/net/xfrm.h             | 20 +++++++
 net/core/dst.c                 |  8 ++-
 net/xfrm/Makefile              |  6 +++
 net/xfrm/xfrm_interface_bpf.c  | 99 ++++++++++++++++++++++++++++++++++
 net/xfrm/xfrm_interface_core.c | 15 ++++++
 6 files changed, 147 insertions(+), 2 deletions(-)
 create mode 100644 net/xfrm/xfrm_interface_bpf.c
include/net/dst_metadata.h +1 -0
index a454cf4327fe..1b7fae4c6b24 100644
@@ -26,6 +26,7 @@ struct macsec_info {
 struct xfrm_md_info {
 	u32 if_id;
 	int link;
+	struct dst_entry *dst_orig;
 };
 
 struct metadata_dst {
include/net/xfrm.h +20 -0
index e0cc6791c001..5e5fea3087b6 100644
@@ -2086,4 +2086,24 @@ static inline bool xfrm6_local_dontfrag(const struct sock *sk)
 	return false;
 }
 #endif
+
+#if (IS_BUILTIN(CONFIG_XFRM_INTERFACE) && IS_ENABLED(CONFIG_DEBUG_INFO_BTF)) || \
+    (IS_MODULE(CONFIG_XFRM_INTERFACE) && IS_ENABLED(CONFIG_DEBUG_INFO_BTF_MODULES))
+
+extern int register_xfrm_interface_bpf(void);
+extern void cleanup_xfrm_interface_bpf(void);
+
+#else
+
+static inline int register_xfrm_interface_bpf(void)
+{
+	return 0;
+}
+
+static inline void cleanup_xfrm_interface_bpf(void)
+{
+}
+
+#endif
+
 #endif	/* _NET_XFRM_H */
net/core/dst.c +6 -2
index bc9c9be4e080..bb14a0392388 100644
@@ -316,6 +316,8 @@ void metadata_dst_free(struct metadata_dst *md_dst)
 	if (md_dst->type == METADATA_IP_TUNNEL)
 		dst_cache_destroy(&md_dst->u.tun_info.dst_cache);
 #endif
+	if (md_dst->type == METADATA_XFRM)
+		dst_release(md_dst->u.xfrm_info.dst_orig);
 	kfree(md_dst);
 }
 EXPORT_SYMBOL_GPL(metadata_dst_free);
@@ -340,16 +342,18 @@ EXPORT_SYMBOL_GPL(metadata_dst_alloc_percpu);
 
 void metadata_dst_free_percpu(struct metadata_dst __percpu *md_dst)
 {
-#ifdef CONFIG_DST_CACHE
 	int cpu;
 
 	for_each_possible_cpu(cpu) {
 		struct metadata_dst *one_md_dst = per_cpu_ptr(md_dst, cpu);
 
+#ifdef CONFIG_DST_CACHE
 		if (one_md_dst->type == METADATA_IP_TUNNEL)
 			dst_cache_destroy(&one_md_dst->u.tun_info.dst_cache);
-	}
 #endif
+		if (one_md_dst->type == METADATA_XFRM)
+			dst_release(one_md_dst->u.xfrm_info.dst_orig);
+	}
 	free_percpu(md_dst);
 }
 EXPORT_SYMBOL_GPL(metadata_dst_free_percpu);
net/xfrm/Makefile +6 -0
index 08a2870fdd36..cd47f88921f5 100644
@@ -5,6 +5,12 @@
 
 xfrm_interface-$(CONFIG_XFRM_INTERFACE) += xfrm_interface_core.o
 
+ifeq ($(CONFIG_XFRM_INTERFACE),m)
+xfrm_interface-$(CONFIG_DEBUG_INFO_BTF_MODULES) += xfrm_interface_bpf.o
+else ifeq ($(CONFIG_XFRM_INTERFACE),y)
+xfrm_interface-$(CONFIG_DEBUG_INFO_BTF) += xfrm_interface_bpf.o
+endif
+
 obj-$(CONFIG_XFRM) := xfrm_policy.o xfrm_state.o xfrm_hash.o \
 		      xfrm_input.o xfrm_output.o \
 		      xfrm_sysctl.o xfrm_replay.o xfrm_device.o
net/xfrm/xfrm_interface_bpf.c new file +99 -0

99 lines not on this page · raw message

net/xfrm/xfrm_interface_core.c +15 -0
index 5a67b120c4db..1e1e8e965939 100644
@@ -396,6 +396,14 @@ xfrmi_xmit2(struct sk_buff *skb, struct net_device *dev, struct flowi *fl)
 
 		if_id = md_info->if_id;
 		fl->flowi_oif = md_info->link;
+		if (md_info->dst_orig) {
+			struct dst_entry *tmp_dst = dst;
+
+			dst = md_info->dst_orig;
+			skb_dst_set(skb, dst);
+			md_info->dst_orig = NULL;
+			dst_release(tmp_dst);
+		}
 	} else {
 		if_id = xi->p.if_id;
 	}
@@ -1162,12 +1170,18 @@ static int __init xfrmi_init(void)
 	if (err < 0)
 		goto rtnl_link_failed;
 
+	err = register_xfrm_interface_bpf();
+	if (err < 0)
+		goto kfunc_failed;
+
 	lwtunnel_encap_add_ops(&xfrmi_encap_ops, LWTUNNEL_ENCAP_XFRM);
 
 	xfrm_if_register_cb(&xfrm_if_cb);
 
 	return err;
 
+kfunc_failed:
+	rtnl_link_unregister(&xfrmi_link_ops);
 rtnl_link_failed:
 	xfrmi6_fini();
 xfrmi6_failed:
@@ -1183,6 +1197,7 @@ static void __exit xfrmi_fini(void)
 {
 	xfrm_if_unregister_cb();
 	lwtunnel_encap_del_ops(&xfrmi_encap_ops, LWTUNNEL_ENCAP_XFRM);
+	cleanup_xfrm_interface_bpf();
 	rtnl_link_unregister(&xfrmi_link_ops);
 	xfrmi4_fini();
 	xfrmi6_fini();
Signature · 1 line
2.34.1
#3 lore raw

Re: [PATCH bpf-next,v3 2/4] xfrm: interface: Add unstable helpers for setting/getting XFRM metadata from TC-BPF

Quoted 5 lines · Eyal Birger

This change adds xfrm metadata helpers using the unstable kfunc call interface for the TC-BPF hooks. This allows steering traffic towards different IPsec connections based on logic implemented in bpf programs.

This object is built based on the availabilty of BTF debug info.

s/availabilty/availability/

Quoted 3 lines

The metadata percpu dsts used on TX take ownership of the original skb dsts so that they may be used as part of the xfrm transmittion logic -

s/transmittion/transmission/

[ ... ]

Quoted 28 lines
net/xfrm/Makefile +6 -0
index 08a2870fdd36..cd47f88921f5 100644
@@ -5,6 +5,12 @@
 
  xfrm_interface-$(CONFIG_XFRM_INTERFACE) += xfrm_interface_core.o
 
+ifeq ($(CONFIG_XFRM_INTERFACE),m)
+xfrm_interface-$(CONFIG_DEBUG_INFO_BTF_MODULES) += xfrm_interface_bpf.o
+else ifeq ($(CONFIG_XFRM_INTERFACE),y)
+xfrm_interface-$(CONFIG_DEBUG_INFO_BTF) += xfrm_interface_bpf.o
+endif
+
  obj-$(CONFIG_XFRM) := xfrm_policy.o xfrm_state.o xfrm_hash.o \
  		      xfrm_input.o xfrm_output.o \
  		      xfrm_sysctl.o xfrm_replay.o xfrm_device.o
net/xfrm/xfrm_interface_bpf.c new file +5 -0
new file mode 100644
index 000000000000..7938bf6cbb32
@@ -0,0 +1,99 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/* Unstable XFRM Helpers for TC-BPF hook
+ *
+ * These are called from SCHED_CLS BPF programs. Note that it is
+ * allowed to break compatibility for these functions since the interface they

net/xfrm/xfrm_interface_bpf.c line 15

Quoted 10 lines
net/xfrm/xfrm_interface_bpf.c new file +10 -0
+ * are exposed through to BPF programs is explicitly unstable.
+ */
+
+#include <linux/bpf.h>
+#include <linux/btf_ids.h>
+
+#include <net/dst_metadata.h>
+#include <net/xfrm.h>
+
+struct bpf_xfrm_info {

Documentation is needed here and...

net/xfrm/xfrm_interface_bpf.c line 26

Quoted 11 lines
net/xfrm/xfrm_interface_bpf.c new file +11 -0
+	u32 if_id;
+	int link;
+};
+
+static struct metadata_dst __percpu *xfrm_md_dst;
+__diag_push();
+__diag_ignore_all("-Wmissing-prototypes",
+		  "Global functions as their definitions will be in xfrm_interface BTF");
+
+__used noinline
+int bpf_skb_get_xfrm_info(struct __sk_buff *skb_ctx, struct bpf_xfrm_info *to)

... also this get and the following set kfunc.

Quoted 54 lines
net/xfrm/xfrm_interface_bpf.c new file +54 -0
+{
+	struct sk_buff *skb = (struct sk_buff *)skb_ctx;
+	struct xfrm_md_info *info;
+
+	memset(to, 0, sizeof(*to));
+
+	info = skb_xfrm_md_info(skb);
+	if (!info)
+		return -EINVAL;
+
+	to->if_id = info->if_id;
+	to->link = info->link;
+	return 0;
+}
+
+__used noinline
+int bpf_skb_set_xfrm_info(struct __sk_buff *skb_ctx,
+			  const struct bpf_xfrm_info *from)
+{
+	struct sk_buff *skb = (struct sk_buff *)skb_ctx;
+	struct metadata_dst *md_dst;
+	struct xfrm_md_info *info;
+
+	if (unlikely(skb_metadata_dst(skb)))
+		return -EINVAL;
+
+	md_dst = this_cpu_ptr(xfrm_md_dst);
+
+	info = &md_dst->u.xfrm_info;
+
+	info->if_id = from->if_id;
+	info->link = from->link;
+	skb_dst_force(skb);
+	info->dst_orig = skb_dst(skb);
+
+	dst_hold((struct dst_entry *)md_dst);
+	skb_dst_set(skb, (struct dst_entry *)md_dst);
+	return 0;
+}
+
+__diag_pop()
+
+BTF_SET8_START(xfrm_ifc_kfunc_set)
+BTF_ID_FLAGS(func, bpf_skb_get_xfrm_info)
+BTF_ID_FLAGS(func, bpf_skb_set_xfrm_info)
+BTF_SET8_END(xfrm_ifc_kfunc_set)
+
+static const struct btf_kfunc_id_set xfrm_interface_kfunc_set = {
+	.owner = THIS_MODULE,
+	.set   = &xfrm_ifc_kfunc_set,
+};
+
+int __init register_xfrm_interface_bpf(void)
+{

net/xfrm/xfrm_interface_bpf.c line 90

Quoted 10 lines
net/xfrm/xfrm_interface_bpf.c new file +10 -0
+	int err;
+
+	xfrm_md_dst = metadata_dst_alloc_percpu(0, METADATA_XFRM,
+						GFP_KERNEL);
+	if (!xfrm_md_dst)
+		return -ENOMEM;
+	err = register_btf_kfunc_id_set(BPF_PROG_TYPE_SCHED_CLS,
+					&xfrm_interface_kfunc_set);
+	if (err < 0) {
+		cleanup_xfrm_interface_bpf();

nit. Directly call metadata_dst_free_percpu(xfrm_md_dst), easier to read.

Quoted 9 lines
net/xfrm/xfrm_interface_bpf.c new file +9 -0
+		return err;
+	}
+	return 0;
+}
+
+void cleanup_xfrm_interface_bpf(void)
+{
+	metadata_dst_free_percpu(xfrm_md_dst);
+}
#4 lore raw

Re: [PATCH bpf-next,v3 2/4] xfrm: interface: Add unstable helpers for setting/getting XFRM metadata from TC-BPF

In reply to Eyal Birger

net/xfrm/xfrm_interface_bpf.c line 31

Quoted 6 lines · Eyal Birger
net/xfrm/xfrm_interface_bpf.c new file +6 -0
+int bpf_skb_get_xfrm_info(struct __sk_buff *skb_ctx, struct bpf_xfrm_info *to)
+{
+	struct sk_buff *skb = (struct sk_buff *)skb_ctx;
+	struct xfrm_md_info *info;
+
+	memset(to, 0, sizeof(*to));

This zero-ing is also not needed. verifier ensures "to" is initialized before calling this kfunc.

Quoted 9 lines
net/xfrm/xfrm_interface_bpf.c new file +9 -0
+
+	info = skb_xfrm_md_info(skb);
+	if (!info)
+		return -EINVAL;
+
+	to->if_id = info->if_id;
+	to->link = info->link;
+	return 0;
+}
#5 lore raw

PATCH v3 3/4 tools: add IFLA_XFRM_COLLECT_METADATA to uapi/linux/if_link.h

In reply to Eyal Birger

Needed for XFRM metadata tests.

---
 tools/include/uapi/linux/if_link.h | 1 +
 1 file changed, 1 insertion(+)
tools/include/uapi/linux/if_link.h +1 -0
index 0242f31e339c..901d98b865a1 100644
@@ -673,6 +673,7 @@ enum {
 	IFLA_XFRM_UNSPEC,
 	IFLA_XFRM_LINK,
 	IFLA_XFRM_IF_ID,
+	IFLA_XFRM_COLLECT_METADATA,
 	__IFLA_XFRM_MAX
 };
 
Signature · 1 line
2.34.1
#6 lore raw

PATCH v3 4/4 selftests/bpf: add xfrm_info tests

In reply to Eyal Birger

Test the xfrm_info kfunc helpers.

Note: the tests require support for xfrmi "external" mode in iproute2.

The test setup creates three name spaces - NS0, NS1, NS2.

XFRM tunnels are setup between NS0 and the two other NSs.

The kfunc helpers are used to steer traffic from NS0 to the other NSs based on a userspace populated bpf global variable and validate that the return traffic had arrived from the desired NS.

---
v3: changes suggested by Martin KaFai Lau:
- rename test_xfrm_info{,_kern}.c -> xfrm_info.c
- avoid running in a separate thread
- simplify test setup
- verify underlay/overlay success
- create NS0 xfrmi external device using netlink to avoid dependency on newer iproute2
- remove use of bpf_trace_printk() in test programs
- add "preserve_access_index" attribute annotation in test program struct definition
- use bss variables access instead of a map for userspace/bpf interface

v2:
- use an lwt route in NS1 for testing that flow as well
- indendation fix

---
 tools/testing/selftests/bpf/config            |   2 +
 .../selftests/bpf/prog_tests/xfrm_info.c      | 365 ++++++++++++++++++
 tools/testing/selftests/bpf/progs/xfrm_info.c |  40 ++
 3 files changed, 407 insertions(+)
 create mode 100644 tools/testing/selftests/bpf/prog_tests/xfrm_info.c
 create mode 100644 tools/testing/selftests/bpf/progs/xfrm_info.c
tools/testing/selftests/bpf/config +2 -0
index f9034ea00bc9..3543c76cef56 100644
@@ -23,6 +23,7 @@ CONFIG_IKCONFIG_PROC=y
 CONFIG_IMA=y
 CONFIG_IMA_READ_POLICY=y
 CONFIG_IMA_WRITE_POLICY=y
+CONFIG_INET_ESP=y
 CONFIG_IP_NF_FILTER=y
 CONFIG_IP_NF_RAW=y
 CONFIG_IP_NF_TARGET_SYNPROXY=y
@@ -74,3 +75,4 @@ CONFIG_TEST_BPF=y
 CONFIG_USERFAULTFD=y
 CONFIG_VXLAN=y
 CONFIG_XDP_SOCKETS=y
+CONFIG_XFRM_INTERFACE=y
tools/testing/selftests/bpf/prog_tests/xfrm_info.c new file +365 -0

365 lines not on this page · raw message

tools/testing/selftests/bpf/progs/xfrm_info.c new file +40 -0
new file mode 100644
index 000000000000..908579310bf9
@@ -0,0 +1,40 @@
+// SPDX-License-Identifier: GPL-2.0
+#include <linux/bpf.h>
+#include <linux/pkt_cls.h>
+#include <bpf/bpf_helpers.h>
+
+__u32 req_if_id;
+__u32 resp_if_id;
+
+struct bpf_xfrm_info {
+	__u32 if_id;
+	int link;
+} __attribute__((preserve_access_index));
+
+int bpf_skb_set_xfrm_info(struct __sk_buff *skb_ctx,
+			  const struct bpf_xfrm_info *from) __ksym;
+int bpf_skb_get_xfrm_info(struct __sk_buff *skb_ctx,
+			  struct bpf_xfrm_info *to) __ksym;
+
+SEC("tc")
+int set_xfrm_info(struct __sk_buff *skb)
+{
+	struct bpf_xfrm_info info = { .if_id = req_if_id };
+
+	return bpf_skb_set_xfrm_info(skb, &info) ? TC_ACT_SHOT : TC_ACT_UNSPEC;
+}
+
+SEC("tc")
+int get_xfrm_info(struct __sk_buff *skb)
+{
+	struct bpf_xfrm_info info = {};
+
+	if (bpf_skb_get_xfrm_info(skb, &info) < 0)
+		return TC_ACT_SHOT;
+
+	resp_if_id = info.if_id;
+
+	return TC_ACT_UNSPEC;
+}
+
+char _license[] SEC("license") = "GPL";
Signature · 1 line
2.34.1
#7 lore raw

Re: [PATCH bpf-next,v3 4/4] selftests/bpf: add xfrm_info tests

Quoted 3 lines · Eyal Birger

Test the xfrm_info kfunc helpers.

Note: the tests require support for xfrmi "external" mode in iproute2.

Not needed now. Please update the commit message.

The test is failing on platform that has no kfunc support yet. Please check the BPF CI result in patchwork after posting to ensure the tests run well:
https://patchwork.kernel.org/project/netdevbpf/patch/[email protected]/

This test needs to be added to the DENYLIST.<arch> for the not yet supported platform. Please refer to selftests/bpf/README.rst for details.

[ ... ]

tools/testing/selftests/bpf/prog_tests/xfrm_info.c line 94

Quoted 13 lines
tools/testing/selftests/bpf/prog_tests/xfrm_info.c new file +13 -0
+#define SYS_NOFAIL(fmt, ...)					\
+	({							\
+		char cmd[1024];					\
+		snprintf(cmd, sizeof(cmd), fmt, ##__VA_ARGS__);	\
+		system(cmd);					\
+	})
+
+static int attach_tc_prog(struct bpf_tc_hook *hook, int igr_fd, int egr_fd)
+{
+	DECLARE_LIBBPF_OPTS(bpf_tc_opts, opts1, .handle = 1,
+			    .priority = 1, .prog_fd = igr_fd);
+	DECLARE_LIBBPF_OPTS(bpf_tc_opts, opts2, .handle = 1,
+			    .priority = 1, .prog_fd = egr_fd);

s/DECLARE_LIBBPF_OPTS/LIBBPF_OPTS/

DECLARE_ is a legacy naming in libbpf_legacy.h

[ ... ]

tools/testing/selftests/bpf/prog_tests/xfrm_info.c line 229

Quoted 10 lines
tools/testing/selftests/bpf/prog_tests/xfrm_info.c new file +10 -0
+static int setup_xfrmi_external_dev(const char *ns)
+{
+	struct {
+		struct nlmsghdr nh;
+		struct ifinfomsg info;
+		unsigned char data[128];
+	} req;
+	struct rtattr *link_info, *info_data;
+	struct nstoken *nstoken;
+	int ret = -1, sock = 0;

sock = -1;

Quoted 7 lines
tools/testing/selftests/bpf/prog_tests/xfrm_info.c new file +7 -0
+	struct nlmsghdr *nh;
+
+	memset(&req, 0, sizeof(req));
+	nh = &req.nh;
+	nh->nlmsg_len = NLMSG_LENGTH(sizeof(req.info));
+	nh->nlmsg_type = RTM_NEWLINK;
+	nh->nlmsg_flags |= NLM_F_CREATE | NLM_F_REQUEST;

tools/testing/selftests/bpf/prog_tests/xfrm_info.c line 246

Quoted 10 lines
tools/testing/selftests/bpf/prog_tests/xfrm_info.c new file +10 -0
+
+	rtattr_add_str(nh, IFLA_IFNAME, "ipsec0");
+	link_info = rtattr_begin(nh, IFLA_LINKINFO);
+	rtattr_add_str(nh, IFLA_INFO_KIND, "xfrm");
+	info_data = rtattr_begin(nh, IFLA_INFO_DATA);
+	rtattr_add(nh, IFLA_XFRM_COLLECT_METADATA, 0);
+	rtattr_end(nh, info_data);
+	rtattr_end(nh, link_info);
+
+	nstoken = open_netns(ns);

Please check error.

tools/testing/selftests/bpf/prog_tests/xfrm_info.c line 249

Quoted 3 lines
tools/testing/selftests/bpf/prog_tests/xfrm_info.c new file +3 -0
+
+	sock = socket(AF_NETLINK, SOCK_RAW | SOCK_CLOEXEC, NETLINK_ROUTE);
+	if (!ASSERT_GT(sock, 0, "netlink socket"))

s/_GT/_GE/

Review

Quoted 1 line
(quoted diff) +1 -0
+		goto Exit;

Please run checkpatch.pl...

CHECK: Avoid CamelCase: <Exit>
#301: FILE: tools/testing/selftests/bpf/prog_tests/xfrm_info.c:250:
+ goto Exit;

tools/testing/selftests/bpf/prog_tests/xfrm_info.c line 257

Quoted 7 lines
tools/testing/selftests/bpf/prog_tests/xfrm_info.c new file +7 -0
+	ret = send(sock, nh, nh->nlmsg_len, 0);
+	if (!ASSERT_EQ(ret, nh->nlmsg_len, "netlink send length"))
+		goto Exit;
+
+	ret = 0;
+Exit:
+	if (sock)

if (sock != -1) ...

Quoted 37 lines
tools/testing/selftests/bpf/prog_tests/xfrm_info.c new file +37 -0
+		close(sock);
+	close_netns(nstoken);
+	return ret;
+}
+
+static int config_overlay(void)
+{
+	if (setup_xfrm_tunnel(NS0, NS1, IP4_ADDR_VETH01, IP4_ADDR_VETH10,
+			      IF_ID_0_TO_1, IF_ID_1))
+		goto fail;
+	if (setup_xfrm_tunnel(NS0, NS2, IP4_ADDR_VETH02, IP4_ADDR_VETH20,
+			      IF_ID_0_TO_2, IF_ID_2))
+		goto fail;
+
+	/* Older iproute2 doesn't support this option */
+	if (!ASSERT_OK(setup_xfrmi_external_dev(NS0), "xfrmi"))
+		goto fail;
+
+	SYS("ip -net " NS0 " addr add 192.168.1.100/24 dev ipsec0");
+	SYS("ip -net " NS0 " link set dev ipsec0 up");
+
+	SYS("ip -net " NS1 " link add ipsec0 type xfrm if_id %d", IF_ID_1);
+	SYS("ip -net " NS1 " addr add 192.168.1.200/24 dev ipsec0");
+	SYS("ip -net " NS1 " link set dev ipsec0 up");
+
+	SYS("ip -net " NS2 " link add ipsec0 type xfrm if_id %d", IF_ID_2);
+	SYS("ip -net " NS2 " addr add 192.168.1.200/24 dev ipsec0");
+	SYS("ip -net " NS2 " link set dev ipsec0 up");
+
+	return 0;
+fail:
+	return -1;
+}
+
+static int test_ping(int family, const char *addr)
+{
+	SYS("%s %s %s > /dev/null", ping_command(family), PING_ARGS, addr);

tools/testing/selftests/bpf/prog_tests/xfrm_info.c line 304

Quoted 10 lines
tools/testing/selftests/bpf/prog_tests/xfrm_info.c new file +10 -0
+	return 0;
+fail:
+	return -1;
+}
+
+static int test_xfrm_ping(struct xfrm_info *skel, u32 if_id)
+{
+	skel->bss->req_if_id = if_id;
+
+	if (test_ping(AF_INET, "192.168.1.200"))

nit. Directly do SYS() here to avoid another reading detour to test_ping() which is almost a one liner and only used once here.

Quoted 4 lines
tools/testing/selftests/bpf/prog_tests/xfrm_info.c new file +4 -0
+		return -1;
+
+	if (!ASSERT_EQ(skel->bss->resp_if_id, if_id, "if_id"))
+		return -1;

tools/testing/selftests/bpf/prog_tests/xfrm_info.c line 318

Quoted 10 lines
tools/testing/selftests/bpf/prog_tests/xfrm_info.c new file +10 -0
+
+	return 0;
+}
+
+static void _test_xfrm_info(void)
+{
+	int get_xfrm_info_prog_fd, set_xfrm_info_prog_fd;
+	struct xfrm_info *skel = NULL;
+	struct nstoken *nstoken = NULL;
+	int ifindex = -1;

nit. Unnecessary init. Nonthing to cleanup at the label "done:" and it is not a return value also. will be easier to review.

tools/testing/selftests/bpf/prog_tests/xfrm_info.c line 326

Quoted 8 lines
tools/testing/selftests/bpf/prog_tests/xfrm_info.c new file +8 -0
+	DECLARE_LIBBPF_OPTS(bpf_tc_hook, tc_hook,
+			    .attach_point = BPF_TC_INGRESS);
+
+	/* load and attach bpf progs to ipsec dev tc hook point */
+	skel = xfrm_info__open_and_load();
+	if (!ASSERT_OK_PTR(skel, "xfrm_info__open_and_load"))
+		goto done;
+	nstoken = open_netns(NS0);

Check error.

Quoted 20 lines
tools/testing/selftests/bpf/prog_tests/xfrm_info.c new file +20 -0
+	ifindex = if_nametoindex("ipsec0");
+	if (!ASSERT_NEQ(ifindex, 0, "ipsec0 ifindex"))
+		goto done;
+	tc_hook.ifindex = ifindex;
+	set_xfrm_info_prog_fd = bpf_program__fd(skel->progs.set_xfrm_info);
+	get_xfrm_info_prog_fd = bpf_program__fd(skel->progs.get_xfrm_info);
+	if (!ASSERT_GE(set_xfrm_info_prog_fd, 0, "bpf_program__fd"))
+		goto done;
+	if (!ASSERT_GE(get_xfrm_info_prog_fd, 0, "bpf_program__fd"))
+		goto done;
+	if (attach_tc_prog(&tc_hook, get_xfrm_info_prog_fd,
+			   set_xfrm_info_prog_fd))
+		goto done;
+	if (!ASSERT_EQ(test_xfrm_ping(skel, IF_ID_0_TO_1), 0, "ping " NS1))
+		goto done;
+	if (!ASSERT_EQ(test_xfrm_ping(skel, IF_ID_0_TO_2), 0, "ping " NS2))
+		goto done;
+
+done:
+	if (nstoken)

tools/testing/selftests/bpf/prog_tests/xfrm_info.c line 356

Quoted 10 lines
tools/testing/selftests/bpf/prog_tests/xfrm_info.c new file +10 -0
+		close_netns(nstoken);
+	if (skel)
+		xfrm_info__destroy(skel);
+}
+
+void test_xfrm_info(void)
+{
+	cleanup();
+
+	if (!ASSERT_OK(config_underlay(), "config_underlay"))

cleanup() is needed on error.

tools/testing/selftests/bpf/prog_tests/xfrm_info.c line 358

Quoted 2 lines
tools/testing/selftests/bpf/prog_tests/xfrm_info.c new file +2 -0
+		return;
+	if (!ASSERT_OK(config_overlay(), "config_overlay"))

Same here.

Quoted 27 lines
tools/testing/selftests/bpf/prog_tests/xfrm_info.c new file +7 -0
+		return;
+
+	if (test__start_subtest("xfrm_info"))
+		_test_xfrm_info();
+
+	cleanup();
+}
tools/testing/selftests/bpf/progs/xfrm_info.c new file +14 -0
new file mode 100644
index 000000000000..908579310bf9
@@ -0,0 +1,40 @@
+// SPDX-License-Identifier: GPL-2.0
+#include <linux/bpf.h>
+#include <linux/pkt_cls.h>
+#include <bpf/bpf_helpers.h>
+
+__u32 req_if_id;
+__u32 resp_if_id;
+
+struct bpf_xfrm_info {
+	__u32 if_id;
+	int link;
+} __attribute__((preserve_access_index));
+
+int bpf_skb_set_xfrm_info(struct __sk_buff *skb_ctx,

tools/testing/selftests/bpf/progs/xfrm_info.c line 24

Quoted 10 lines
tools/testing/selftests/bpf/progs/xfrm_info.c new file +10 -0
+			  const struct bpf_xfrm_info *from) __ksym;
+int bpf_skb_get_xfrm_info(struct __sk_buff *skb_ctx,
+			  struct bpf_xfrm_info *to) __ksym;
+
+SEC("tc")
+int set_xfrm_info(struct __sk_buff *skb)
+{
+	struct bpf_xfrm_info info = { .if_id = req_if_id };
+
+	return bpf_skb_set_xfrm_info(skb, &info) ? TC_ACT_SHOT : TC_ACT_UNSPEC;

Add these TC_ACT_* to bpf_tracing_net.h and then vmlinux.h can be used. Take a look at some of the bpf_tracing_net.h in selftests.

Quoted 16 lines
tools/testing/selftests/bpf/progs/xfrm_info.c new file +16 -0
+}
+
+SEC("tc")
+int get_xfrm_info(struct __sk_buff *skb)
+{
+	struct bpf_xfrm_info info = {};
+
+	if (bpf_skb_get_xfrm_info(skb, &info) < 0)
+		return TC_ACT_SHOT;
+
+	resp_if_id = info.if_id;
+
+	return TC_ACT_UNSPEC;
+}
+
+char _license[] SEC("license") = "GPL";