Skip to contentSkip to conversation

net: new ipv6 header extension parsing function

PATCH Superseded Marcin Smoczynski 4 participants 23 messages May 8, 2019 – Jul 2, 2019

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

PATCH 1/3 net: new ipv6 header extension parsing function

Introduce new function for IPv6 header extension parsing able to determine extension length and next protocol number.

This function is helpful when implementing IPv6 header traversing.

---
 lib/librte_net/rte_ip.h | 49 +++++++++++++++++++++++++++++++++++++++++
 1 file changed, 49 insertions(+)
lib/librte_net/rte_ip.h +49 -0
index f9b909090..be64da662 100644
@@ -425,6 +425,55 @@ rte_ipv6_udptcp_cksum(const struct ipv6_hdr *ipv6_hdr, const void *l4_hdr)
 	return (uint16_t)cksum;
 }
 
+/* IPv6 fragmentation header size */
+#define RTE_IPV6_FRAG_HDR_SIZE 8
+
+/**
+ * Parse next IPv6 header extension
+ *
+ * This function checks if proto number is an IPv6 extensions and parses its
+ * data if so, providing information on next header and extension length.
+ *
+ * @param p
+ *   Pointer to an extension raw data.
+ * @param proto
+ *   Protocol number extracted from the "next header" field from
+ *   the IPv6 header or the previous extension.
+ * @param ext_len
+ *   Extension data length.
+ * @return
+ *   next protocol number if proto is an IPv6 extension, -EINVAL otherwise
+ */
+static inline int __rte_experimental
+rte_ipv6_get_next_ext(uint8_t *p, int proto, size_t *ext_len)
+{
+	int next_proto;
+
+	switch (proto) {
+	case IPPROTO_AH:
+		next_proto = *p++;
+		*ext_len = (*p + 2) * sizeof(uint32_t);
+		break;
+
+	case IPPROTO_HOPOPTS:
+	case IPPROTO_ROUTING:
+	case IPPROTO_DSTOPTS:
+		next_proto = *p++;
+		*ext_len = (*p + 1) * sizeof(uint64_t);
+		break;
+
+	case IPPROTO_FRAGMENT:
+		next_proto = *p;
+		*ext_len = RTE_IPV6_FRAG_HDR_SIZE;
+		break;
+
+	default:
+		return -EINVAL;
+	}
+
+	return next_proto;
+}
+
 #ifdef __cplusplus
 }
 #endif
Signature · 1 line
2.21.0.windows.1
#1 lore raw

PATCH 2/3 ipsec: fix transport mode for ipv6 with extensions

Reconstructing IPv6 header after encryption or decryption requires updating 'next header' value in the preceding protocol header, which is determined by parsing IPv6 header and iteratively looking for next IPv6 header extension.

It is required that 'l3_len' in the mbuf metadata contains a total length of the IPv6 header with header extensions up to ESP header.

---
 lib/Makefile           |  3 ++-
 lib/librte_ipsec/iph.h | 55 ++++++++++++++++++++++++++++++++++++------
 2 files changed, 49 insertions(+), 9 deletions(-)
lib/Makefile +2 -1
index 791e0d991..3ad579f68 100644
@@ -108,7 +108,8 @@ DEPDIRS-librte_gso += librte_mempool
 DIRS-$(CONFIG_RTE_LIBRTE_BPF) += librte_bpf
 DEPDIRS-librte_bpf := librte_eal librte_mempool librte_mbuf librte_ethdev
 DIRS-$(CONFIG_RTE_LIBRTE_IPSEC) += librte_ipsec
-DEPDIRS-librte_ipsec := librte_eal librte_mbuf librte_cryptodev librte_security
+DEPDIRS-librte_ipsec := librte_eal librte_mbuf librte_cryptodev librte_security \
+			librte_net
 DIRS-$(CONFIG_RTE_LIBRTE_TELEMETRY) += librte_telemetry
 DEPDIRS-librte_telemetry := librte_eal librte_metrics librte_ethdev
 DIRS-$(CONFIG_RTE_LIBRTE_RCU) += librte_rcu
lib/librte_ipsec/iph.h +47 -8

77 lines not on this page · raw message

Signature · 1 line
2.21.0.windows.1
Acked-by on v1 2/3
#2 lore raw

Re: [dpdk-dev] [PATCH 2/3] ipsec: fix transport mode for ipv6 with extensions

Quoted 8 lines

Reconstructing IPv6 header after encryption or decryption requires updating 'next header' value in the preceding protocol header, which is determined by parsing IPv6 header and iteratively looking for next IPv6 header extension.

Quoted 5 lines

It is required that 'l3_len' in the mbuf metadata contains a total length of the IPv6 header with header extensions up to ESP header.

---

#3 lore raw

Re: [dpdk-dev] [PATCH 2/3] ipsec: fix transport mode for ipv6 with extensions

In reply to Marcin Smoczynski

Hi Marcin,

Quoted 20 lines

Reconstructing IPv6 header after encryption or decryption requires updating 'next header' value in the preceding protocol header, which is determined by parsing IPv6 header and iteratively looking for next IPv6 header extension.

It is required that 'l3_len' in the mbuf metadata contains a total length of the IPv6 header with header extensions up to ESP header.

---
 lib/Makefile           |  3 ++-
 lib/librte_ipsec/iph.h | 55 ++++++++++++++++++++++++++++++++++++------
 2 files changed, 49 insertions(+), 9 deletions(-)
lib/Makefile +0 -0
index 791e0d991..3ad579f68 100644

lib/Makefile DEPDIRS-librte_gso += librte_mempool · line 112

Quoted 6 lines
lib/Makefile +2 -1
@@ -108,7 +108,8 @@ DEPDIRS-librte_gso += librte_mempool
 DIRS-$(CONFIG_RTE_LIBRTE_BPF) += librte_bpf
 DEPDIRS-librte_bpf := librte_eal librte_mempool librte_mbuf librte_ethdev
 DIRS-$(CONFIG_RTE_LIBRTE_IPSEC) += librte_ipsec
-DEPDIRS-librte_ipsec := librte_eal librte_mbuf librte_cryptodev librte_security
+DEPDIRS-librte_ipsec := librte_eal librte_mbuf librte_cryptodev librte_security \
+			librte_net

A nit.
Please update the comment in lib/meson.build file for the dependencies. Currently it is only for crypto and security.

Quoted 35 lines
lib/Makefile +0 -0
 DIRS-$(CONFIG_RTE_LIBRTE_TELEMETRY) += librte_telemetry
 DEPDIRS-librte_telemetry := librte_eal librte_metrics librte_ethdev
 DIRS-$(CONFIG_RTE_LIBRTE_RCU) += librte_rcu
lib/librte_ipsec/iph.h +7 -3
index 58930cf18..082e4e73e 100644
@@ -5,6 +5,8 @@
 #ifndef _IPH_H_
 #define _IPH_H_
 
+#include <rte_ip.h>
+
 /**
  * @file iph.h
  * Contains functions/structures/macros to manipulate IPv4/IPv6 headers
@@ -40,24 +42,61 @@ static inline int
 update_trs_l3hdr(const struct rte_ipsec_sa *sa, void *p, uint32_t plen,
 		uint32_t l2len, uint32_t l3len, uint8_t proto)
 {
-	struct ipv4_hdr *v4h;
-	struct ipv6_hdr *v6h;
 	int32_t rc;
 
+	/* IPv4 */
 	if ((sa->type & RTE_IPSEC_SATP_IPV_MASK) == RTE_IPSEC_SATP_IPV4) {
+		struct ipv4_hdr *v4h;
+
 		v4h = p;
 		rc = v4h->next_proto_id;
 		v4h->next_proto_id = proto;
 		v4h->total_length = rte_cpu_to_be_16(plen - l2len);
-	} else if (l3len == sizeof(*v6h)) {
+	/* IPv6 */
+	} else {

lib/librte_ipsec/iph.h static inline int · line 64

Quoted 10 lines
lib/librte_ipsec/iph.h +7 -2
+		struct ipv6_hdr *v6h;
+		uint8_t *next_proto_off;
+
 		v6h = p;
-		rc = v6h->proto;
-		v6h->proto = proto;
+
+		/* basic IPv6 header with no extensions */
+		if (l3len == sizeof(struct ipv6_hdr))
+			next_proto_off = &v6h->proto;

Is this next_proto_off a pointer to an offset or the value of the next_proto. So IMO the name should be next_proto or it should be p_nh

Quoted 43 lines
lib/librte_ipsec/iph.h +33 -4
+
+		/* IPv6 with extensions */
+		else {
+			size_t ext_len;
+			int nh;
+			uint8_t *pd, *plimit;
+
+			/* locate last extension within l3len bytes */
+			pd = (uint8_t *)p;
+			plimit = pd + l3len;
+			ext_len = sizeof(struct ipv6_hdr);
+			nh = v6h->proto;
+			while (pd + ext_len < plimit) {
+				pd += ext_len;
+				nh = rte_ipv6_get_next_ext(pd, nh, &ext_len);
+				if (unlikely(nh < 0))
+					return -EINVAL;
+			}
+
+			/* invalid l3len - extension exceeds header length */
+			if (unlikely(pd + ext_len != plimit))
+				return -EINVAL;
+
+			/* save last extension offset */
+			next_proto_off = pd;
+		}
+
+		/* update header type; return original value */
+		rc = *next_proto_off;
+		*next_proto_off = proto;
+
+		/* fix packet length */
 		v6h->payload_len = rte_cpu_to_be_16(plen - l2len -
 				sizeof(*v6h));
-	/* need to add support for IPv6 with options */
-	} else
-		rc = -ENOTSUP;
+	}
 
 	return rc;
 }
--

2.21.0.windows.1

#4 lore raw

PATCH 3/3 examples/ipsec-secgw: add support for ipv6 options

In reply to Marcin Smoczynski

Using transport with IPv6 and header extensions requires calculating total header length including extensions up to ESP header which is achieved with iteratively parsing extensions when preparing traffic for processing. Calculated l3_len is later used to determine SPI field offset for an inbound traffic and to reconstruct L3 header by librte_ipsec.

A simple unittest script is provided to test various headers for the IPv6 transport mode. Within each test case a test packet is crafted with Scapy and sent as an inbound or outbound traffic. Application response is then checked with a set of assertions.

---
 examples/ipsec-secgw/ipsec-secgw.c      |  33 +++-
 examples/ipsec-secgw/sa.c               |   5 +-
 examples/ipsec-secgw/test/test-scapy.py | 231 ++++++++++++++++++++++++
 3 files changed, 260 insertions(+), 9 deletions(-)
 create mode 100755 examples/ipsec-secgw/test/test-scapy.py
examples/ipsec-secgw/ipsec-secgw.c +28 -5
index 478dd80c2..1c49aa22c 100644
@@ -41,6 +41,7 @@
 #include <rte_jhash.h>
 #include <rte_cryptodev.h>
 #include <rte_security.h>
+#include <rte_ip.h>
 
 #include "ipsec.h"
 #include "parser.h"
@@ -248,16 +249,38 @@ prepare_one_packet(struct rte_mbuf *pkt, struct ipsec_traffic *t)
 		pkt->l2_len = 0;
 		pkt->l3_len = sizeof(struct ip);
 	} else if (eth->ether_type == rte_cpu_to_be_16(ETHER_TYPE_IPv6)) {
-		nlp = (uint8_t *)rte_pktmbuf_adj(pkt, ETHER_HDR_LEN);
-		nlp = RTE_PTR_ADD(nlp, offsetof(struct ip6_hdr, ip6_nxt));
-		if (*nlp == IPPROTO_ESP)
+		int next_proto;
+		size_t l3len, ext_len;
+		struct ipv6_hdr *v6h;
+		uint8_t *p;
+
+		/* get protocol type */
+		v6h = (struct ipv6_hdr *)rte_pktmbuf_adj(pkt, ETHER_HDR_LEN);
+		next_proto = v6h->proto;
+
+		/* determine l3 header size up to ESP extension */
+		l3len = sizeof(struct ip6_hdr);
+		p = rte_pktmbuf_mtod(pkt, uint8_t *);
+		while (next_proto != IPPROTO_ESP && l3len < pkt->data_len &&
+			(next_proto = rte_ipv6_get_next_ext(p + l3len,
+						next_proto, &ext_len)) >= 0)
+			l3len += ext_len;
+
+		/* drop packet when IPv6 header exceeds first segment length */
+		if (unlikely(l3len > pkt->data_len)) {
+			rte_pktmbuf_free(pkt);
+			return;
+		}
+
+		if (next_proto == IPPROTO_ESP)
 			t->ipsec.pkts[(t->ipsec.num)++] = pkt;
 		else {
-			t->ip6.data[t->ip6.num] = nlp;
+			t->ip6.data[t->ip6.num] = rte_pktmbuf_mtod_offset(pkt,
+				uint8_t *, offsetof(struct ipv6_hdr, proto));
 			t->ip6.pkts[(t->ip6.num)++] = pkt;
 		}
 		pkt->l2_len = 0;
-		pkt->l3_len = sizeof(struct ip6_hdr);
+		pkt->l3_len = l3len;
 	} else {
 		/* Unknown/Unsupported type, drop the packet */
 		RTE_LOG(ERR, IPSEC, "Unsupported packet type 0x%x\n",
examples/ipsec-secgw/sa.c +1 -4
index b850e9839..607527d08 100644
@@ -1228,10 +1228,7 @@ single_inbound_lookup(struct ipsec_sa *sadb, struct rte_mbuf *pkt,
 	*sa_ret = NULL;
 
 	ip = rte_pktmbuf_mtod(pkt, struct ip *);
-	if (ip->ip_v == IPVERSION)
-		esp = (struct esp_hdr *)(ip + 1);
-	else
-		esp = (struct esp_hdr *)(((struct ip6_hdr *)ip) + 1);
+	esp = rte_pktmbuf_mtod_offset(pkt, struct esp_hdr *, pkt->l3_len);
 
 	if (esp->spi == INVALID_SPI)
 		return;
examples/ipsec-secgw/test/test-scapy.py new file +231 -0

231 lines not on this page · raw message

Signature · 1 line
2.21.0.windows.1
#5 lore raw

Re: [dpdk-dev] [PATCH 3/3] examples/ipsec-secgw: add support for ipv6 options

Quoted 10 lines

Using transport with IPv6 and header extensions requires calculating total header length including extensions up to ESP header which is achieved with iteratively parsing extensions when preparing traffic for processing. Calculated l3_len is later used to determine SPI field offset for an inbound traffic and to reconstruct L3 header by librte_ipsec.

Quoted 4 lines

A simple unittest script is provided to test various headers for the IPv6 transport mode. Within each test case a test packet is crafted with Scapy and sent as an inbound or outbound traffic. Application response is then checked with a set of assertions.

ipsec-secgw changes itself looks good to me.
One comment I have - would be good to incorporate your test script into run_test.sh, so user can run all tests in one go. Konstantin

Quoted 323 lines
---
 examples/ipsec-secgw/ipsec-secgw.c      |  33 +++-
 examples/ipsec-secgw/sa.c               |   5 +-
 examples/ipsec-secgw/test/test-scapy.py | 231 ++++++++++++++++++++++++
 3 files changed, 260 insertions(+), 9 deletions(-)
 create mode 100755 examples/ipsec-secgw/test/test-scapy.py
examples/ipsec-secgw/ipsec-secgw.c +28 -5
index 478dd80c2..1c49aa22c 100644
@@ -41,6 +41,7 @@
 #include <rte_jhash.h>
 #include <rte_cryptodev.h>
 #include <rte_security.h>
+#include <rte_ip.h>
 
 #include "ipsec.h"
 #include "parser.h"
@@ -248,16 +249,38 @@ prepare_one_packet(struct rte_mbuf *pkt, struct ipsec_traffic *t)
 		pkt->l2_len = 0;
 		pkt->l3_len = sizeof(struct ip);
 	} else if (eth->ether_type == rte_cpu_to_be_16(ETHER_TYPE_IPv6)) {
-		nlp = (uint8_t *)rte_pktmbuf_adj(pkt, ETHER_HDR_LEN);
-		nlp = RTE_PTR_ADD(nlp, offsetof(struct ip6_hdr, ip6_nxt));
-		if (*nlp == IPPROTO_ESP)
+		int next_proto;
+		size_t l3len, ext_len;
+		struct ipv6_hdr *v6h;
+		uint8_t *p;
+
+		/* get protocol type */
+		v6h = (struct ipv6_hdr *)rte_pktmbuf_adj(pkt, ETHER_HDR_LEN);
+		next_proto = v6h->proto;
+
+		/* determine l3 header size up to ESP extension */
+		l3len = sizeof(struct ip6_hdr);
+		p = rte_pktmbuf_mtod(pkt, uint8_t *);
+		while (next_proto != IPPROTO_ESP && l3len < pkt->data_len &&
+			(next_proto = rte_ipv6_get_next_ext(p + l3len,
+						next_proto, &ext_len)) >= 0)
+			l3len += ext_len;
+
+		/* drop packet when IPv6 header exceeds first segment length */
+		if (unlikely(l3len > pkt->data_len)) {
+			rte_pktmbuf_free(pkt);
+			return;
+		}
+
+		if (next_proto == IPPROTO_ESP)
 			t->ipsec.pkts[(t->ipsec.num)++] = pkt;
 		else {
-			t->ip6.data[t->ip6.num] = nlp;
+			t->ip6.data[t->ip6.num] = rte_pktmbuf_mtod_offset(pkt,
+				uint8_t *, offsetof(struct ipv6_hdr, proto));
 			t->ip6.pkts[(t->ip6.num)++] = pkt;
 		}
 		pkt->l2_len = 0;
-		pkt->l3_len = sizeof(struct ip6_hdr);
+		pkt->l3_len = l3len;
 	} else {
 		/* Unknown/Unsupported type, drop the packet */
 		RTE_LOG(ERR, IPSEC, "Unsupported packet type 0x%x\n",
examples/ipsec-secgw/sa.c +1 -4
index b850e9839..607527d08 100644
@@ -1228,10 +1228,7 @@ single_inbound_lookup(struct ipsec_sa *sadb, struct rte_mbuf *pkt,
 	*sa_ret = NULL;
 
 	ip = rte_pktmbuf_mtod(pkt, struct ip *);
-	if (ip->ip_v == IPVERSION)
-		esp = (struct esp_hdr *)(ip + 1);
-	else
-		esp = (struct esp_hdr *)(((struct ip6_hdr *)ip) + 1);
+	esp = rte_pktmbuf_mtod_offset(pkt, struct esp_hdr *, pkt->l3_len);
 
 	if (esp->spi == INVALID_SPI)
 		return;
examples/ipsec-secgw/test/test-scapy.py new file +4 -0
new file mode 100755
index 000000000..d7f66b734
@@ -0,0 +1,231 @@
+#!/usr/bin/env python3
+
+# Run DPDK IPsec example with following arguments:
+# ./dpdk-ipsec-secgw --log-level=31 -l 0 --vdev=crypto_openssl --vdev=net_tap0 --vdev=net_tap1 -- -P -p 0x3 -u 0x1 --config

"(0,0,0),(1,0,0)" -f test-transport.cfg -l

examples/ipsec-secgw/test/test-scapy.py new file +94 -0
+# Two tap ports are expected: 0: unprotected (remote), 1: protected (local)
+
+# sample configuration:
+#	sp ipv6 out esp protect 5 pri 1 \
+#	src 1111:0000:0000:0000:0000:0000:0000:0000/64 \
+#	dst 2222:0000:0000:0000:0000:0000:0000:0000/64 \
+#	sport 0:65535 dport 0:65535
+#
+#	sp ipv6 in esp protect 6 pri 1 \
+#	src 2222:0000:0000:0000:0000:0000:0000:0000/64 \
+#	dst 1111:0000:0000:0000:0000:0000:0000:0000/64 \
+#	sport 0:65535 dport 0:65535
+#
+#	sa out 5 cipher_algo aes-128-cbc cipher_key 0:0:0:0:0:0:0:0:0:0:0:0:0:0:0:0 \
+#	auth_algo sha1-hmac auth_key 0:0:0:0:0:0:0:0:0:0:0:0:0:0:0:0:0:0:0:0 \
+#	mode transport
+#
+#	sa in 6 cipher_algo aes-128-cbc cipher_key 0:0:0:0:0:0:0:0:0:0:0:0:0:0:0:0 \
+#	auth_algo sha1-hmac auth_key 0:0:0:0:0:0:0:0:0:0:0:0:0:0:0:0:0:0:0:0 \
+#	mode transport
+#
+#	rt ipv6 dst 1111:0000:0000:0000:0000:0000:0000:0000/64 port 1
+#	rt ipv6 dst 2222:0000:0000:0000:0000:0000:0000:0000/64 port 0
+#
+# run tests with:
+# python3 -m unittest test-scapy
+
+
+import socket
+import sys
+import unittest
+from scapy.all import *
+
+
+SRC_ETHER = "52:54:00:00:00:01"
+DST_ETHER = "52:54:00:00:00:02"
+SRC_ADDR = "1111::1"
+DST_ADDR = "2222::1"
+LOCAL_IFACE = "dtap1"
+REMOTE_IFACE = "dtap0"
+
+
+class Interface(object):
+    ETH_P_ALL = 3
+    MAX_PACKET_SIZE = 1280
+    def __init__(self, ifname):
+        self.name = ifname
+        self.s = socket.socket(socket.AF_PACKET, socket.SOCK_RAW, socket.htons(ETH_P_ALL))
+        self.s.bind((self.name, 0, socket.PACKET_OTHERHOST))
+
+    def __del__(self):
+        self.s.close()
+
+    def send_packet(self, pkt):
+        self.send_bytes(bytes(pkt))
+
+    def send_bytes(self, bytedata):
+        self.s.send(bytedata)
+
+    def recv_packet(self):
+        return Ether(self.recv_bytes())
+
+    def recv_bytes(self):
+        return self.s.recv(Interface.MAX_PACKET_SIZE)
+
+
+class TestTransportMode(unittest.TestCase):
+    # There is a bug in the IPsec Scapy implementation
+    # which causes invalid packet reconstruction after
+    # successful decryption. This method is a workaround.
+    @staticmethod
+    def decrypt(pkt, sa):
+        esp = pkt[ESP]
+
+        # decrypt dummy packet with no extensions
+        d = sa.decrypt(IPv6()/esp)
+
+        # fix 'next header' in the preceding header of the original
+        # packet and remove ESP
+        pkt[ESP].underlayer.nh = d[IPv6].nh
+        pkt[ESP].underlayer.remove_payload()
+
+        # combine L3 header with decrypted payload
+        npkt = pkt/d[IPv6].payload
+
+        # fix length
+        npkt[IPv6].plen = d[IPv6].plen + len(pkt[IPv6].payload)
+
+        return npkt
+
+    def setUp(self):
+        self.ilocal = Interface(LOCAL_IFACE)
+        self.iremote = Interface(REMOTE_IFACE)
+        self.outb_sa = SecurityAssociation(ESP, spi=5, crypt_algo='AES-CBC', crypt_key='\x00'*16, auth_algo='HMAC-SHA1-96',

auth_key='\x00'*20)
+ self.inb_sa = SecurityAssociation(ESP, spi=6, crypt_algo='AES-CBC', crypt_key='\x00'*16, auth_algo='HMAC-SHA1-96',
auth_key='\x00'*20)

examples/ipsec-secgw/test/test-scapy.py new file +132 -1
+
+    def test_outb_ipv6_noopt(self):
+        pkt = Ether(src=SRC_ETHER, dst=DST_ETHER)
+        pkt /= IPv6(src=SRC_ADDR, dst=DST_ADDR)
+        pkt /= UDP(sport=123,dport=456)/Raw(load="abc")
+        self.ilocal.send_packet(pkt)
+
+        # check response
+        resp = self.iremote.recv_packet()
+        self.assertEqual(resp[IPv6].nh, socket.IPPROTO_ESP)
+        self.assertEqual(resp[ESP].spi, 5)
+
+        # decrypt response, check packet after decryption
+        d = TestTransportMode.decrypt(resp[IPv6], self.outb_sa)
+        self.assertEqual(d[IPv6].nh, socket.IPPROTO_UDP)
+        self.assertEqual(d[UDP].sport, 123)
+        self.assertEqual(d[UDP].dport, 456)
+        self.assertEqual(bytes(d[UDP].payload), b'abc')
+
+    def test_outb_ipv6_opt(self):
+        hoptions = []
+        hoptions.append(RouterAlert(value=2))
+        hoptions.append(Jumbo(jumboplen=5000))
+        hoptions.append(Pad1())
+
+        doptions = []
+        doptions.append(HAO(hoa="1234::4321"))
+
+        pkt = Ether(src=SRC_ETHER, dst=DST_ETHER)
+        pkt /= IPv6(src=SRC_ADDR, dst=DST_ADDR)
+        pkt /= IPv6ExtHdrHopByHop(options=hoptions)
+        pkt /= IPv6ExtHdrRouting(addresses=["3333::3","4444::4"])
+        pkt /= IPv6ExtHdrDestOpt(options=doptions)
+        pkt /= UDP(sport=123,dport=456)/Raw(load="abc")
+        self.ilocal.send_packet(pkt)
+
+        # check response
+        resp = self.iremote.recv_packet()
+        self.assertEqual(resp[IPv6].nh, socket.IPPROTO_HOPOPTS)
+
+        # check extensions
+        self.assertEqual(resp[IPv6ExtHdrHopByHop].nh, socket.IPPROTO_ROUTING)
+        self.assertEqual(resp[IPv6ExtHdrRouting].nh, socket.IPPROTO_DSTOPTS)
+        self.assertEqual(resp[IPv6ExtHdrDestOpt].nh, socket.IPPROTO_ESP)
+
+        # check ESP
+        self.assertEqual(resp[ESP].spi, 5)
+
+        # decrypt response, check packet after decryption
+        d = TestTransportMode.decrypt(resp[IPv6], self.outb_sa)
+        self.assertEqual(d[IPv6].nh, socket.IPPROTO_HOPOPTS)
+        self.assertEqual(d[IPv6ExtHdrHopByHop].nh, socket.IPPROTO_ROUTING)
+        self.assertEqual(d[IPv6ExtHdrRouting].nh, socket.IPPROTO_DSTOPTS)
+        self.assertEqual(d[IPv6ExtHdrDestOpt].nh, socket.IPPROTO_UDP)
+
+        # check UDP
+        self.assertEqual(d[UDP].sport, 123)
+        self.assertEqual(d[UDP].dport, 456)
+        self.assertEqual(bytes(d[UDP].payload), b'abc')
+
+    def test_inb_ipv6_noopt(self):
+        # encrypt and send raw UDP packet
+        pkt = IPv6(src=DST_ADDR, dst=SRC_ADDR)
+        pkt /= UDP(sport=123,dport=456)/Raw(load="abc")
+        e = self.inb_sa.encrypt(pkt)
+        e = Ether(src=DST_ETHER, dst=SRC_ETHER)/e
+        self.iremote.send_packet(e)
+
+        # check response
+        resp = self.ilocal.recv_packet()
+        self.assertEqual(resp[IPv6].nh, socket.IPPROTO_UDP)
+
+        # check UDP packet
+        self.assertEqual(resp[UDP].sport, 123)
+        self.assertEqual(resp[UDP].dport, 456)
+        self.assertEqual(bytes(resp[UDP].payload), b'abc')
+
+    def test_inb_ipv6_opt(self):
+        hoptions = []
+        hoptions.append(RouterAlert(value=2))
+        hoptions.append(Jumbo(jumboplen=5000))
+        hoptions.append(Pad1())
+
+        doptions = []
+        doptions.append(HAO(hoa="1234::4321"))
+
+        # prepare packet with options
+        pkt = IPv6(src=DST_ADDR, dst=SRC_ADDR)
+        pkt /= IPv6ExtHdrHopByHop(options=hoptions)
+        pkt /= IPv6ExtHdrRouting(addresses=["3333::3","4444::4"])
+        pkt /= IPv6ExtHdrDestOpt(options=doptions)
+        pkt /= UDP(sport=123,dport=456)/Raw(load="abc")
+
+        # encrypt and send packet
+        e = self.inb_sa.encrypt(pkt)
+        e = Ether(src=DST_ETHER, dst=SRC_ETHER)/e
+        self.iremote.send_packet(e)
+
+        # check response
+        resp = self.ilocal.recv_packet()
+        self.assertEqual(resp[IPv6].nh, socket.IPPROTO_HOPOPTS)
+        self.assertEqual(resp[IPv6ExtHdrHopByHop].nh, socket.IPPROTO_ROUTING)
+        self.assertEqual(resp[IPv6ExtHdrRouting].nh, socket.IPPROTO_DSTOPTS)
+        self.assertEqual(resp[IPv6ExtHdrDestOpt].nh, socket.IPPROTO_UDP)
+
+        # check UDP
+        self.assertEqual(resp[UDP].sport, 123)
+        self.assertEqual(resp[UDP].dport, 456)
+        self.assertEqual(bytes(resp[UDP].payload), b'abc')
+
+    def test_inb_ipv6_frag(self):
+        # prepare ESP payload
+        pkt = UDP(sport=123,dport=456)/Raw(load="abc")
+        e = self.inb_sa.encrypt(IPv6()/pkt)
+
+        # craft and send inbound packet
+        e = Ether(src=DST_ETHER, dst=SRC_ETHER)/IPv6(src=DST_ADDR, dst=SRC_ADDR)/IPv6ExtHdrFragment()/e[IPv6].payload
+        self.iremote.send_packet(e)
+
+        # check response
+        resp = self.ilocal.recv_packet()
+        self.assertEqual(resp[IPv6].nh, socket.IPPROTO_FRAGMENT)
+        self.assertEqual(resp[IPv6ExtHdrFragment].nh, socket.IPPROTO_UDP)
+
+        # check UDP
+        self.assertEqual(resp[UDP].sport, 123)
+        self.assertEqual(resp[UDP].dport, 456)
+        self.assertEqual(bytes(resp[UDP].payload), b'abc')
+
+
+if __name__ == "__main__":
+    unittest.main()
--

2.21.0.windows.1

#7 lore raw
In reply to Marcin Smoczynski

Hi Marcin,

Quoted 30 lines

Introduce new function for IPv6 header extension parsing able to determine extension length and next protocol number.

This function is helpful when implementing IPv6 header traversing.

---
 lib/librte_net/rte_ip.h | 49 +++++++++++++++++++++++++++++++++++++++++
 1 file changed, 49 insertions(+)
lib/librte_net/rte_ip.h +10 -0
index f9b909090..be64da662 100644
@@ -425,6 +425,55 @@ rte_ipv6_udptcp_cksum(const struct ipv6_hdr
 *ipv6_hdr, const void *l4_hdr)
 	return (uint16_t)cksum;
 }
 
+/* IPv6 fragmentation header size */
+#define RTE_IPV6_FRAG_HDR_SIZE 8
+
+/**
+ * Parse next IPv6 header extension
+ *
+ * This function checks if proto number is an IPv6 extensions and parses its
+ * data if so, providing information on next header and extension length.
+ *
+ * @param p

lib/librte_net/rte_ip.h rte_ipv6_udptcp_cksum(const struct ipv6_hdr · line 448

Quoted 10 lines
lib/librte_net/rte_ip.h +10 -0
+ *   Pointer to an extension raw data.
+ * @param proto
+ *   Protocol number extracted from the "next header" field from
+ *   the IPv6 header or the previous extension.
+ * @param ext_len
+ *   Extension data length.
+ * @return
+ *   next protocol number if proto is an IPv6 extension, -EINVAL otherwise
+ */
+static inline int __rte_experimental

Rte_experimental may not be required for inline functions.

Quoted 8 lines
lib/librte_net/rte_ip.h +8 -0
+rte_ipv6_get_next_ext(uint8_t *p, int proto, size_t *ext_len)
+{
+	int next_proto;
+
+	switch (proto) {
+	case IPPROTO_AH:
+		next_proto = *p++;
+		*ext_len = (*p + 2) * sizeof(uint32_t);

lib/librte_net/rte_ip.h rte_ipv6_udptcp_cksum(const struct ipv6_hdr *ipv6_hdr, const void *l4_hdr) · line 465

Quoted 10 lines
lib/librte_net/rte_ip.h +10 -0
+		break;
+
+	case IPPROTO_HOPOPTS:
+	case IPPROTO_ROUTING:
+	case IPPROTO_DSTOPTS:
+		next_proto = *p++;
+		*ext_len = (*p + 1) * sizeof(uint64_t);
+		break;
+
+	case IPPROTO_FRAGMENT:

I see that there is some compilation issues with respect to IPPROTO_xxx fields. These are reported in patchworks if you need details.

Could you please fix these and send next rev.

Thanks,
Akhil

#8 lore raw

Hi Akhil,

Quoted 38 lines

Hi Marcin,

Quoted 35 lines

Introduce new function for IPv6 header extension parsing able to determine extension length and next protocol number.

This function is helpful when implementing IPv6 header traversing.

---
 lib/librte_net/rte_ip.h | 49 +++++++++++++++++++++++++++++++++++++++++
 1 file changed, 49 insertions(+)
lib/librte_net/rte_ip.h +15 -0
index f9b909090..be64da662 100644
@@ -425,6 +425,55 @@ rte_ipv6_udptcp_cksum(const struct ipv6_hdr
 *ipv6_hdr, const void *l4_hdr)
 	return (uint16_t)cksum;
 }
 
+/* IPv6 fragmentation header size */
+#define RTE_IPV6_FRAG_HDR_SIZE 8
+
+/**
+ * Parse next IPv6 header extension
+ *
+ * This function checks if proto number is an IPv6 extensions and parses its
+ * data if so, providing information on next header and extension length.
+ *
+ * @param p
+ *   Pointer to an extension raw data.
+ * @param proto
+ *   Protocol number extracted from the "next header" field from
+ *   the IPv6 header or the previous extension.
+ * @param ext_len
Quoted 8 lines

lib/librte_net/rte_ip.h rte_ipv6_udptcp_cksum(const struct ipv6_hdr *ipv6_hdr, const void *l4_hdr) · line 447

Quoted 5 lines
lib/librte_net/rte_ip.h +5 -0
+ *   Extension data length.
+ * @return
+ *   next protocol number if proto is an IPv6 extension, -EINVAL otherwise
+ */
+static inline int __rte_experimental

Rte_experimental may not be required for inline functions.

AFAIK we do need that tag for both inline and non-inline functions, till API will be transferred to 'stable' state:

$ find lib -type f | xargs grep __rte_experimental | grep inline | wc -l
57
Quoted 26 lines
Quoted 10 lines
lib/librte_net/rte_ip.h +8 -0
+rte_ipv6_get_next_ext(uint8_t *p, int proto, size_t *ext_len)
+{
+	int next_proto;
+
+	switch (proto) {
+	case IPPROTO_AH:
+		next_proto = *p++;
+		*ext_len = (*p + 2) * sizeof(uint32_t);

lib/librte_net/rte_ip.h rte_ipv6_udptcp_cksum(const struct ipv6_hdr *ipv6_hdr, const void *l4_hdr) · line 465

Quoted 10 lines
lib/librte_net/rte_ip.h +10 -0
+		break;
+
+	case IPPROTO_HOPOPTS:
+	case IPPROTO_ROUTING:
+	case IPPROTO_DSTOPTS:
+		next_proto = *p++;
+		*ext_len = (*p + 1) * sizeof(uint64_t);
+		break;
+
+	case IPPROTO_FRAGMENT:

I see that there is some compilation issues with respect to IPPROTO_xxx fields. These are reported in patchworks if you need details.

Could you please fix these and send next rev.

Thanks,
Akhil

#9 lore raw

Hi Konstantin,

Quoted 51 lines

Hi Akhil,

Quoted 12 lines

Hi Marcin,

Quoted 9 lines

Introduce new function for IPv6 header extension parsing able to determine extension length and next protocol number.

This function is helpful when implementing IPv6 header traversing.

---
 lib/librte_net/rte_ip.h | 49

+++++++++++++++++++++++++++++++++++++++++

Quoted 26 lines
Quoted 26 lines
 1 file changed, 49 insertions(+)
lib/librte_net/rte_ip.h +15 -0
index f9b909090..be64da662 100644
@@ -425,6 +425,55 @@ rte_ipv6_udptcp_cksum(const struct ipv6_hdr
 *ipv6_hdr, const void *l4_hdr)
 	return (uint16_t)cksum;
 }
 
+/* IPv6 fragmentation header size */
+#define RTE_IPV6_FRAG_HDR_SIZE 8
+
+/**
+ * Parse next IPv6 header extension
+ *
+ * This function checks if proto number is an IPv6 extensions and parses its
+ * data if so, providing information on next header and extension length.
+ *
+ * @param p
+ *   Pointer to an extension raw data.
+ * @param proto
+ *   Protocol number extracted from the "next header" field from
+ *   the IPv6 header or the previous extension.
+ * @param ext_len
Quoted 8 lines

lib/librte_net/rte_ip.h rte_ipv6_udptcp_cksum(const struct ipv6_hdr *ipv6_hdr, const void *l4_hdr) · line 447

Quoted 5 lines
lib/librte_net/rte_ip.h +5 -0
+ *   Extension data length.
+ * @return
+ *   next protocol number if proto is an IPv6 extension, -EINVAL otherwise
+ */
+static inline int __rte_experimental

Rte_experimental may not be required for inline functions.

Quoted 5 lines

AFAIK we do need that tag for both inline and non-inline functions, till API will be transferred to 'stable' state:

$ find lib -type f | xargs grep __rte_experimental | grep inline | wc -l
57

My first impression was that this API was not exposed. Inline APIs are normally not exposed, So I thought rte_experimental is not required. I believe only exposed APIs Need to be tagged with rte_experimental.

Now since this API is exposed to the application as well, it should have an entry in the .map file. Otherwise shared build will fail I think for application.

Quoted 28 lines
Quoted 14 lines
Quoted 14 lines
lib/librte_net/rte_ip.h +11 -0
+rte_ipv6_get_next_ext(uint8_t *p, int proto, size_t *ext_len)
+{
+	int next_proto;
+
+	switch (proto) {
+	case IPPROTO_AH:
+		next_proto = *p++;
+		*ext_len = (*p + 2) * sizeof(uint32_t);
+		break;
+
+	case IPPROTO_HOPOPTS:
Quoted 8 lines

lib/librte_net/rte_ip.h rte_ipv6_udptcp_cksum(const struct ipv6_hdr *ipv6_hdr, const void *l4_hdr) · line 465

Quoted 7 lines
lib/librte_net/rte_ip.h +7 -0
+	case IPPROTO_ROUTING:
+	case IPPROTO_DSTOPTS:
+		next_proto = *p++;
+		*ext_len = (*p + 1) * sizeof(uint64_t);
+		break;
+
+	case IPPROTO_FRAGMENT:

I see that there is some compilation issues with respect to IPPROTO_xxx fields.

These are reported in patchworks if you need details.

Quoted 5 lines

Could you please fix these and send next rev.

Thanks,
Akhil

#10 lore raw

Hi Akhil,
I've double checked and there is no problem with linking application against shared libraries, because inline functions are being inlined during compilation and before linking is done. To be more specific: there is not such symbol as rte_ipv6_get_next_ext in ipsec-secgw.o (which uses this function in prepare_one_packet function).

There are a number of inline experimental functions defined today and they are mostly (43 out of 59) not listed in maps, e.g. rte_ticketlock functions family defined here:
lib/librte_eal/common/include/generic/rte_ticketlock.h

Is there any particular reason we should put inlines in maps? Maps are used by linker to control versions of shared libraries and inlines are not processed by linker.

I am aware this patchset causes build failures on BSDs - this is a project-wide issue which I have addressed here:
http://patchwork.dpdk.org/patch/53406/
I'll put a note in the next version of this patchset's cover that it depends on abovementioned patchset.

Marcin

Quoted 100 lines

Hi Konstantin,

Quoted 27 lines

Hi Akhil,

Quoted 12 lines

Hi Marcin,

Quoted 9 lines

Introduce new function for IPv6 header extension parsing able to determine extension length and next protocol number.

This function is helpful when implementing IPv6 header traversing.

---
 lib/librte_net/rte_ip.h | 49

+++++++++++++++++++++++++++++++++++++++++

Quoted 11 lines
Quoted 11 lines
 1 file changed, 49 insertions(+)
lib/librte_net/rte_ip.h +0 -0
index f9b909090..be64da662 100644
@@ -425,6 +425,55 @@ rte_ipv6_udptcp_cksum(const struct ipv6_hdr
 *ipv6_hdr, const void *l4_hdr)
 	return (uint16_t)cksum;
 }
 
Quoted 9 lines
Quoted 9 lines

lib/librte_net/rte_ip.h rte_ipv6_udptcp_cksum(const struct ipv6_hdr · line 437

Quoted 9 lines
lib/librte_net/rte_ip.h +9 -0
+/* IPv6 fragmentation header size */ #define
+RTE_IPV6_FRAG_HDR_SIZE 8
+
+/**
+ * Parse next IPv6 header extension
+ *
+ * This function checks if proto number is an IPv6 extensions and
+parses its
+ * data if so, providing information on next header and extension

length.

Quoted 10 lines
Quoted 10 lines

lib/librte_net/rte_ip.h rte_ipv6_udptcp_cksum(const struct ipv6_hdr *ipv6_hdr, const void *l4_hdr)

Quoted 10 lines
lib/librte_net/rte_ip.h +10 -0
+ *
+ * @param p
+ *   Pointer to an extension raw data.
+ * @param proto
+ *   Protocol number extracted from the "next header" field from
+ *   the IPv6 header or the previous extension.
+ * @param ext_len
+ *   Extension data length.
+ * @return
+ *   next protocol number if proto is an IPv6 extension, -EINVAL

otherwise

Quoted 4 lines
Quoted 4 lines
Quoted 2 lines
lib/librte_net/rte_ip.h +2 -0
+ */
+static inline int __rte_experimental
Quoted 8 lines
Quoted 1 line

Rte_experimental may not be required for inline functions.

AFAIK we do need that tag for both inline and non-inline functions, till API will be transferred to 'stable' state:

$ find lib -type f | xargs grep __rte_experimental | grep inline | wc
-l
57

My first impression was that this API was not exposed. Inline APIs are normally not exposed, So I thought rte_experimental is not required. I believe only exposed APIs Need to be tagged with rte_experimental.

Now since this API is exposed to the application as well, it should have an entry in the .map file.
Otherwise shared build will fail I think for application.

Quoted 13 lines
Quoted 13 lines
Quoted 13 lines
lib/librte_net/rte_ip.h +10 -0
+rte_ipv6_get_next_ext(uint8_t *p, int proto, size_t *ext_len) {
+	int next_proto;
+
+	switch (proto) {
+	case IPPROTO_AH:
+		next_proto = *p++;
+		*ext_len = (*p + 2) * sizeof(uint32_t);
+		break;
+
+	case IPPROTO_HOPOPTS:
Quoted 8 lines
Quoted 8 lines

lib/librte_net/rte_ip.h rte_ipv6_udptcp_cksum(const struct ipv6_hdr *ipv6_hdr, const void *l4_hdr) · line 465

Quoted 7 lines
lib/librte_net/rte_ip.h +7 -0
+	case IPPROTO_ROUTING:
+	case IPPROTO_DSTOPTS:
+		next_proto = *p++;
+		*ext_len = (*p + 1) * sizeof(uint64_t);
+		break;
+
+	case IPPROTO_FRAGMENT:

I see that there is some compilation issues with respect to IPPROTO_xxx

fields.

Quoted 6 lines

These are reported in patchworks if you need details.

Quoted 5 lines

Could you please fix these and send next rev.

Thanks,
Akhil

#11 lore raw

Hi Marcin,

Quoted 17 lines

Hi Akhil,
I've double checked and there is no problem with linking application against shared libraries, because inline functions are being inlined during compilation and before linking is done. To be more specific: there is not such symbol as rte_ipv6_get_next_ext in ipsec-secgw.o (which uses this function in prepare_one_packet function).

There are a number of inline experimental functions defined today and they are mostly (43 out of 59) not listed in maps, e.g. rte_ticketlock functions family defined here:
lib/librte_eal/common/include/generic/rte_ticketlock.h

Is there any particular reason we should put inlines in maps? Maps are used by linker to control versions of shared libraries and inlines are not processed by linker.

Quoted 7 lines

I am aware this patchset causes build failures on BSDs - this is a project-wide issue which I have addressed here:
http://patchwork.dpdk.org/patch/53406/
I'll put a note in the next version of this patchset's cover that it depends on abovementioned patchset.

Marcin

Thanks for correcting me.

#12 lore raw

v2 cover IPv6 with options support for IPsec transport

In reply to Marcin Smoczynski

Add support for IPv6 with header extensions (IPv6 options) and IPsec transport mode:
1. Add IPv6 header extension parsing method in librte_net
2. Fix IPv6 header with extension parsing in both librte_ipsec and ipsec sample application using abovementioned function
3. Add new testing mechanism for ipsec-secgw sample application which allows to automatically check how custom packets are processed.

This patchset depends on the following patch:
http://patchwork.dpdk.org/patch/53406/

Marcin Smoczynski (4):
net: new ipv6 header extension parsing function
ipsec: fix transport mode for ipv6 with extensions examples/ipsec-secgw: add support for ipv6 options examples/ipsec-secgw: add scapy based unittests

 examples/ipsec-secgw/ipsec-secgw.c            |  35 +++-
 examples/ipsec-secgw/sa.c                     |   5 +-
 examples/ipsec-secgw/test/common_defs.sh      |  58 +-----
 .../ipsec-secgw/test/common_defs_secgw.sh     |  65 +++++++
 examples/ipsec-secgw/test/pkttest.py          | 127 ++++++++++++
 examples/ipsec-secgw/test/pkttest.sh          |  65 +++++++
 examples/ipsec-secgw/test/run_test.sh         | 108 ++++++++---
 examples/ipsec-secgw/test/trs_ipv6opts.py     | 181 ++++++++++++++++++
 lib/Makefile                                  |   3 +-
 lib/librte_ipsec/iph.h                        |  55 +++++-
 lib/librte_net/rte_ip.h                       |  49 +++++
 lib/meson.build                               |   2 +-
 12 files changed, 649 insertions(+), 104 deletions(-)
 create mode 100644 examples/ipsec-secgw/test/common_defs_secgw.sh
 create mode 100755 examples/ipsec-secgw/test/pkttest.py
 create mode 100755 examples/ipsec-secgw/test/pkttest.sh
 mode change 100644 => 100755 examples/ipsec-secgw/test/run_test.sh
 create mode 100755 examples/ipsec-secgw/test/trs_ipv6opts.py
Signature · 1 line
2.17.1
#13 lore raw

PATCH v2 1/4 net: new ipv6 header extension parsing function

Introduce new function for IPv6 header extension parsing able to determine extension length and next protocol number.

This function is helpful when implementing IPv6 header traversing.

---
 lib/librte_net/rte_ip.h | 49 +++++++++++++++++++++++++++++++++++++++++
 1 file changed, 49 insertions(+)
lib/librte_net/rte_ip.h +49 -0
index ae3b7e730..c2c67b85d 100644
@@ -428,6 +428,55 @@ rte_ipv6_udptcp_cksum(const struct rte_ipv6_hdr *ipv6_hdr, const void *l4_hdr)
 	return (uint16_t)cksum;
 }
 
+/* IPv6 fragmentation header size */
+#define RTE_IPV6_FRAG_HDR_SIZE 8
+
+/**
+ * Parse next IPv6 header extension
+ *
+ * This function checks if proto number is an IPv6 extensions and parses its
+ * data if so, providing information on next header and extension length.
+ *
+ * @param p
+ *   Pointer to an extension raw data.
+ * @param proto
+ *   Protocol number extracted from the "next header" field from
+ *   the IPv6 header or the previous extension.
+ * @param ext_len
+ *   Extension data length.
+ * @return
+ *   next protocol number if proto is an IPv6 extension, -EINVAL otherwise
+ */
+static inline int __rte_experimental
+rte_ipv6_get_next_ext(uint8_t *p, int proto, size_t *ext_len)
+{
+	int next_proto;
+
+	switch (proto) {
+	case IPPROTO_AH:
+		next_proto = *p++;
+		*ext_len = (*p + 2) * sizeof(uint32_t);
+		break;
+
+	case IPPROTO_HOPOPTS:
+	case IPPROTO_ROUTING:
+	case IPPROTO_DSTOPTS:
+		next_proto = *p++;
+		*ext_len = (*p + 1) * sizeof(uint64_t);
+		break;
+
+	case IPPROTO_FRAGMENT:
+		next_proto = *p;
+		*ext_len = RTE_IPV6_FRAG_HDR_SIZE;
+		break;
+
+	default:
+		return -EINVAL;
+	}
+
+	return next_proto;
+}
+
 #ifdef __cplusplus
 }
 #endif
Signature · 1 line
2.17.1
Acked-by Tested-by on v2 1/4
#14 lore raw
Quoted 73 lines

-----Original Message-----
From: Smoczynski, MarcinX
Sent: Monday, June 24, 2019 2:40 PM
To: Kovacevic, Marko <[email protected]>; [email protected]; Richardson, Bruce <[email protected]>; De Lara Guarch, Pablo <[email protected]>; Nicolau, Radu <[email protected]>; [email protected]; Kantecki, Tomasz <[email protected]>; Ananyev, Konstantin <[email protected]>; Iremonger, Bernard <[email protected]>; [email protected]
Cc: [email protected]; Smoczynski, MarcinX <[email protected]>
Subject: [PATCH v2 1/4] net: new ipv6 header extension parsing function

Introduce new function for IPv6 header extension parsing able to determine extension length and next protocol number.

This function is helpful when implementing IPv6 header traversing.

---
 lib/librte_net/rte_ip.h | 49 +++++++++++++++++++++++++++++++++++++++++
 1 file changed, 49 insertions(+)
lib/librte_net/rte_ip.h +43 -0
index ae3b7e730..c2c67b85d 100644
@@ -428,6 +428,55 @@ rte_ipv6_udptcp_cksum(const struct rte_ipv6_hdr *ipv6_hdr, const void *l4_hdr)
 	return (uint16_t)cksum;
 }
 
+/* IPv6 fragmentation header size */
+#define RTE_IPV6_FRAG_HDR_SIZE 8
+
+/**
+ * Parse next IPv6 header extension
+ *
+ * This function checks if proto number is an IPv6 extensions and parses its
+ * data if so, providing information on next header and extension length.
+ *
+ * @param p
+ *   Pointer to an extension raw data.
+ * @param proto
+ *   Protocol number extracted from the "next header" field from
+ *   the IPv6 header or the previous extension.
+ * @param ext_len
+ *   Extension data length.
+ * @return
+ *   next protocol number if proto is an IPv6 extension, -EINVAL otherwise
+ */
+static inline int __rte_experimental
+rte_ipv6_get_next_ext(uint8_t *p, int proto, size_t *ext_len)
+{
+	int next_proto;
+
+	switch (proto) {
+	case IPPROTO_AH:
+		next_proto = *p++;
+		*ext_len = (*p + 2) * sizeof(uint32_t);
+		break;
+
+	case IPPROTO_HOPOPTS:
+	case IPPROTO_ROUTING:
+	case IPPROTO_DSTOPTS:
+		next_proto = *p++;
+		*ext_len = (*p + 1) * sizeof(uint64_t);
+		break;
+
+	case IPPROTO_FRAGMENT:
+		next_proto = *p;
+		*ext_len = RTE_IPV6_FRAG_HDR_SIZE;
+		break;
+
+	default:

lib/librte_net/rte_ip.h rte_ipv6_udptcp_cksum(const struct rte_ipv6_hdr *ipv6_hdr, const void *l4_hdr)

Quoted 10 lines
lib/librte_net/rte_ip.h +6 -1
+		return -EINVAL;
+	}
+
+	return next_proto;
+}
+
 #ifdef __cplusplus
 }
 #endif
--
Quoted 1 line

2.17.1

#15 lore raw
In reply to Marcin Smoczynski

Hi,

Quoted 60 lines · Marcin Smoczynski

Introduce new function for IPv6 header extension parsing able to determine extension length and next protocol number.

This function is helpful when implementing IPv6 header traversing.

---
 lib/librte_net/rte_ip.h | 49 +++++++++++++++++++++++++++++++++++++++++
 1 file changed, 49 insertions(+)
lib/librte_net/rte_ip.h +42 -0
index ae3b7e730..c2c67b85d 100644
@@ -428,6 +428,55 @@ rte_ipv6_udptcp_cksum(const struct rte_ipv6_hdr *ipv6_hdr, const void *l4_hdr)
 	return (uint16_t)cksum;
 }
 
+/* IPv6 fragmentation header size */
+#define RTE_IPV6_FRAG_HDR_SIZE 8
+
+/**
+ * Parse next IPv6 header extension
+ *
+ * This function checks if proto number is an IPv6 extensions and parses its
+ * data if so, providing information on next header and extension length.
+ *
+ * @param p
+ *   Pointer to an extension raw data.
+ * @param proto
+ *   Protocol number extracted from the "next header" field from
+ *   the IPv6 header or the previous extension.
+ * @param ext_len
+ *   Extension data length.
+ * @return
+ *   next protocol number if proto is an IPv6 extension, -EINVAL otherwise
+ */
+static inline int __rte_experimental
+rte_ipv6_get_next_ext(uint8_t *p, int proto, size_t *ext_len)
+{
+	int next_proto;
+
+	switch (proto) {
+	case IPPROTO_AH:
+		next_proto = *p++;
+		*ext_len = (*p + 2) * sizeof(uint32_t);
+		break;
+
+	case IPPROTO_HOPOPTS:
+	case IPPROTO_ROUTING:
+	case IPPROTO_DSTOPTS:
+		next_proto = *p++;
+		*ext_len = (*p + 1) * sizeof(uint64_t);
+		break;
+
+	case IPPROTO_FRAGMENT:
+		next_proto = *p;
+		*ext_len = RTE_IPV6_FRAG_HDR_SIZE;
+		break;
+

lib/librte_net/rte_ip.h rte_ipv6_udptcp_cksum(const struct rte_ipv6_hdr *ipv6_hdr, const void *l4_hdr) · line 482

Quoted 10 lines
lib/librte_net/rte_ip.h +7 -0
+	default:
+		return -EINVAL;
+	}
+
+	return next_proto;
+}
+
 #ifdef __cplusplus
 }
 #endif

I just want to highlight the potential danger of this kind of function: then length is read from the packet (controlled by the peer), and returned without check.

My initial fear was that an attacker forge an IPv6 packet, resulting in *ext_len being 0. For instance, calling the function with (proto = IPPROTO_ROUTING, buffer = [ 0, 255, ... ]). Fortunatly, this does not happen, due to the implicit promotion of *p to a larger integer in the (*p + 1) expression.

That said, what about adding some comments in the description, saying that:
- the pointer p must point to a buffer whose length is at least 2
- the returned length must be checked by the caller to ensure it does not overflow the packet buffer

You can add my ack in the next version.

Thanks,
Olivier

#16 lore raw

PATCH v2 2/4 ipsec: fix transport mode for ipv6 with extensions

In reply to Marcin Smoczynski

Reconstructing IPv6 header after encryption or decryption requires updating 'next header' value in the preceding protocol header, which is determined by parsing IPv6 header and iteratively looking for next IPv6 header extension.

It is required that 'l3_len' in the mbuf metadata contains a total length of the IPv6 header with header extensions up to ESP header.

---
 lib/Makefile           |  3 ++-
 lib/librte_ipsec/iph.h | 55 ++++++++++++++++++++++++++++++++++++------
 lib/meson.build        |  2 +-
 3 files changed, 50 insertions(+), 10 deletions(-)
lib/Makefile +2 -1
index 791e0d991..3ad579f68 100644
@@ -108,7 +108,8 @@ DEPDIRS-librte_gso += librte_mempool
 DIRS-$(CONFIG_RTE_LIBRTE_BPF) += librte_bpf
 DEPDIRS-librte_bpf := librte_eal librte_mempool librte_mbuf librte_ethdev
 DIRS-$(CONFIG_RTE_LIBRTE_IPSEC) += librte_ipsec
-DEPDIRS-librte_ipsec := librte_eal librte_mbuf librte_cryptodev librte_security
+DEPDIRS-librte_ipsec := librte_eal librte_mbuf librte_cryptodev librte_security \
+			librte_net
 DIRS-$(CONFIG_RTE_LIBRTE_TELEMETRY) += librte_telemetry
 DEPDIRS-librte_telemetry := librte_eal librte_metrics librte_ethdev
 DIRS-$(CONFIG_RTE_LIBRTE_RCU) += librte_rcu
lib/librte_ipsec/iph.h +47 -8

77 lines not on this page · raw message

lib/meson.build +1 -1
index 992091a94..e93797ede 100644
@@ -23,7 +23,7 @@ libraries = [
 	'kni', 'latencystats', 'lpm', 'member',
 	'power', 'pdump', 'rawdev',
 	'rcu', 'reorder', 'sched', 'security', 'stack', 'vhost',
-	#ipsec lib depends on crypto and security
+	# ipsec lib depends on net, crypto and security
 	'ipsec',
 	# add pkt framework libs which use other libs from above
 	'port', 'table', 'pipeline',
Signature · 1 line
2.17.1
Acked-by Tested-by on v2 2/4
#17 lore raw

Re: [dpdk-dev] [PATCH v2 2/4] ipsec: fix transport mode for ipv6 with extensions

Quoted 7 lines

Reconstructing IPv6 header after encryption or decryption requires updating 'next header' value in the preceding protocol header, which is determined by parsing IPv6 header and iteratively looking for next IPv6 header extension.

Quoted 5 lines

It is required that 'l3_len' in the mbuf metadata contains a total length of the IPv6 header with header extensions up to ESP header.

---

Quoted 1 line

2.17.1

#18 lore raw

PATCH v2 3/4 examples/ipsec-secgw: add support for ipv6 options

In reply to Marcin Smoczynski

Using transport with IPv6 and header extensions requires calculating total header length including extensions up to ESP header which is achieved with iteratively parsing extensions when preparing traffic for processing. Calculated l3_len is later used to determine SPI field offset for an inbound traffic and to reconstruct L3 header by librte_ipsec.

---
 examples/ipsec-secgw/ipsec-secgw.c | 35 +++++++++++++++++++++++++-----
 examples/ipsec-secgw/sa.c          |  5 +----
 2 files changed, 31 insertions(+), 9 deletions(-)
examples/ipsec-secgw/ipsec-secgw.c +30 -5
index 6c626fa5f..17012caf9 100644
@@ -41,6 +41,7 @@
 #include <rte_jhash.h>
 #include <rte_cryptodev.h>
 #include <rte_security.h>
+#include <rte_ip.h>
 
 #include "ipsec.h"
 #include "parser.h"
@@ -248,16 +249,40 @@ prepare_one_packet(struct rte_mbuf *pkt, struct ipsec_traffic *t)
 		pkt->l2_len = 0;
 		pkt->l3_len = sizeof(struct ip);
 	} else if (eth->ether_type == rte_cpu_to_be_16(RTE_ETHER_TYPE_IPV6)) {
-		nlp = (uint8_t *)rte_pktmbuf_adj(pkt, RTE_ETHER_HDR_LEN);
-		nlp = RTE_PTR_ADD(nlp, offsetof(struct ip6_hdr, ip6_nxt));
-		if (*nlp == IPPROTO_ESP)
+		int next_proto;
+		size_t l3len, ext_len;
+		struct rte_ipv6_hdr *v6h;
+		uint8_t *p;
+
+		/* get protocol type */
+		v6h = (struct rte_ipv6_hdr *)rte_pktmbuf_adj(pkt,
+			RTE_ETHER_HDR_LEN);
+		next_proto = v6h->proto;
+
+		/* determine l3 header size up to ESP extension */
+		l3len = sizeof(struct ip6_hdr);
+		p = rte_pktmbuf_mtod(pkt, uint8_t *);
+		while (next_proto != IPPROTO_ESP && l3len < pkt->data_len &&
+			(next_proto = rte_ipv6_get_next_ext(p + l3len,
+						next_proto, &ext_len)) >= 0)
+			l3len += ext_len;
+
+		/* drop packet when IPv6 header exceeds first segment length */
+		if (unlikely(l3len > pkt->data_len)) {
+			rte_pktmbuf_free(pkt);
+			return;
+		}
+
+		if (next_proto == IPPROTO_ESP)
 			t->ipsec.pkts[(t->ipsec.num)++] = pkt;
 		else {
-			t->ip6.data[t->ip6.num] = nlp;
+			t->ip6.data[t->ip6.num] = rte_pktmbuf_mtod_offset(pkt,
+				uint8_t *,
+				offsetof(struct rte_ipv6_hdr, proto));
 			t->ip6.pkts[(t->ip6.num)++] = pkt;
 		}
 		pkt->l2_len = 0;
-		pkt->l3_len = sizeof(struct ip6_hdr);
+		pkt->l3_len = l3len;
 	} else {
 		/* Unknown/Unsupported type, drop the packet */
 		RTE_LOG(ERR, IPSEC, "Unsupported packet type 0x%x\n",
examples/ipsec-secgw/sa.c +1 -4
index 8d47d1def..7262ccee8 100644
@@ -1228,10 +1228,7 @@ single_inbound_lookup(struct ipsec_sa *sadb, struct rte_mbuf *pkt,
 	*sa_ret = NULL;
 
 	ip = rte_pktmbuf_mtod(pkt, struct ip *);
-	if (ip->ip_v == IPVERSION)
-		esp = (struct rte_esp_hdr *)(ip + 1);
-	else
-		esp = (struct rte_esp_hdr *)(((struct ip6_hdr *)ip) + 1);
+	esp = rte_pktmbuf_mtod_offset(pkt, struct rte_esp_hdr *, pkt->l3_len);
 
 	if (esp->spi == INVALID_SPI)
 		return;
Signature · 1 line
2.17.1
Acked-by Tested-by on v2 3/4
#19 lore raw

Re: [dpdk-dev] [PATCH v2 3/4] examples/ipsec-secgw: add support for ipv6 options

Quoted 11 lines

Using transport with IPv6 and header extensions requires calculating total header length including extensions up to ESP header which is achieved with iteratively parsing extensions when preparing traffic for processing. Calculated l3_len is later used to determine SPI field offset for an inbound traffic and to reconstruct L3 header by librte_ipsec.

---

Quoted 1 line

2.17.1

#20 lore raw

PATCH v2 4/4 examples/ipsec-secgw: add scapy based unittests

In reply to Marcin Smoczynski

Add new unittest-like mechanism which uses scapy to craft custom packets and a set of assertions to check how ipsec-secgw example application is processing them. Python3 with scapy module is required by pkttest.sh to run test scripts.

A new mechanism is used to test IPv6 transport mode traffic with header extensions (trs_ipv6opts.py).

Fix incomplete test log problem by disabling buffering of ipsec-secgw standard output with stdbuf application.

---
 examples/ipsec-secgw/test/common_defs.sh      |  58 +-----
 .../ipsec-secgw/test/common_defs_secgw.sh     |  65 +++++++
 examples/ipsec-secgw/test/pkttest.py          | 127 ++++++++++++
 examples/ipsec-secgw/test/pkttest.sh          |  65 +++++++
 examples/ipsec-secgw/test/run_test.sh         | 108 ++++++++---
 examples/ipsec-secgw/test/trs_ipv6opts.py     | 181 ++++++++++++++++++
 6 files changed, 519 insertions(+), 85 deletions(-)
 create mode 100644 examples/ipsec-secgw/test/common_defs_secgw.sh
 create mode 100755 examples/ipsec-secgw/test/pkttest.py
 create mode 100755 examples/ipsec-secgw/test/pkttest.sh
 mode change 100644 => 100755 examples/ipsec-secgw/test/run_test.sh
 create mode 100755 examples/ipsec-secgw/test/trs_ipv6opts.py
examples/ipsec-secgw/test/common_defs.sh +4 -54
index 8dc574b50..63ad5415d 100644
@@ -1,22 +1,10 @@
 #! /bin/bash
 
-#check that env vars are properly defined
-
-#check SGW_PATH
-if [[ -z "${SGW_PATH}" || ! -x ${SGW_PATH} ]]; then
-	echo "SGW_PATH is invalid"
-	exit 127
-fi
-
 #check ETH_DEV
 if [[ -z "${ETH_DEV}" ]]; then
 	echo "ETH_DEV is invalid"
 	exit 127
 fi
-
-#setup SGW_LCORE
-SGW_LCORE=${SGW_LCORE:-0}
-
 #check that REMOTE_HOST is reachable
 ssh ${REMOTE_HOST} echo
 st=$?
@@ -47,14 +35,6 @@ LOCAL_IPV6=fd12:3456:789a:0031:0000:0000:0000:0092
 DPDK_PATH=${RTE_SDK:-${PWD}}
 DPDK_BUILD=${RTE_TARGET:-x86_64-native-linux-gcc}
 
-SGW_OUT_FILE=./ipsec-secgw.out1
-
-SGW_CMD_EAL_PRM="--lcores=${SGW_LCORE} -n 4 ${ETH_DEV}"
-SGW_CMD_CFG="(0,0,${SGW_LCORE}),(1,0,${SGW_LCORE})"
-SGW_CMD_PRM="-p 0x3 -u 1 -P --config=\"${SGW_CMD_CFG}\""
-
-SGW_CFG_FILE=$(mktemp)
-
 # configure local host/ifaces
 config_local_iface()
 {
@@ -126,37 +106,7 @@ config6_iface()
 	config6_remote_iface
 }
 
-#start ipsec-secgw
-secgw_start()
-{
-	SGW_EXEC_FILE=$(mktemp)
-	cat <<EOF > ${SGW_EXEC_FILE}
-${SGW_PATH} ${SGW_CMD_EAL_PRM} ${CRYPTO_DEV} \
---vdev="net_tap0,mac=fixed" \
--- ${SGW_CMD_PRM} ${SGW_CMD_XPRM} -f ${SGW_CFG_FILE} > \
-${SGW_OUT_FILE} 2>&1 &
-p=\$!
-echo \$p
-EOF
-
-	cat ${SGW_EXEC_FILE}
-	SGW_PID=`/bin/bash -x ${SGW_EXEC_FILE}`
-
-	# wait till ipsec-secgw start properly
-	i=0
-	st=1
-	while [[ $i -ne 10 && st -ne 0 ]]; do
-		sleep 1
-		ifconfig ${LOCAL_IFACE}
-		st=$?
-		let i++
-	done
-}
-
-#stop ipsec-secgw and cleanup
-secgw_stop()
-{
-	kill ${SGW_PID}
-	rm -f ${SGW_EXEC_FILE}
-	rm -f ${SGW_CFG_FILE}
-}
+# secgw application parameters setup
+SGW_PORT_CFG="--vdev=\"net_tap0,mac=fixed\" ${ETH_DEV}"
+SGW_WAIT_DEV="${LOCAL_IFACE}"
+. ${DIR}/common_defs_secgw.sh
examples/ipsec-secgw/test/common_defs_secgw.sh new file +65 -0

65 lines not on this page · raw message

examples/ipsec-secgw/test/pkttest.py new file +127 -0

127 lines not on this page · raw message

examples/ipsec-secgw/test/pkttest.sh new file +65 -0

65 lines not on this page · raw message

examples/ipsec-secgw/test/run_test.sh +77 -31

130 lines not on this page · raw message

examples/ipsec-secgw/test/trs_ipv6opts.py new file +181 -0

181 lines not on this page · raw message

Signature · 1 line
2.17.1
Acked-by Tested-by on v2 4/4
#21 lore raw

Re: [dpdk-dev] [PATCH v2 4/4] examples/ipsec-secgw: add scapy based unittests

Quoted 7 lines

Add new unittest-like mechanism which uses scapy to craft custom packets and a set of assertions to check how ipsec-secgw example application is processing them. Python3 with scapy module is required by pkttest.sh to run test scripts.

Quoted 8 lines

A new mechanism is used to test IPv6 transport mode traffic with header extensions (trs_ipv6opts.py).

Fix incomplete test log problem by disabling buffering of ipsec-secgw standard output with stdbuf application.

---

Quoted 1 line

2.17.1

#22 lore raw

Re: [dpdk-dev] [PATCH v2 0/4] IPv6 with options support for IPsec transport

In reply to Marcin Smoczynski
Quoted 39 lines

Add support for IPv6 with header extensions (IPv6 options) and IPsec transport mode:
1. Add IPv6 header extension parsing method in librte_net
2. Fix IPv6 header with extension parsing in both librte_ipsec and ipsec sample application using abovementioned function
3. Add new testing mechanism for ipsec-secgw sample application which allows to automatically check how custom packets are processed.

This patchset depends on the following patch:
http://patchwork.dpdk.org/patch/53406/

Marcin Smoczynski (4):
net: new ipv6 header extension parsing function
ipsec: fix transport mode for ipv6 with extensions examples/ipsec-secgw: add support for ipv6 options examples/ipsec-secgw: add scapy based unittests

 examples/ipsec-secgw/ipsec-secgw.c            |  35 +++-
 examples/ipsec-secgw/sa.c                     |   5 +-
 examples/ipsec-secgw/test/common_defs.sh      |  58 +-----
 .../ipsec-secgw/test/common_defs_secgw.sh     |  65 +++++++
 examples/ipsec-secgw/test/pkttest.py          | 127 ++++++++++++
 examples/ipsec-secgw/test/pkttest.sh          |  65 +++++++
 examples/ipsec-secgw/test/run_test.sh         | 108 ++++++++---
 examples/ipsec-secgw/test/trs_ipv6opts.py     | 181 ++++++++++++++++++
 lib/Makefile                                  |   3 +-
 lib/librte_ipsec/iph.h                        |  55 +++++-
 lib/librte_net/rte_ip.h                       |  49 +++++
 lib/meson.build                               |   2 +-
 12 files changed, 649 insertions(+), 104 deletions(-)
 create mode 100644 examples/ipsec-secgw/test/common_defs_secgw.sh
 create mode 100755 examples/ipsec-secgw/test/pkttest.py
 create mode 100755 examples/ipsec-secgw/test/pkttest.sh
 mode change 100644 => 100755 examples/ipsec-secgw/test/run_test.sh
 create mode 100755 examples/ipsec-secgw/test/trs_ipv6opts.py
Quoted 2 lines

--
2.17.1

Series Acked-by: Akhil Goyal <[email protected]>

The dependent patch will go in mainline and is a generic patch to fix the compilation. So not holding this series. Series Applied to dpdk-next-crypto

Thanks.