Files
vpp/test/template_ipsec.py

3310 lines
120 KiB
Python
Raw Normal View History

import unittest
import socket
import struct
import re
import os
from scapy.layers.inet import IP, ICMP, TCP, UDP
from scapy.layers.ipsec import SecurityAssociation, ESP
from scapy.layers.l2 import Ether
from scapy.packet import raw, Raw, Padding
from scapy.layers.inet6 import (
IPv6,
ICMPv6EchoRequest,
IPv6ExtHdrHopByHop,
IPv6ExtHdrFragment,
IPv6ExtHdrDestOpt,
)
2023-08-31 00:47:44 -04:00
from framework import VppTestCase
from asfframework import VppTestRunner
from util import ppp, reassemble4, fragment_rfc791, fragment_rfc8200
from vpp_papi import VppEnum
from vpp_ipsec import VppIpsecSpd, VppIpsecSpdEntry, VppIpsecSpdItfBinding
ipsec: Performance improvement of ipsec4_output_node using flow cache Adding flow cache support to improve outbound IPv4/IPSec SPD lookup performance. Details about flow cache: Mechanism: 1. First packet of a flow will undergo linear search in SPD table. Once a policy match is found, a new entry will be added into the flow cache. From 2nd packet onwards, the policy lookup will happen in flow cache. 2. The flow cache is implemented using bihash without collision handling. This will avoid the logic to age out or recycle the old flows in flow cache. Whenever a collision occurs, old entry will be overwritten by the new entry. Worst case is when all the 256 packets in a batch result in collision and fall back to linear search. Average and best case will be O(1). 3. The size of flow cache is fixed and decided based on the number of flows to be supported. The default is set to 1 million flows. This can be made as a configurable option as a next step. 4. Whenever a SPD rule is added/deleted by the control plane, the flow cache entries will be completely deleted (reset) in the control plane. The assumption here is that SPD rule add/del is not a frequent operation from control plane. Flow cache reset is done, by putting the data plane in fall back mode, to bypass flow cache and do linear search till the SPD rule add/delete operation is complete. Once the rule is successfully added/deleted, the data plane will be allowed to make use of the flow cache. The flow cache will be reset only after flushing out the inflight packets from all the worker cores using vlib_worker_wait_one_loop(). Details about bihash usage: 1. A new bihash template (16_8) is added to support IPv4 5 tuple. BIHASH_KVP_PER_PAGE and BIHASH_KVP_AT_BUCKET_LEVEL are set to 1 in the new template. It means only one KVP is supported per bucket. 2. Collision handling is avoided by calling BV (clib_bihash_add_or_overwrite_stale) function. Through the stale callback function pointer, the KVP entry will be overwritten during collision. 3. Flow cache reset is done using BV (clib_bihash_foreach_key_value_pair) function. Through the callback function pointer, the KVP value is reset to ~0ULL. MRR performance numbers with 1 core, 1 ESP Tunnel, null-encrypt, 64B for different SPD policy matching indices: SPD Policy index : 1 10 100 1000 Throughput : MPPS/MPPS MPPS/MPPS MPPS/MPPS KPPS/MPPS (Baseline/Optimized) ARM Neoverse N1 : 5.2/4.84 4.55/4.84 2.11/4.84 329.5/4.84 ARM TX2 : 2.81/2.6 2.51/2.6 1.27/2.6 176.62/2.6 INTEL SKX : 4.93/4.48 4.29/4.46 2.05/4.48 336.79/4.47 Next Steps: Following can be made as a configurable option through startup conf at IPSec level: 1. Enable/Disable Flow cache. 2. Bihash configuration like number of buckets and memory size. 3. Dual/Quad loop unroll can be applied around bihash to further improve the performance. 4. The same flow cache logic can be applied for IPv6 as well as in IPSec inbound direction. A deeper and wider flow cache using bihash_40_8 can replace existing bihash_16_8, to make it common for both IPv4 and IPv6 in both outbound and inbound directions. Following changes are made based on the review comments: 1. ON/OFF flow cache through startup conf. Default: OFF 2. Flow cache stale entry detection using epoch counter. 3. Avoid host order endianness conversion during flow cache lookup. 4. Move IPSec startup conf to a common file. 5. Added SPD flow cache unit test case 6. Replaced bihash with vectors to implement flow cache. 7. ipsec_add_del_policy API is not mpsafe. Cleaned up inflight packets check in control plane. Type: improvement Signed-off-by: mgovind <govindarajan.Mohandoss@arm.com> Signed-off-by: Zachary Leaf <zachary.leaf@arm.com> Tested-by: Jieqiang Wang <jieqiang.wang@arm.com> Change-Id: I62b4d6625fbc6caf292427a5d2046aa5672b2006
2021-03-19 19:20:49 +00:00
from ipaddress import ip_address
from config import config
ipsec: Performance improvement of ipsec4_output_node using flow cache Adding flow cache support to improve outbound IPv4/IPSec SPD lookup performance. Details about flow cache: Mechanism: 1. First packet of a flow will undergo linear search in SPD table. Once a policy match is found, a new entry will be added into the flow cache. From 2nd packet onwards, the policy lookup will happen in flow cache. 2. The flow cache is implemented using bihash without collision handling. This will avoid the logic to age out or recycle the old flows in flow cache. Whenever a collision occurs, old entry will be overwritten by the new entry. Worst case is when all the 256 packets in a batch result in collision and fall back to linear search. Average and best case will be O(1). 3. The size of flow cache is fixed and decided based on the number of flows to be supported. The default is set to 1 million flows. This can be made as a configurable option as a next step. 4. Whenever a SPD rule is added/deleted by the control plane, the flow cache entries will be completely deleted (reset) in the control plane. The assumption here is that SPD rule add/del is not a frequent operation from control plane. Flow cache reset is done, by putting the data plane in fall back mode, to bypass flow cache and do linear search till the SPD rule add/delete operation is complete. Once the rule is successfully added/deleted, the data plane will be allowed to make use of the flow cache. The flow cache will be reset only after flushing out the inflight packets from all the worker cores using vlib_worker_wait_one_loop(). Details about bihash usage: 1. A new bihash template (16_8) is added to support IPv4 5 tuple. BIHASH_KVP_PER_PAGE and BIHASH_KVP_AT_BUCKET_LEVEL are set to 1 in the new template. It means only one KVP is supported per bucket. 2. Collision handling is avoided by calling BV (clib_bihash_add_or_overwrite_stale) function. Through the stale callback function pointer, the KVP entry will be overwritten during collision. 3. Flow cache reset is done using BV (clib_bihash_foreach_key_value_pair) function. Through the callback function pointer, the KVP value is reset to ~0ULL. MRR performance numbers with 1 core, 1 ESP Tunnel, null-encrypt, 64B for different SPD policy matching indices: SPD Policy index : 1 10 100 1000 Throughput : MPPS/MPPS MPPS/MPPS MPPS/MPPS KPPS/MPPS (Baseline/Optimized) ARM Neoverse N1 : 5.2/4.84 4.55/4.84 2.11/4.84 329.5/4.84 ARM TX2 : 2.81/2.6 2.51/2.6 1.27/2.6 176.62/2.6 INTEL SKX : 4.93/4.48 4.29/4.46 2.05/4.48 336.79/4.47 Next Steps: Following can be made as a configurable option through startup conf at IPSec level: 1. Enable/Disable Flow cache. 2. Bihash configuration like number of buckets and memory size. 3. Dual/Quad loop unroll can be applied around bihash to further improve the performance. 4. The same flow cache logic can be applied for IPv6 as well as in IPSec inbound direction. A deeper and wider flow cache using bihash_40_8 can replace existing bihash_16_8, to make it common for both IPv4 and IPv6 in both outbound and inbound directions. Following changes are made based on the review comments: 1. ON/OFF flow cache through startup conf. Default: OFF 2. Flow cache stale entry detection using epoch counter. 3. Avoid host order endianness conversion during flow cache lookup. 4. Move IPSec startup conf to a common file. 5. Added SPD flow cache unit test case 6. Replaced bihash with vectors to implement flow cache. 7. ipsec_add_del_policy API is not mpsafe. Cleaned up inflight packets check in control plane. Type: improvement Signed-off-by: mgovind <govindarajan.Mohandoss@arm.com> Signed-off-by: Zachary Leaf <zachary.leaf@arm.com> Tested-by: Jieqiang Wang <jieqiang.wang@arm.com> Change-Id: I62b4d6625fbc6caf292427a5d2046aa5672b2006
2021-03-19 19:20:49 +00:00
class IPsecIPv4Params:
addr_type = socket.AF_INET
addr_any = "0.0.0.0"
addr_bcast = "255.255.255.255"
addr_len = 32
is_ipv6 = 0
def __init__(self):
self.remote_tun_if_host = "1.1.1.1"
self.remote_tun_if_host6 = "1111::1"
self.scapy_tun_sa_id = 100
self.scapy_tun_spi = 1000
self.vpp_tun_sa_id = 200
self.vpp_tun_spi = 2000
self.scapy_tra_sa_id = 300
self.scapy_tra_spi = 3000
self.vpp_tra_sa_id = 400
self.vpp_tra_spi = 4000
self.outer_hop_limit = 64
self.inner_hop_limit = 255
self.outer_flow_label = 0
self.inner_flow_label = 0x12345
self.anti_replay_window_size = 64
self.auth_algo_vpp_id = (
VppEnum.vl_api_ipsec_integ_alg_t.IPSEC_API_INTEG_ALG_SHA1_96
)
self.auth_algo = "HMAC-SHA1-96" # scapy name
self.auth_key = b"C91KUR9GYMm5GfkEvNjX"
self.crypt_algo_vpp_id = (
VppEnum.vl_api_ipsec_crypto_alg_t.IPSEC_API_CRYPTO_ALG_AES_CBC_128
)
self.crypt_algo = "AES-CBC" # scapy name
self.crypt_key = b"JPjyOWBeVEQiMe7h"
self.salt = 0
self.flags = 0
self.nat_header = None
self.tun_flags = (
VppEnum.vl_api_tunnel_encap_decap_flags_t.TUNNEL_API_ENCAP_DECAP_FLAG_NONE
)
self.dscp = 0
self.async_mode = False
class IPsecIPv6Params:
addr_type = socket.AF_INET6
addr_any = "0::0"
addr_bcast = "ffff:ffff:ffff:ffff:ffff:ffff:ffff:ffff"
addr_len = 128
is_ipv6 = 1
def __init__(self):
self.remote_tun_if_host = "1111:1111:1111:1111:1111:1111:1111:1111"
self.remote_tun_if_host4 = "1.1.1.1"
self.scapy_tun_sa_id = 500
self.scapy_tun_spi = 3001
self.vpp_tun_sa_id = 600
self.vpp_tun_spi = 3000
self.scapy_tra_sa_id = 700
self.scapy_tra_spi = 4001
self.vpp_tra_sa_id = 800
self.vpp_tra_spi = 4000
self.outer_hop_limit = 64
self.inner_hop_limit = 255
self.outer_flow_label = 0
self.inner_flow_label = 0x12345
self.anti_replay_window_size = 64
self.auth_algo_vpp_id = (
VppEnum.vl_api_ipsec_integ_alg_t.IPSEC_API_INTEG_ALG_SHA1_96
)
self.auth_algo = "HMAC-SHA1-96" # scapy name
self.auth_key = b"C91KUR9GYMm5GfkEvNjX"
self.crypt_algo_vpp_id = (
VppEnum.vl_api_ipsec_crypto_alg_t.IPSEC_API_CRYPTO_ALG_AES_CBC_128
)
self.crypt_algo = "AES-CBC" # scapy name
self.crypt_key = b"JPjyOWBeVEQiMe7h"
self.salt = 0
self.flags = 0
self.nat_header = None
self.tun_flags = (
VppEnum.vl_api_tunnel_encap_decap_flags_t.TUNNEL_API_ENCAP_DECAP_FLAG_NONE
)
self.dscp = 0
self.async_mode = False
def mk_scapy_crypt_key(p):
if p.crypt_algo in ("AES-GCM", "AES-CTR", "AES-NULL-GMAC"):
return p.crypt_key + struct.pack("!I", p.salt)
else:
return p.crypt_key
def config_tun_params(p, encryption_type, tun_if):
ip_class_by_addr_type = {socket.AF_INET: IP, socket.AF_INET6: IPv6}
esn_en = bool(
p.flags & (VppEnum.vl_api_ipsec_sad_flags_t.IPSEC_API_SAD_FLAG_USE_ESN)
)
p.tun_dst = tun_if.remote_addr[p.addr_type]
p.tun_src = tun_if.local_addr[p.addr_type]
crypt_key = mk_scapy_crypt_key(p)
p.scapy_tun_sa = SecurityAssociation(
encryption_type,
spi=p.scapy_tun_spi,
crypt_algo=p.crypt_algo,
crypt_key=crypt_key,
auth_algo=p.auth_algo,
auth_key=p.auth_key,
tunnel_header=ip_class_by_addr_type[p.addr_type](src=p.tun_dst, dst=p.tun_src),
nat_t_header=p.nat_header,
esn_en=esn_en,
)
p.vpp_tun_sa = SecurityAssociation(
encryption_type,
spi=p.vpp_tun_spi,
crypt_algo=p.crypt_algo,
crypt_key=crypt_key,
auth_algo=p.auth_algo,
auth_key=p.auth_key,
tunnel_header=ip_class_by_addr_type[p.addr_type](dst=p.tun_dst, src=p.tun_src),
nat_t_header=p.nat_header,
esn_en=esn_en,
)
def config_tra_params(p, encryption_type):
esn_en = bool(
p.flags & (VppEnum.vl_api_ipsec_sad_flags_t.IPSEC_API_SAD_FLAG_USE_ESN)
)
crypt_key = mk_scapy_crypt_key(p)
p.scapy_tra_sa = SecurityAssociation(
encryption_type,
spi=p.scapy_tra_spi,
crypt_algo=p.crypt_algo,
crypt_key=crypt_key,
auth_algo=p.auth_algo,
auth_key=p.auth_key,
nat_t_header=p.nat_header,
esn_en=esn_en,
)
p.vpp_tra_sa = SecurityAssociation(
encryption_type,
spi=p.vpp_tra_spi,
crypt_algo=p.crypt_algo,
crypt_key=crypt_key,
auth_algo=p.auth_algo,
auth_key=p.auth_key,
nat_t_header=p.nat_header,
esn_en=esn_en,
)
class TemplateIpsec(VppTestCase):
"""
TRANSPORT MODE::
------ encrypt ---
|tra_if| <-------> |VPP|
------ decrypt ---
TUNNEL MODE::
------ encrypt --- plain ---
|tun_if| <------- |VPP| <------ |pg1|
------ --- ---
------ decrypt --- plain ---
|tun_if| -------> |VPP| ------> |pg1|
------ --- ---
"""
tun_spd_id = 1
tra_spd_id = 2
def ipsec_select_backend(self):
"""empty method to be overloaded when necessary"""
pass
@classmethod
def setUpClass(cls):
super(TemplateIpsec, cls).setUpClass()
@classmethod
def tearDownClass(cls):
super(TemplateIpsec, cls).tearDownClass()
def setup_params(self):
if not hasattr(self, "ipv4_params"):
self.ipv4_params = IPsecIPv4Params()
if not hasattr(self, "ipv6_params"):
self.ipv6_params = IPsecIPv6Params()
self.params = {
self.ipv4_params.addr_type: self.ipv4_params,
self.ipv6_params.addr_type: self.ipv6_params,
}
def config_interfaces(self):
self.create_pg_interfaces(range(3))
self.interfaces = list(self.pg_interfaces)
for i in self.interfaces:
i.admin_up()
i.config_ip4()
i.resolve_arp()
i.config_ip6()
i.resolve_ndp()
def setUp(self):
super(TemplateIpsec, self).setUp()
self.setup_params()
self.vpp_esp_protocol = VppEnum.vl_api_ipsec_proto_t.IPSEC_API_PROTO_ESP
self.vpp_ah_protocol = VppEnum.vl_api_ipsec_proto_t.IPSEC_API_PROTO_AH
self.config_interfaces()
self.ipsec_select_backend()
def unconfig_interfaces(self):
for i in self.interfaces:
i.admin_down()
i.unconfig_ip4()
i.unconfig_ip6()
def tearDown(self):
super(TemplateIpsec, self).tearDown()
self.unconfig_interfaces()
def show_commands_at_teardown(self):
self.logger.info(self.vapi.cli("show hardware"))
def gen_encrypt_pkts(self, p, sa, sw_intf, src, dst, count=1, payload_size=54):
return [
Ether(src=sw_intf.remote_mac, dst=sw_intf.local_mac)
/ sa.encrypt(IP(src=src, dst=dst) / ICMP() / Raw(b"X" * payload_size))
for i in range(count)
]
def gen_encrypt_pkts6(self, p, sa, sw_intf, src, dst, count=1, payload_size=54):
return [
Ether(src=sw_intf.remote_mac, dst=sw_intf.local_mac)
/ sa.encrypt(
IPv6(src=src, dst=dst, hlim=p.inner_hop_limit, fl=p.inner_flow_label)
/ ICMPv6EchoRequest(id=0, seq=1, data="X" * payload_size)
)
for i in range(count)
]
def gen_pkts(self, sw_intf, src, dst, count=1, payload_size=54):
return [
Ether(src=sw_intf.remote_mac, dst=sw_intf.local_mac)
/ IP(src=src, dst=dst)
/ ICMP()
/ Raw(b"X" * payload_size)
for i in range(count)
]
def gen_pkts6(self, p, sw_intf, src, dst, count=1, payload_size=54):
return [
Ether(src=sw_intf.remote_mac, dst=sw_intf.local_mac)
/ IPv6(src=src, dst=dst, hlim=p.inner_hop_limit, fl=p.inner_flow_label)
/ ICMPv6EchoRequest(id=0, seq=1, data="X" * payload_size)
for i in range(count)
]
class IpsecTcp(object):
def verify_tcp_checksum(self):
# start http cli server listener on http://0.0.0.0:80
self.vapi.cli("http cli server")
p = self.params[socket.AF_INET]
send = Ether(
src=self.tun_if.remote_mac, dst=self.tun_if.local_mac
) / p.scapy_tun_sa.encrypt(
IP(src=p.remote_tun_if_host, dst=self.tun_if.local_ip4)
/ TCP(flags="S", dport=80)
)
self.logger.debug(ppp("Sending packet:", send))
recv = self.send_and_expect(self.tun_if, [send], self.tun_if)
recv = recv[0]
decrypted = p.vpp_tun_sa.decrypt(recv[IP])
self.assert_packet_checksums_valid(decrypted)
@unittest.skipIf(
"hs_apps" in config.excluded_plugins, "Exclude tests requiring hs_apps plugin"
)
class IpsecTcpTests(IpsecTcp):
def test_tcp_checksum(self):
"""verify checksum correctness for vpp generated packets"""
self.verify_tcp_checksum()
class IpsecTra4(object):
"""verify methods for Transport v4"""
def get_replay_counts(self, p):
replay_node_name = "/err/%s/replay" % self.tra4_decrypt_node_name[0]
count = self.statistics.get_err_counter(replay_node_name)
if p.async_mode:
replay_post_node_name = (
"/err/%s/replay" % self.tra4_decrypt_node_name[p.async_mode]
)
count += self.statistics.get_err_counter(replay_post_node_name)
return count
def get_hash_failed_counts(self, p):
if ESP == self.encryption_type and p.crypt_algo in ("AES-GCM", "AES-NULL-GMAC"):
hash_failed_node_name = (
"/err/%s/decryption_failed" % self.tra4_decrypt_node_name[p.async_mode]
)
else:
hash_failed_node_name = (
"/err/%s/integ_error" % self.tra4_decrypt_node_name[p.async_mode]
)
count = self.statistics.get_err_counter(hash_failed_node_name)
if p.async_mode:
count += self.statistics.get_err_counter("/err/crypto-dispatch/bad-hmac")
return count
ipsec: Fix setting the hi-sequence number for decrypt Type: fix two problems; 1 - just because anti-reply is not enabled doesn't mean the high sequence number should not be used. - fix, there needs to be some means to detect a wrapped packet, so we use a window size of 2^30. 2 - The SA object was used as a scratch pad for the high-sequence number used during decryption. That means that once the batch has been processed the high-sequence number used is lost. This means it is not possible to distinguish this case: if (seq < IPSEC_SA_ANTI_REPLAY_WINDOW_LOWER_BOUND (tl)) { ... if (post_decrypt) { if (hi_seq_used == sa->seq_hi) /* the high sequence number used to succesfully decrypt this * packet is the same as the last-sequnence number of the SA. * that means this packet did not cause a wrap. * this packet is thus out of window and should be dropped */ return 1; else /* The packet decrypted with a different high sequence number * to the SA, that means it is the wrap packet and should be * accepted */ return 0; } - fix: don't use the SA as a scratch pad, use the 'packet_data' - the same place that is used as the scratch pad for the low sequence number. other consequences: - An SA doesn't have seq and last_seq, it has only seq; the sequence numnber of the last packet tx'd or rx'd. - there's 64bits of space available on the SA's first cache line. move the AES CTR mode IV there. - test the ESN/AR combinations to catch the bugs this fixes. This doubles the amount of tests, but without AR on they only run for 2 seconds. In the AR tests, the time taken to wait for packets that won't arrive is dropped from 1 to 0.2 seconds thus reducing the runtime of these tests from 10-15 to about 5 sceonds. Signed-off-by: Neale Ranns <neale@graphiant.com> Change-Id: Iaac78905289a272dc01930d70decd8109cf5e7a5
2021-06-28 13:31:28 +00:00
def verify_hi_seq_num(self):
p = self.params[socket.AF_INET]
saf = VppEnum.vl_api_ipsec_sad_flags_t
esn_on = p.vpp_tra_sa.esn_en
ar_on = p.flags & saf.IPSEC_API_SAD_FLAG_USE_ANTI_REPLAY
seq_cycle_node_name = "/err/%s/seq_cycled" % self.tra4_encrypt_node_name
ipsec: Fix setting the hi-sequence number for decrypt Type: fix two problems; 1 - just because anti-reply is not enabled doesn't mean the high sequence number should not be used. - fix, there needs to be some means to detect a wrapped packet, so we use a window size of 2^30. 2 - The SA object was used as a scratch pad for the high-sequence number used during decryption. That means that once the batch has been processed the high-sequence number used is lost. This means it is not possible to distinguish this case: if (seq < IPSEC_SA_ANTI_REPLAY_WINDOW_LOWER_BOUND (tl)) { ... if (post_decrypt) { if (hi_seq_used == sa->seq_hi) /* the high sequence number used to succesfully decrypt this * packet is the same as the last-sequnence number of the SA. * that means this packet did not cause a wrap. * this packet is thus out of window and should be dropped */ return 1; else /* The packet decrypted with a different high sequence number * to the SA, that means it is the wrap packet and should be * accepted */ return 0; } - fix: don't use the SA as a scratch pad, use the 'packet_data' - the same place that is used as the scratch pad for the low sequence number. other consequences: - An SA doesn't have seq and last_seq, it has only seq; the sequence numnber of the last packet tx'd or rx'd. - there's 64bits of space available on the SA's first cache line. move the AES CTR mode IV there. - test the ESN/AR combinations to catch the bugs this fixes. This doubles the amount of tests, but without AR on they only run for 2 seconds. In the AR tests, the time taken to wait for packets that won't arrive is dropped from 1 to 0.2 seconds thus reducing the runtime of these tests from 10-15 to about 5 sceonds. Signed-off-by: Neale Ranns <neale@graphiant.com> Change-Id: Iaac78905289a272dc01930d70decd8109cf5e7a5
2021-06-28 13:31:28 +00:00
replay_count = self.get_replay_counts(p)
hash_failed_count = self.get_hash_failed_counts(p)
seq_cycle_count = self.statistics.get_err_counter(seq_cycle_node_name)
# a few packets so we get the rx seq number above the window size and
# thus can simulate a wrap with an out of window packet
pkts = [
(
Ether(src=self.tra_if.remote_mac, dst=self.tra_if.local_mac)
/ p.scapy_tra_sa.encrypt(
IP(src=self.tra_if.remote_ip4, dst=self.tra_if.local_ip4) / ICMP(),
seq_num=seq,
)
)
for seq in range(63, 80)
]
ipsec: Fix setting the hi-sequence number for decrypt Type: fix two problems; 1 - just because anti-reply is not enabled doesn't mean the high sequence number should not be used. - fix, there needs to be some means to detect a wrapped packet, so we use a window size of 2^30. 2 - The SA object was used as a scratch pad for the high-sequence number used during decryption. That means that once the batch has been processed the high-sequence number used is lost. This means it is not possible to distinguish this case: if (seq < IPSEC_SA_ANTI_REPLAY_WINDOW_LOWER_BOUND (tl)) { ... if (post_decrypt) { if (hi_seq_used == sa->seq_hi) /* the high sequence number used to succesfully decrypt this * packet is the same as the last-sequnence number of the SA. * that means this packet did not cause a wrap. * this packet is thus out of window and should be dropped */ return 1; else /* The packet decrypted with a different high sequence number * to the SA, that means it is the wrap packet and should be * accepted */ return 0; } - fix: don't use the SA as a scratch pad, use the 'packet_data' - the same place that is used as the scratch pad for the low sequence number. other consequences: - An SA doesn't have seq and last_seq, it has only seq; the sequence numnber of the last packet tx'd or rx'd. - there's 64bits of space available on the SA's first cache line. move the AES CTR mode IV there. - test the ESN/AR combinations to catch the bugs this fixes. This doubles the amount of tests, but without AR on they only run for 2 seconds. In the AR tests, the time taken to wait for packets that won't arrive is dropped from 1 to 0.2 seconds thus reducing the runtime of these tests from 10-15 to about 5 sceonds. Signed-off-by: Neale Ranns <neale@graphiant.com> Change-Id: Iaac78905289a272dc01930d70decd8109cf5e7a5
2021-06-28 13:31:28 +00:00
recv_pkts = self.send_and_expect(self.tra_if, pkts, self.tra_if)
# these 4 packets will all choose seq-num 0 to decrpyt since none
# are out of window when first checked. however, once #200 has
# decrypted it will move the window to 200 and has #81 is out of
# window. this packet should be dropped.
pkts = [
(
Ether(src=self.tra_if.remote_mac, dst=self.tra_if.local_mac)
/ p.scapy_tra_sa.encrypt(
IP(src=self.tra_if.remote_ip4, dst=self.tra_if.local_ip4) / ICMP(),
seq_num=200,
)
),
(
Ether(src=self.tra_if.remote_mac, dst=self.tra_if.local_mac)
/ p.scapy_tra_sa.encrypt(
IP(src=self.tra_if.remote_ip4, dst=self.tra_if.local_ip4) / ICMP(),
seq_num=81,
)
),
(
Ether(src=self.tra_if.remote_mac, dst=self.tra_if.local_mac)
/ p.scapy_tra_sa.encrypt(
IP(src=self.tra_if.remote_ip4, dst=self.tra_if.local_ip4) / ICMP(),
seq_num=201,
)
),
(
Ether(src=self.tra_if.remote_mac, dst=self.tra_if.local_mac)
/ p.scapy_tra_sa.encrypt(
IP(src=self.tra_if.remote_ip4, dst=self.tra_if.local_ip4) / ICMP(),
seq_num=202,
)
),
]
ipsec: Fix setting the hi-sequence number for decrypt Type: fix two problems; 1 - just because anti-reply is not enabled doesn't mean the high sequence number should not be used. - fix, there needs to be some means to detect a wrapped packet, so we use a window size of 2^30. 2 - The SA object was used as a scratch pad for the high-sequence number used during decryption. That means that once the batch has been processed the high-sequence number used is lost. This means it is not possible to distinguish this case: if (seq < IPSEC_SA_ANTI_REPLAY_WINDOW_LOWER_BOUND (tl)) { ... if (post_decrypt) { if (hi_seq_used == sa->seq_hi) /* the high sequence number used to succesfully decrypt this * packet is the same as the last-sequnence number of the SA. * that means this packet did not cause a wrap. * this packet is thus out of window and should be dropped */ return 1; else /* The packet decrypted with a different high sequence number * to the SA, that means it is the wrap packet and should be * accepted */ return 0; } - fix: don't use the SA as a scratch pad, use the 'packet_data' - the same place that is used as the scratch pad for the low sequence number. other consequences: - An SA doesn't have seq and last_seq, it has only seq; the sequence numnber of the last packet tx'd or rx'd. - there's 64bits of space available on the SA's first cache line. move the AES CTR mode IV there. - test the ESN/AR combinations to catch the bugs this fixes. This doubles the amount of tests, but without AR on they only run for 2 seconds. In the AR tests, the time taken to wait for packets that won't arrive is dropped from 1 to 0.2 seconds thus reducing the runtime of these tests from 10-15 to about 5 sceonds. Signed-off-by: Neale Ranns <neale@graphiant.com> Change-Id: Iaac78905289a272dc01930d70decd8109cf5e7a5
2021-06-28 13:31:28 +00:00
# if anti-replay is off then we won't drop #81
n_rx = 3 if ar_on else 4
self.send_and_expect(self.tra_if, pkts, self.tra_if, n_rx=n_rx)
# this packet is one before the wrap
pkts = [
(
Ether(src=self.tra_if.remote_mac, dst=self.tra_if.local_mac)
/ p.scapy_tra_sa.encrypt(
IP(src=self.tra_if.remote_ip4, dst=self.tra_if.local_ip4) / ICMP(),
seq_num=203,
)
)
]
ipsec: Fix setting the hi-sequence number for decrypt Type: fix two problems; 1 - just because anti-reply is not enabled doesn't mean the high sequence number should not be used. - fix, there needs to be some means to detect a wrapped packet, so we use a window size of 2^30. 2 - The SA object was used as a scratch pad for the high-sequence number used during decryption. That means that once the batch has been processed the high-sequence number used is lost. This means it is not possible to distinguish this case: if (seq < IPSEC_SA_ANTI_REPLAY_WINDOW_LOWER_BOUND (tl)) { ... if (post_decrypt) { if (hi_seq_used == sa->seq_hi) /* the high sequence number used to succesfully decrypt this * packet is the same as the last-sequnence number of the SA. * that means this packet did not cause a wrap. * this packet is thus out of window and should be dropped */ return 1; else /* The packet decrypted with a different high sequence number * to the SA, that means it is the wrap packet and should be * accepted */ return 0; } - fix: don't use the SA as a scratch pad, use the 'packet_data' - the same place that is used as the scratch pad for the low sequence number. other consequences: - An SA doesn't have seq and last_seq, it has only seq; the sequence numnber of the last packet tx'd or rx'd. - there's 64bits of space available on the SA's first cache line. move the AES CTR mode IV there. - test the ESN/AR combinations to catch the bugs this fixes. This doubles the amount of tests, but without AR on they only run for 2 seconds. In the AR tests, the time taken to wait for packets that won't arrive is dropped from 1 to 0.2 seconds thus reducing the runtime of these tests from 10-15 to about 5 sceonds. Signed-off-by: Neale Ranns <neale@graphiant.com> Change-Id: Iaac78905289a272dc01930d70decd8109cf5e7a5
2021-06-28 13:31:28 +00:00
recv_pkts = self.send_and_expect(self.tra_if, pkts, self.tra_if)
# a replayed packet, then an out of window, then a legit
# tests that a early failure on the batch doesn't affect subsequent packets.
pkts = [
(
Ether(src=self.tra_if.remote_mac, dst=self.tra_if.local_mac)
/ p.scapy_tra_sa.encrypt(
IP(src=self.tra_if.remote_ip4, dst=self.tra_if.local_ip4) / ICMP(),
seq_num=203,
)
),
(
Ether(src=self.tra_if.remote_mac, dst=self.tra_if.local_mac)
/ p.scapy_tra_sa.encrypt(
IP(src=self.tra_if.remote_ip4, dst=self.tra_if.local_ip4) / ICMP(),
seq_num=81,
)
),
(
Ether(src=self.tra_if.remote_mac, dst=self.tra_if.local_mac)
/ p.scapy_tra_sa.encrypt(
IP(src=self.tra_if.remote_ip4, dst=self.tra_if.local_ip4) / ICMP(),
seq_num=204,
)
),
]
n_rx = 1 if ar_on else 3
recv_pkts = self.send_and_expect(self.tra_if, pkts, self.tra_if, n_rx=n_rx)
ipsec: Fix setting the hi-sequence number for decrypt Type: fix two problems; 1 - just because anti-reply is not enabled doesn't mean the high sequence number should not be used. - fix, there needs to be some means to detect a wrapped packet, so we use a window size of 2^30. 2 - The SA object was used as a scratch pad for the high-sequence number used during decryption. That means that once the batch has been processed the high-sequence number used is lost. This means it is not possible to distinguish this case: if (seq < IPSEC_SA_ANTI_REPLAY_WINDOW_LOWER_BOUND (tl)) { ... if (post_decrypt) { if (hi_seq_used == sa->seq_hi) /* the high sequence number used to succesfully decrypt this * packet is the same as the last-sequnence number of the SA. * that means this packet did not cause a wrap. * this packet is thus out of window and should be dropped */ return 1; else /* The packet decrypted with a different high sequence number * to the SA, that means it is the wrap packet and should be * accepted */ return 0; } - fix: don't use the SA as a scratch pad, use the 'packet_data' - the same place that is used as the scratch pad for the low sequence number. other consequences: - An SA doesn't have seq and last_seq, it has only seq; the sequence numnber of the last packet tx'd or rx'd. - there's 64bits of space available on the SA's first cache line. move the AES CTR mode IV there. - test the ESN/AR combinations to catch the bugs this fixes. This doubles the amount of tests, but without AR on they only run for 2 seconds. In the AR tests, the time taken to wait for packets that won't arrive is dropped from 1 to 0.2 seconds thus reducing the runtime of these tests from 10-15 to about 5 sceonds. Signed-off-by: Neale Ranns <neale@graphiant.com> Change-Id: Iaac78905289a272dc01930d70decd8109cf5e7a5
2021-06-28 13:31:28 +00:00
# move the window over half way to a wrap
pkts = [
(
Ether(src=self.tra_if.remote_mac, dst=self.tra_if.local_mac)
/ p.scapy_tra_sa.encrypt(
IP(src=self.tra_if.remote_ip4, dst=self.tra_if.local_ip4) / ICMP(),
seq_num=0x80000001,
)
)
]
ipsec: Fix setting the hi-sequence number for decrypt Type: fix two problems; 1 - just because anti-reply is not enabled doesn't mean the high sequence number should not be used. - fix, there needs to be some means to detect a wrapped packet, so we use a window size of 2^30. 2 - The SA object was used as a scratch pad for the high-sequence number used during decryption. That means that once the batch has been processed the high-sequence number used is lost. This means it is not possible to distinguish this case: if (seq < IPSEC_SA_ANTI_REPLAY_WINDOW_LOWER_BOUND (tl)) { ... if (post_decrypt) { if (hi_seq_used == sa->seq_hi) /* the high sequence number used to succesfully decrypt this * packet is the same as the last-sequnence number of the SA. * that means this packet did not cause a wrap. * this packet is thus out of window and should be dropped */ return 1; else /* The packet decrypted with a different high sequence number * to the SA, that means it is the wrap packet and should be * accepted */ return 0; } - fix: don't use the SA as a scratch pad, use the 'packet_data' - the same place that is used as the scratch pad for the low sequence number. other consequences: - An SA doesn't have seq and last_seq, it has only seq; the sequence numnber of the last packet tx'd or rx'd. - there's 64bits of space available on the SA's first cache line. move the AES CTR mode IV there. - test the ESN/AR combinations to catch the bugs this fixes. This doubles the amount of tests, but without AR on they only run for 2 seconds. In the AR tests, the time taken to wait for packets that won't arrive is dropped from 1 to 0.2 seconds thus reducing the runtime of these tests from 10-15 to about 5 sceonds. Signed-off-by: Neale Ranns <neale@graphiant.com> Change-Id: Iaac78905289a272dc01930d70decd8109cf5e7a5
2021-06-28 13:31:28 +00:00
recv_pkts = self.send_and_expect(self.tra_if, pkts, self.tra_if)
# anti-replay will drop old packets, no anti-replay will not
pkts = [
(
Ether(src=self.tra_if.remote_mac, dst=self.tra_if.local_mac)
/ p.scapy_tra_sa.encrypt(
IP(src=self.tra_if.remote_ip4, dst=self.tra_if.local_ip4) / ICMP(),
seq_num=0x44000001,
)
)
]
ipsec: Fix setting the hi-sequence number for decrypt Type: fix two problems; 1 - just because anti-reply is not enabled doesn't mean the high sequence number should not be used. - fix, there needs to be some means to detect a wrapped packet, so we use a window size of 2^30. 2 - The SA object was used as a scratch pad for the high-sequence number used during decryption. That means that once the batch has been processed the high-sequence number used is lost. This means it is not possible to distinguish this case: if (seq < IPSEC_SA_ANTI_REPLAY_WINDOW_LOWER_BOUND (tl)) { ... if (post_decrypt) { if (hi_seq_used == sa->seq_hi) /* the high sequence number used to succesfully decrypt this * packet is the same as the last-sequnence number of the SA. * that means this packet did not cause a wrap. * this packet is thus out of window and should be dropped */ return 1; else /* The packet decrypted with a different high sequence number * to the SA, that means it is the wrap packet and should be * accepted */ return 0; } - fix: don't use the SA as a scratch pad, use the 'packet_data' - the same place that is used as the scratch pad for the low sequence number. other consequences: - An SA doesn't have seq and last_seq, it has only seq; the sequence numnber of the last packet tx'd or rx'd. - there's 64bits of space available on the SA's first cache line. move the AES CTR mode IV there. - test the ESN/AR combinations to catch the bugs this fixes. This doubles the amount of tests, but without AR on they only run for 2 seconds. In the AR tests, the time taken to wait for packets that won't arrive is dropped from 1 to 0.2 seconds thus reducing the runtime of these tests from 10-15 to about 5 sceonds. Signed-off-by: Neale Ranns <neale@graphiant.com> Change-Id: Iaac78905289a272dc01930d70decd8109cf5e7a5
2021-06-28 13:31:28 +00:00
if ar_on:
self.send_and_assert_no_replies(self.tra_if, pkts)
else:
recv_pkts = self.send_and_expect(self.tra_if, pkts, self.tra_if)
if esn_on:
#
# validate wrapping the ESN
#
# wrap scapy's TX SA SN
p.scapy_tra_sa.seq_num = 0x100000005
# send a packet that wraps the window for both AR and no AR
pkts = [
(
Ether(src=self.tra_if.remote_mac, dst=self.tra_if.local_mac)
/ p.scapy_tra_sa.encrypt(
IP(src=self.tra_if.remote_ip4, dst=self.tra_if.local_ip4)
/ ICMP(),
seq_num=0x100000005,
)
)
]
ipsec: Fix setting the hi-sequence number for decrypt Type: fix two problems; 1 - just because anti-reply is not enabled doesn't mean the high sequence number should not be used. - fix, there needs to be some means to detect a wrapped packet, so we use a window size of 2^30. 2 - The SA object was used as a scratch pad for the high-sequence number used during decryption. That means that once the batch has been processed the high-sequence number used is lost. This means it is not possible to distinguish this case: if (seq < IPSEC_SA_ANTI_REPLAY_WINDOW_LOWER_BOUND (tl)) { ... if (post_decrypt) { if (hi_seq_used == sa->seq_hi) /* the high sequence number used to succesfully decrypt this * packet is the same as the last-sequnence number of the SA. * that means this packet did not cause a wrap. * this packet is thus out of window and should be dropped */ return 1; else /* The packet decrypted with a different high sequence number * to the SA, that means it is the wrap packet and should be * accepted */ return 0; } - fix: don't use the SA as a scratch pad, use the 'packet_data' - the same place that is used as the scratch pad for the low sequence number. other consequences: - An SA doesn't have seq and last_seq, it has only seq; the sequence numnber of the last packet tx'd or rx'd. - there's 64bits of space available on the SA's first cache line. move the AES CTR mode IV there. - test the ESN/AR combinations to catch the bugs this fixes. This doubles the amount of tests, but without AR on they only run for 2 seconds. In the AR tests, the time taken to wait for packets that won't arrive is dropped from 1 to 0.2 seconds thus reducing the runtime of these tests from 10-15 to about 5 sceonds. Signed-off-by: Neale Ranns <neale@graphiant.com> Change-Id: Iaac78905289a272dc01930d70decd8109cf5e7a5
2021-06-28 13:31:28 +00:00
rxs = self.send_and_expect(self.tra_if, pkts, self.tra_if)
for rx in rxs:
decrypted = p.vpp_tra_sa.decrypt(rx[0][IP])
# move the window forward to half way to the next wrap
pkts = [
(
Ether(src=self.tra_if.remote_mac, dst=self.tra_if.local_mac)
/ p.scapy_tra_sa.encrypt(
IP(src=self.tra_if.remote_ip4, dst=self.tra_if.local_ip4)
/ ICMP(),
seq_num=0x180000005,
)
)
]
ipsec: Fix setting the hi-sequence number for decrypt Type: fix two problems; 1 - just because anti-reply is not enabled doesn't mean the high sequence number should not be used. - fix, there needs to be some means to detect a wrapped packet, so we use a window size of 2^30. 2 - The SA object was used as a scratch pad for the high-sequence number used during decryption. That means that once the batch has been processed the high-sequence number used is lost. This means it is not possible to distinguish this case: if (seq < IPSEC_SA_ANTI_REPLAY_WINDOW_LOWER_BOUND (tl)) { ... if (post_decrypt) { if (hi_seq_used == sa->seq_hi) /* the high sequence number used to succesfully decrypt this * packet is the same as the last-sequnence number of the SA. * that means this packet did not cause a wrap. * this packet is thus out of window and should be dropped */ return 1; else /* The packet decrypted with a different high sequence number * to the SA, that means it is the wrap packet and should be * accepted */ return 0; } - fix: don't use the SA as a scratch pad, use the 'packet_data' - the same place that is used as the scratch pad for the low sequence number. other consequences: - An SA doesn't have seq and last_seq, it has only seq; the sequence numnber of the last packet tx'd or rx'd. - there's 64bits of space available on the SA's first cache line. move the AES CTR mode IV there. - test the ESN/AR combinations to catch the bugs this fixes. This doubles the amount of tests, but without AR on they only run for 2 seconds. In the AR tests, the time taken to wait for packets that won't arrive is dropped from 1 to 0.2 seconds thus reducing the runtime of these tests from 10-15 to about 5 sceonds. Signed-off-by: Neale Ranns <neale@graphiant.com> Change-Id: Iaac78905289a272dc01930d70decd8109cf5e7a5
2021-06-28 13:31:28 +00:00
rxs = self.send_and_expect(self.tra_if, pkts, self.tra_if)
# a packet less than 2^30 from the current position is:
# - AR: out of window and dropped
# - non-AR: accepted
pkts = [
(
Ether(src=self.tra_if.remote_mac, dst=self.tra_if.local_mac)
/ p.scapy_tra_sa.encrypt(
IP(src=self.tra_if.remote_ip4, dst=self.tra_if.local_ip4)
/ ICMP(),
seq_num=0x170000005,
)
)
]
ipsec: Fix setting the hi-sequence number for decrypt Type: fix two problems; 1 - just because anti-reply is not enabled doesn't mean the high sequence number should not be used. - fix, there needs to be some means to detect a wrapped packet, so we use a window size of 2^30. 2 - The SA object was used as a scratch pad for the high-sequence number used during decryption. That means that once the batch has been processed the high-sequence number used is lost. This means it is not possible to distinguish this case: if (seq < IPSEC_SA_ANTI_REPLAY_WINDOW_LOWER_BOUND (tl)) { ... if (post_decrypt) { if (hi_seq_used == sa->seq_hi) /* the high sequence number used to succesfully decrypt this * packet is the same as the last-sequnence number of the SA. * that means this packet did not cause a wrap. * this packet is thus out of window and should be dropped */ return 1; else /* The packet decrypted with a different high sequence number * to the SA, that means it is the wrap packet and should be * accepted */ return 0; } - fix: don't use the SA as a scratch pad, use the 'packet_data' - the same place that is used as the scratch pad for the low sequence number. other consequences: - An SA doesn't have seq and last_seq, it has only seq; the sequence numnber of the last packet tx'd or rx'd. - there's 64bits of space available on the SA's first cache line. move the AES CTR mode IV there. - test the ESN/AR combinations to catch the bugs this fixes. This doubles the amount of tests, but without AR on they only run for 2 seconds. In the AR tests, the time taken to wait for packets that won't arrive is dropped from 1 to 0.2 seconds thus reducing the runtime of these tests from 10-15 to about 5 sceonds. Signed-off-by: Neale Ranns <neale@graphiant.com> Change-Id: Iaac78905289a272dc01930d70decd8109cf5e7a5
2021-06-28 13:31:28 +00:00
if ar_on:
self.send_and_assert_no_replies(self.tra_if, pkts)
else:
self.send_and_expect(self.tra_if, pkts, self.tra_if)
# a packet more than 2^30 from the current position is:
# - AR: out of window and dropped
# - non-AR: considered a wrap, but since it's not a wrap
# it won't decrpyt and so will be dropped
pkts = [
(
Ether(src=self.tra_if.remote_mac, dst=self.tra_if.local_mac)
/ p.scapy_tra_sa.encrypt(
IP(src=self.tra_if.remote_ip4, dst=self.tra_if.local_ip4)
/ ICMP(),
seq_num=0x130000005,
)
)
]
ipsec: Fix setting the hi-sequence number for decrypt Type: fix two problems; 1 - just because anti-reply is not enabled doesn't mean the high sequence number should not be used. - fix, there needs to be some means to detect a wrapped packet, so we use a window size of 2^30. 2 - The SA object was used as a scratch pad for the high-sequence number used during decryption. That means that once the batch has been processed the high-sequence number used is lost. This means it is not possible to distinguish this case: if (seq < IPSEC_SA_ANTI_REPLAY_WINDOW_LOWER_BOUND (tl)) { ... if (post_decrypt) { if (hi_seq_used == sa->seq_hi) /* the high sequence number used to succesfully decrypt this * packet is the same as the last-sequnence number of the SA. * that means this packet did not cause a wrap. * this packet is thus out of window and should be dropped */ return 1; else /* The packet decrypted with a different high sequence number * to the SA, that means it is the wrap packet and should be * accepted */ return 0; } - fix: don't use the SA as a scratch pad, use the 'packet_data' - the same place that is used as the scratch pad for the low sequence number. other consequences: - An SA doesn't have seq and last_seq, it has only seq; the sequence numnber of the last packet tx'd or rx'd. - there's 64bits of space available on the SA's first cache line. move the AES CTR mode IV there. - test the ESN/AR combinations to catch the bugs this fixes. This doubles the amount of tests, but without AR on they only run for 2 seconds. In the AR tests, the time taken to wait for packets that won't arrive is dropped from 1 to 0.2 seconds thus reducing the runtime of these tests from 10-15 to about 5 sceonds. Signed-off-by: Neale Ranns <neale@graphiant.com> Change-Id: Iaac78905289a272dc01930d70decd8109cf5e7a5
2021-06-28 13:31:28 +00:00
self.send_and_assert_no_replies(self.tra_if, pkts)
# a packet less than 2^30 from the current position and is a
# wrap; (the seq is currently at 0x180000005).
# - AR: out of window so considered a wrap, so accepted
# - non-AR: not considered a wrap, so won't decrypt
p.scapy_tra_sa.seq_num = 0x260000005
pkts = [
(
Ether(src=self.tra_if.remote_mac, dst=self.tra_if.local_mac)
/ p.scapy_tra_sa.encrypt(
IP(src=self.tra_if.remote_ip4, dst=self.tra_if.local_ip4)
/ ICMP(),
seq_num=0x260000005,
)
)
]
ipsec: Fix setting the hi-sequence number for decrypt Type: fix two problems; 1 - just because anti-reply is not enabled doesn't mean the high sequence number should not be used. - fix, there needs to be some means to detect a wrapped packet, so we use a window size of 2^30. 2 - The SA object was used as a scratch pad for the high-sequence number used during decryption. That means that once the batch has been processed the high-sequence number used is lost. This means it is not possible to distinguish this case: if (seq < IPSEC_SA_ANTI_REPLAY_WINDOW_LOWER_BOUND (tl)) { ... if (post_decrypt) { if (hi_seq_used == sa->seq_hi) /* the high sequence number used to succesfully decrypt this * packet is the same as the last-sequnence number of the SA. * that means this packet did not cause a wrap. * this packet is thus out of window and should be dropped */ return 1; else /* The packet decrypted with a different high sequence number * to the SA, that means it is the wrap packet and should be * accepted */ return 0; } - fix: don't use the SA as a scratch pad, use the 'packet_data' - the same place that is used as the scratch pad for the low sequence number. other consequences: - An SA doesn't have seq and last_seq, it has only seq; the sequence numnber of the last packet tx'd or rx'd. - there's 64bits of space available on the SA's first cache line. move the AES CTR mode IV there. - test the ESN/AR combinations to catch the bugs this fixes. This doubles the amount of tests, but without AR on they only run for 2 seconds. In the AR tests, the time taken to wait for packets that won't arrive is dropped from 1 to 0.2 seconds thus reducing the runtime of these tests from 10-15 to about 5 sceonds. Signed-off-by: Neale Ranns <neale@graphiant.com> Change-Id: Iaac78905289a272dc01930d70decd8109cf5e7a5
2021-06-28 13:31:28 +00:00
if ar_on:
self.send_and_expect(self.tra_if, pkts, self.tra_if)
else:
self.send_and_assert_no_replies(self.tra_if, pkts)
#
# window positions are different now for AR/non-AR
# move non-AR forward
#
if not ar_on:
# a packet more than 2^30 from the current position and is a
# wrap; (the seq is currently at 0x180000005).
# - AR: accepted
# - non-AR: not considered a wrap, so won't decrypt
pkts = [
(
Ether(src=self.tra_if.remote_mac, dst=self.tra_if.local_mac)
/ p.scapy_tra_sa.encrypt(
IP(src=self.tra_if.remote_ip4, dst=self.tra_if.local_ip4)
/ ICMP(),
seq_num=0x200000005,
)
),
(
Ether(src=self.tra_if.remote_mac, dst=self.tra_if.local_mac)
/ p.scapy_tra_sa.encrypt(
IP(src=self.tra_if.remote_ip4, dst=self.tra_if.local_ip4)
/ ICMP(),
seq_num=0x200000006,
)
),
]
ipsec: Fix setting the hi-sequence number for decrypt Type: fix two problems; 1 - just because anti-reply is not enabled doesn't mean the high sequence number should not be used. - fix, there needs to be some means to detect a wrapped packet, so we use a window size of 2^30. 2 - The SA object was used as a scratch pad for the high-sequence number used during decryption. That means that once the batch has been processed the high-sequence number used is lost. This means it is not possible to distinguish this case: if (seq < IPSEC_SA_ANTI_REPLAY_WINDOW_LOWER_BOUND (tl)) { ... if (post_decrypt) { if (hi_seq_used == sa->seq_hi) /* the high sequence number used to succesfully decrypt this * packet is the same as the last-sequnence number of the SA. * that means this packet did not cause a wrap. * this packet is thus out of window and should be dropped */ return 1; else /* The packet decrypted with a different high sequence number * to the SA, that means it is the wrap packet and should be * accepted */ return 0; } - fix: don't use the SA as a scratch pad, use the 'packet_data' - the same place that is used as the scratch pad for the low sequence number. other consequences: - An SA doesn't have seq and last_seq, it has only seq; the sequence numnber of the last packet tx'd or rx'd. - there's 64bits of space available on the SA's first cache line. move the AES CTR mode IV there. - test the ESN/AR combinations to catch the bugs this fixes. This doubles the amount of tests, but without AR on they only run for 2 seconds. In the AR tests, the time taken to wait for packets that won't arrive is dropped from 1 to 0.2 seconds thus reducing the runtime of these tests from 10-15 to about 5 sceonds. Signed-off-by: Neale Ranns <neale@graphiant.com> Change-Id: Iaac78905289a272dc01930d70decd8109cf5e7a5
2021-06-28 13:31:28 +00:00
self.send_and_expect(self.tra_if, pkts, self.tra_if)
pkts = [
(
Ether(src=self.tra_if.remote_mac, dst=self.tra_if.local_mac)
/ p.scapy_tra_sa.encrypt(
IP(src=self.tra_if.remote_ip4, dst=self.tra_if.local_ip4)
/ ICMP(),
seq_num=0x260000005,
)
)
]
ipsec: Fix setting the hi-sequence number for decrypt Type: fix two problems; 1 - just because anti-reply is not enabled doesn't mean the high sequence number should not be used. - fix, there needs to be some means to detect a wrapped packet, so we use a window size of 2^30. 2 - The SA object was used as a scratch pad for the high-sequence number used during decryption. That means that once the batch has been processed the high-sequence number used is lost. This means it is not possible to distinguish this case: if (seq < IPSEC_SA_ANTI_REPLAY_WINDOW_LOWER_BOUND (tl)) { ... if (post_decrypt) { if (hi_seq_used == sa->seq_hi) /* the high sequence number used to succesfully decrypt this * packet is the same as the last-sequnence number of the SA. * that means this packet did not cause a wrap. * this packet is thus out of window and should be dropped */ return 1; else /* The packet decrypted with a different high sequence number * to the SA, that means it is the wrap packet and should be * accepted */ return 0; } - fix: don't use the SA as a scratch pad, use the 'packet_data' - the same place that is used as the scratch pad for the low sequence number. other consequences: - An SA doesn't have seq and last_seq, it has only seq; the sequence numnber of the last packet tx'd or rx'd. - there's 64bits of space available on the SA's first cache line. move the AES CTR mode IV there. - test the ESN/AR combinations to catch the bugs this fixes. This doubles the amount of tests, but without AR on they only run for 2 seconds. In the AR tests, the time taken to wait for packets that won't arrive is dropped from 1 to 0.2 seconds thus reducing the runtime of these tests from 10-15 to about 5 sceonds. Signed-off-by: Neale Ranns <neale@graphiant.com> Change-Id: Iaac78905289a272dc01930d70decd8109cf5e7a5
2021-06-28 13:31:28 +00:00
self.send_and_expect(self.tra_if, pkts, self.tra_if)
def verify_tra_anti_replay(self):
p = self.params[socket.AF_INET]
esn_en = p.vpp_tra_sa.esn_en
anti_replay_window_size = p.anti_replay_window_size
seq_cycle_node_name = "/err/%s/seq_cycled" % self.tra4_encrypt_node_name
replay_count = self.get_replay_counts(p)
initial_sa_node_replay_diff = replay_count - p.tra_sa_in.get_err("replay")
hash_failed_count = self.get_hash_failed_counts(p)
seq_cycle_count = self.statistics.get_err_counter(seq_cycle_node_name)
initial_sa_node_cycled_diff = seq_cycle_count - p.tra_sa_in.get_err(
"seq_cycled"
)
hash_err = "integ_error"
if ESP == self.encryption_type:
undersize_node_name = "/err/%s/runt" % self.tra4_decrypt_node_name[0]
undersize_count = self.statistics.get_err_counter(undersize_node_name)
initial_sa_node_undersize_diff = undersize_count - p.tra_sa_in.get_err(
"runt"
)
# For AES-GCM an error in the hash is reported as a decryption failure
if p.crypt_algo in ("AES-GCM", "AES-NULL-GMAC"):
hash_err = "decryption_failed"
# In async mode, we don't report errors in the hash.
if p.async_mode:
hash_err = ""
else:
initial_sa_node_hash_diff = hash_failed_count - p.tra_sa_in.get_err(
hash_err
)
#
# send packets with seq numbers 1->34
# this means the window size is still in Case B (see RFC4303
# Appendix A)
#
# for reasons i haven't investigated Scapy won't create a packet with
# seq_num=0
#
pkts = [
(
Ether(src=self.tra_if.remote_mac, dst=self.tra_if.local_mac)
/ p.scapy_tra_sa.encrypt(
IP(src=self.tra_if.remote_ip4, dst=self.tra_if.local_ip4) / ICMP(),
seq_num=seq,
)
)
for seq in range(1, 34)
]
recv_pkts = self.send_and_expect(self.tra_if, pkts, self.tra_if)
# replayed packets are dropped
ipsec: Fix setting the hi-sequence number for decrypt Type: fix two problems; 1 - just because anti-reply is not enabled doesn't mean the high sequence number should not be used. - fix, there needs to be some means to detect a wrapped packet, so we use a window size of 2^30. 2 - The SA object was used as a scratch pad for the high-sequence number used during decryption. That means that once the batch has been processed the high-sequence number used is lost. This means it is not possible to distinguish this case: if (seq < IPSEC_SA_ANTI_REPLAY_WINDOW_LOWER_BOUND (tl)) { ... if (post_decrypt) { if (hi_seq_used == sa->seq_hi) /* the high sequence number used to succesfully decrypt this * packet is the same as the last-sequnence number of the SA. * that means this packet did not cause a wrap. * this packet is thus out of window and should be dropped */ return 1; else /* The packet decrypted with a different high sequence number * to the SA, that means it is the wrap packet and should be * accepted */ return 0; } - fix: don't use the SA as a scratch pad, use the 'packet_data' - the same place that is used as the scratch pad for the low sequence number. other consequences: - An SA doesn't have seq and last_seq, it has only seq; the sequence numnber of the last packet tx'd or rx'd. - there's 64bits of space available on the SA's first cache line. move the AES CTR mode IV there. - test the ESN/AR combinations to catch the bugs this fixes. This doubles the amount of tests, but without AR on they only run for 2 seconds. In the AR tests, the time taken to wait for packets that won't arrive is dropped from 1 to 0.2 seconds thus reducing the runtime of these tests from 10-15 to about 5 sceonds. Signed-off-by: Neale Ranns <neale@graphiant.com> Change-Id: Iaac78905289a272dc01930d70decd8109cf5e7a5
2021-06-28 13:31:28 +00:00
self.send_and_assert_no_replies(self.tra_if, pkts, timeout=0.2)
replay_count += len(pkts)
self.assertEqual(self.get_replay_counts(p), replay_count)
err = p.tra_sa_in.get_err("replay") + initial_sa_node_replay_diff
self.assertEqual(err, replay_count)
#
# now send a batch of packets all with the same sequence number
# the first packet in the batch is legitimate, the rest bogus
#
self.vapi.cli("clear error")
self.vapi.cli("clear node counters")
pkts = Ether(
src=self.tra_if.remote_mac, dst=self.tra_if.local_mac
) / p.scapy_tra_sa.encrypt(
IP(src=self.tra_if.remote_ip4, dst=self.tra_if.local_ip4) / ICMP(),
seq_num=35,
)
recv_pkts = self.send_and_expect(self.tra_if, pkts * 8, self.tra_if, n_rx=1)
replay_count += 7
self.assertEqual(self.get_replay_counts(p), replay_count)
err = p.tra_sa_in.get_err("replay") + initial_sa_node_replay_diff
self.assertEqual(err, replay_count)
#
# now move the window over to anti_replay_window_size + 100 and into Case A
#
self.vapi.cli("clear error")
pkt = Ether(
src=self.tra_if.remote_mac, dst=self.tra_if.local_mac
) / p.scapy_tra_sa.encrypt(
IP(src=self.tra_if.remote_ip4, dst=self.tra_if.local_ip4) / ICMP(),
seq_num=anti_replay_window_size + 100,
)
recv_pkts = self.send_and_expect(self.tra_if, [pkt], self.tra_if)
self.logger.info(self.vapi.ppcli("show ipsec sa 1"))
# replayed packets are dropped
ipsec: Fix setting the hi-sequence number for decrypt Type: fix two problems; 1 - just because anti-reply is not enabled doesn't mean the high sequence number should not be used. - fix, there needs to be some means to detect a wrapped packet, so we use a window size of 2^30. 2 - The SA object was used as a scratch pad for the high-sequence number used during decryption. That means that once the batch has been processed the high-sequence number used is lost. This means it is not possible to distinguish this case: if (seq < IPSEC_SA_ANTI_REPLAY_WINDOW_LOWER_BOUND (tl)) { ... if (post_decrypt) { if (hi_seq_used == sa->seq_hi) /* the high sequence number used to succesfully decrypt this * packet is the same as the last-sequnence number of the SA. * that means this packet did not cause a wrap. * this packet is thus out of window and should be dropped */ return 1; else /* The packet decrypted with a different high sequence number * to the SA, that means it is the wrap packet and should be * accepted */ return 0; } - fix: don't use the SA as a scratch pad, use the 'packet_data' - the same place that is used as the scratch pad for the low sequence number. other consequences: - An SA doesn't have seq and last_seq, it has only seq; the sequence numnber of the last packet tx'd or rx'd. - there's 64bits of space available on the SA's first cache line. move the AES CTR mode IV there. - test the ESN/AR combinations to catch the bugs this fixes. This doubles the amount of tests, but without AR on they only run for 2 seconds. In the AR tests, the time taken to wait for packets that won't arrive is dropped from 1 to 0.2 seconds thus reducing the runtime of these tests from 10-15 to about 5 sceonds. Signed-off-by: Neale Ranns <neale@graphiant.com> Change-Id: Iaac78905289a272dc01930d70decd8109cf5e7a5
2021-06-28 13:31:28 +00:00
self.send_and_assert_no_replies(self.tra_if, pkt * 3, timeout=0.2)
replay_count += 3
self.assertEqual(self.get_replay_counts(p), replay_count)
err = p.tra_sa_in.get_err("replay") + initial_sa_node_replay_diff
self.assertEqual(err, replay_count)
# the window size is anti_replay_window_size packets
# in window are still accepted
pkt = Ether(
src=self.tra_if.remote_mac, dst=self.tra_if.local_mac
) / p.scapy_tra_sa.encrypt(
IP(src=self.tra_if.remote_ip4, dst=self.tra_if.local_ip4) / ICMP(),
seq_num=200,
)
# a packet that does not decrypt does not move the window forward
bogus_sa = SecurityAssociation(
self.encryption_type,
p.scapy_tra_spi,
crypt_algo=p.crypt_algo,
crypt_key=mk_scapy_crypt_key(p)[::-1],
auth_algo=p.auth_algo,
auth_key=p.auth_key[::-1],
)
pkt = Ether(
src=self.tra_if.remote_mac, dst=self.tra_if.local_mac
) / bogus_sa.encrypt(
IP(src=self.tra_if.remote_ip4, dst=self.tra_if.local_ip4) / ICMP(),
seq_num=anti_replay_window_size + 200,
)
ipsec: Fix setting the hi-sequence number for decrypt Type: fix two problems; 1 - just because anti-reply is not enabled doesn't mean the high sequence number should not be used. - fix, there needs to be some means to detect a wrapped packet, so we use a window size of 2^30. 2 - The SA object was used as a scratch pad for the high-sequence number used during decryption. That means that once the batch has been processed the high-sequence number used is lost. This means it is not possible to distinguish this case: if (seq < IPSEC_SA_ANTI_REPLAY_WINDOW_LOWER_BOUND (tl)) { ... if (post_decrypt) { if (hi_seq_used == sa->seq_hi) /* the high sequence number used to succesfully decrypt this * packet is the same as the last-sequnence number of the SA. * that means this packet did not cause a wrap. * this packet is thus out of window and should be dropped */ return 1; else /* The packet decrypted with a different high sequence number * to the SA, that means it is the wrap packet and should be * accepted */ return 0; } - fix: don't use the SA as a scratch pad, use the 'packet_data' - the same place that is used as the scratch pad for the low sequence number. other consequences: - An SA doesn't have seq and last_seq, it has only seq; the sequence numnber of the last packet tx'd or rx'd. - there's 64bits of space available on the SA's first cache line. move the AES CTR mode IV there. - test the ESN/AR combinations to catch the bugs this fixes. This doubles the amount of tests, but without AR on they only run for 2 seconds. In the AR tests, the time taken to wait for packets that won't arrive is dropped from 1 to 0.2 seconds thus reducing the runtime of these tests from 10-15 to about 5 sceonds. Signed-off-by: Neale Ranns <neale@graphiant.com> Change-Id: Iaac78905289a272dc01930d70decd8109cf5e7a5
2021-06-28 13:31:28 +00:00
self.send_and_assert_no_replies(self.tra_if, pkt * 17, timeout=0.2)
hash_failed_count += 17
self.assertEqual(self.get_hash_failed_counts(p), hash_failed_count)
if hash_err != "":
err = p.tra_sa_in.get_err(hash_err) + initial_sa_node_hash_diff
self.assertEqual(err, hash_failed_count)
# a malformed 'runt' packet
# created by a mis-constructed SA
if ESP == self.encryption_type and p.crypt_algo != "NULL":
bogus_sa = SecurityAssociation(self.encryption_type, p.scapy_tra_spi)
pkt = Ether(
src=self.tra_if.remote_mac, dst=self.tra_if.local_mac
) / bogus_sa.encrypt(
IP(src=self.tra_if.remote_ip4, dst=self.tra_if.local_ip4) / ICMP(),
seq_num=anti_replay_window_size + 200,
)
ipsec: Fix setting the hi-sequence number for decrypt Type: fix two problems; 1 - just because anti-reply is not enabled doesn't mean the high sequence number should not be used. - fix, there needs to be some means to detect a wrapped packet, so we use a window size of 2^30. 2 - The SA object was used as a scratch pad for the high-sequence number used during decryption. That means that once the batch has been processed the high-sequence number used is lost. This means it is not possible to distinguish this case: if (seq < IPSEC_SA_ANTI_REPLAY_WINDOW_LOWER_BOUND (tl)) { ... if (post_decrypt) { if (hi_seq_used == sa->seq_hi) /* the high sequence number used to succesfully decrypt this * packet is the same as the last-sequnence number of the SA. * that means this packet did not cause a wrap. * this packet is thus out of window and should be dropped */ return 1; else /* The packet decrypted with a different high sequence number * to the SA, that means it is the wrap packet and should be * accepted */ return 0; } - fix: don't use the SA as a scratch pad, use the 'packet_data' - the same place that is used as the scratch pad for the low sequence number. other consequences: - An SA doesn't have seq and last_seq, it has only seq; the sequence numnber of the last packet tx'd or rx'd. - there's 64bits of space available on the SA's first cache line. move the AES CTR mode IV there. - test the ESN/AR combinations to catch the bugs this fixes. This doubles the amount of tests, but without AR on they only run for 2 seconds. In the AR tests, the time taken to wait for packets that won't arrive is dropped from 1 to 0.2 seconds thus reducing the runtime of these tests from 10-15 to about 5 sceonds. Signed-off-by: Neale Ranns <neale@graphiant.com> Change-Id: Iaac78905289a272dc01930d70decd8109cf5e7a5
2021-06-28 13:31:28 +00:00
self.send_and_assert_no_replies(self.tra_if, pkt * 17, timeout=0.2)
undersize_count += 17
self.assert_error_counter_equal(undersize_node_name, undersize_count)
err = p.tra_sa_in.get_err("runt") + initial_sa_node_undersize_diff
self.assertEqual(err, undersize_count)
# which we can determine since this packet is still in the window
pkt = Ether(
src=self.tra_if.remote_mac, dst=self.tra_if.local_mac
) / p.scapy_tra_sa.encrypt(
IP(src=self.tra_if.remote_ip4, dst=self.tra_if.local_ip4) / ICMP(),
seq_num=234,
)
self.send_and_expect(self.tra_if, [pkt], self.tra_if)
#
# out of window are dropped
# this is Case B. So VPP will consider this to be a high seq num wrap
# and so the decrypt attempt will fail
#
pkt = Ether(
src=self.tra_if.remote_mac, dst=self.tra_if.local_mac
) / p.scapy_tra_sa.encrypt(
IP(src=self.tra_if.remote_ip4, dst=self.tra_if.local_ip4) / ICMP(),
seq_num=17,
)
ipsec: Fix setting the hi-sequence number for decrypt Type: fix two problems; 1 - just because anti-reply is not enabled doesn't mean the high sequence number should not be used. - fix, there needs to be some means to detect a wrapped packet, so we use a window size of 2^30. 2 - The SA object was used as a scratch pad for the high-sequence number used during decryption. That means that once the batch has been processed the high-sequence number used is lost. This means it is not possible to distinguish this case: if (seq < IPSEC_SA_ANTI_REPLAY_WINDOW_LOWER_BOUND (tl)) { ... if (post_decrypt) { if (hi_seq_used == sa->seq_hi) /* the high sequence number used to succesfully decrypt this * packet is the same as the last-sequnence number of the SA. * that means this packet did not cause a wrap. * this packet is thus out of window and should be dropped */ return 1; else /* The packet decrypted with a different high sequence number * to the SA, that means it is the wrap packet and should be * accepted */ return 0; } - fix: don't use the SA as a scratch pad, use the 'packet_data' - the same place that is used as the scratch pad for the low sequence number. other consequences: - An SA doesn't have seq and last_seq, it has only seq; the sequence numnber of the last packet tx'd or rx'd. - there's 64bits of space available on the SA's first cache line. move the AES CTR mode IV there. - test the ESN/AR combinations to catch the bugs this fixes. This doubles the amount of tests, but without AR on they only run for 2 seconds. In the AR tests, the time taken to wait for packets that won't arrive is dropped from 1 to 0.2 seconds thus reducing the runtime of these tests from 10-15 to about 5 sceonds. Signed-off-by: Neale Ranns <neale@graphiant.com> Change-Id: Iaac78905289a272dc01930d70decd8109cf5e7a5
2021-06-28 13:31:28 +00:00
self.send_and_assert_no_replies(self.tra_if, pkt * 17, timeout=0.2)
if esn_en:
# an out of window error with ESN looks like a high sequence
# wrap. but since it isn't then the verify will fail.
hash_failed_count += 17
self.assertEqual(self.get_hash_failed_counts(p), hash_failed_count)
if hash_err != "":
err = p.tra_sa_in.get_err(hash_err) + initial_sa_node_hash_diff
self.assertEqual(err, hash_failed_count)
else:
replay_count += 17
self.assertEqual(self.get_replay_counts(p), replay_count)
err = p.tra_sa_in.get_err("replay") + initial_sa_node_replay_diff
self.assertEqual(err, replay_count)
# valid packet moves the window over to anti_replay_window_size + 258
pkt = Ether(
src=self.tra_if.remote_mac, dst=self.tra_if.local_mac
) / p.scapy_tra_sa.encrypt(
IP(src=self.tra_if.remote_ip4, dst=self.tra_if.local_ip4) / ICMP(),
seq_num=anti_replay_window_size + 258,
)
rx = self.send_and_expect(self.tra_if, [pkt], self.tra_if)
decrypted = p.vpp_tra_sa.decrypt(rx[0][IP])
#
# move VPP's SA TX seq-num to just before the seq-number wrap.
# then fire in a packet that VPP should drop on TX because it
# causes the TX seq number to wrap; unless we're using extened sequence
# numbers.
#
self.vapi.cli("test ipsec sa %d seq 0xffffffff" % p.vpp_tra_sa_id)
self.logger.info(self.vapi.ppcli("show ipsec sa 0"))
self.logger.info(self.vapi.ppcli("show ipsec sa 1"))
pkts = [
(
Ether(src=self.tra_if.remote_mac, dst=self.tra_if.local_mac)
/ p.scapy_tra_sa.encrypt(
IP(src=self.tra_if.remote_ip4, dst=self.tra_if.local_ip4) / ICMP(),
seq_num=seq,
)
)
for seq in range(259, 280)
]
if esn_en:
rxs = self.send_and_expect(self.tra_if, pkts, self.tra_if)
#
# in order for scapy to decrypt its SA's high order number needs
# to wrap
#
p.vpp_tra_sa.seq_num = 0x100000000
for rx in rxs:
decrypted = p.vpp_tra_sa.decrypt(rx[0][IP])
#
# wrap scapy's TX high sequence number. VPP is in case B, so it
# will consider this a high seq wrap also.
# The low seq num we set it to will place VPP's RX window in Case A
#
p.scapy_tra_sa.seq_num = 0x100000005
pkt = Ether(
src=self.tra_if.remote_mac, dst=self.tra_if.local_mac
) / p.scapy_tra_sa.encrypt(
IP(src=self.tra_if.remote_ip4, dst=self.tra_if.local_ip4) / ICMP(),
seq_num=0x100000005,
)
rx = self.send_and_expect(self.tra_if, [pkt], self.tra_if)
ipsec: Fix setting the hi-sequence number for decrypt Type: fix two problems; 1 - just because anti-reply is not enabled doesn't mean the high sequence number should not be used. - fix, there needs to be some means to detect a wrapped packet, so we use a window size of 2^30. 2 - The SA object was used as a scratch pad for the high-sequence number used during decryption. That means that once the batch has been processed the high-sequence number used is lost. This means it is not possible to distinguish this case: if (seq < IPSEC_SA_ANTI_REPLAY_WINDOW_LOWER_BOUND (tl)) { ... if (post_decrypt) { if (hi_seq_used == sa->seq_hi) /* the high sequence number used to succesfully decrypt this * packet is the same as the last-sequnence number of the SA. * that means this packet did not cause a wrap. * this packet is thus out of window and should be dropped */ return 1; else /* The packet decrypted with a different high sequence number * to the SA, that means it is the wrap packet and should be * accepted */ return 0; } - fix: don't use the SA as a scratch pad, use the 'packet_data' - the same place that is used as the scratch pad for the low sequence number. other consequences: - An SA doesn't have seq and last_seq, it has only seq; the sequence numnber of the last packet tx'd or rx'd. - there's 64bits of space available on the SA's first cache line. move the AES CTR mode IV there. - test the ESN/AR combinations to catch the bugs this fixes. This doubles the amount of tests, but without AR on they only run for 2 seconds. In the AR tests, the time taken to wait for packets that won't arrive is dropped from 1 to 0.2 seconds thus reducing the runtime of these tests from 10-15 to about 5 sceonds. Signed-off-by: Neale Ranns <neale@graphiant.com> Change-Id: Iaac78905289a272dc01930d70decd8109cf5e7a5
2021-06-28 13:31:28 +00:00
decrypted = p.vpp_tra_sa.decrypt(rx[0][IP])
#
# A packet that has seq num between (2^32-anti_replay_window_size)+4 and 5 is within
# the window
#
p.scapy_tra_sa.seq_num = 0xFFFFFFFD
pkt = Ether(
src=self.tra_if.remote_mac, dst=self.tra_if.local_mac
) / p.scapy_tra_sa.encrypt(
IP(src=self.tra_if.remote_ip4, dst=self.tra_if.local_ip4) / ICMP(),
seq_num=0xFFFFFFFD,
)
rx = self.send_and_expect(self.tra_if, [pkt], self.tra_if)
decrypted = p.vpp_tra_sa.decrypt(rx[0][IP])
#
# While in case A we cannot wrap the high sequence number again
ipsec: Fix setting the hi-sequence number for decrypt Type: fix two problems; 1 - just because anti-reply is not enabled doesn't mean the high sequence number should not be used. - fix, there needs to be some means to detect a wrapped packet, so we use a window size of 2^30. 2 - The SA object was used as a scratch pad for the high-sequence number used during decryption. That means that once the batch has been processed the high-sequence number used is lost. This means it is not possible to distinguish this case: if (seq < IPSEC_SA_ANTI_REPLAY_WINDOW_LOWER_BOUND (tl)) { ... if (post_decrypt) { if (hi_seq_used == sa->seq_hi) /* the high sequence number used to succesfully decrypt this * packet is the same as the last-sequnence number of the SA. * that means this packet did not cause a wrap. * this packet is thus out of window and should be dropped */ return 1; else /* The packet decrypted with a different high sequence number * to the SA, that means it is the wrap packet and should be * accepted */ return 0; } - fix: don't use the SA as a scratch pad, use the 'packet_data' - the same place that is used as the scratch pad for the low sequence number. other consequences: - An SA doesn't have seq and last_seq, it has only seq; the sequence numnber of the last packet tx'd or rx'd. - there's 64bits of space available on the SA's first cache line. move the AES CTR mode IV there. - test the ESN/AR combinations to catch the bugs this fixes. This doubles the amount of tests, but without AR on they only run for 2 seconds. In the AR tests, the time taken to wait for packets that won't arrive is dropped from 1 to 0.2 seconds thus reducing the runtime of these tests from 10-15 to about 5 sceonds. Signed-off-by: Neale Ranns <neale@graphiant.com> Change-Id: Iaac78905289a272dc01930d70decd8109cf5e7a5
2021-06-28 13:31:28 +00:00
# because VPP will consider this packet to be one that moves the
# window forward
#
pkt = Ether(
src=self.tra_if.remote_mac, dst=self.tra_if.local_mac
) / p.scapy_tra_sa.encrypt(
IP(src=self.tra_if.remote_ip4, dst=self.tra_if.local_ip4) / ICMP(),
seq_num=0x200000999,
)
self.send_and_assert_no_replies(
self.tra_if, [pkt], self.tra_if, timeout=0.2
)
hash_failed_count += 1
self.assertEqual(self.get_hash_failed_counts(p), hash_failed_count)
if hash_err != "":
err = p.tra_sa_in.get_err(hash_err) + initial_sa_node_hash_diff
self.assertEqual(err, hash_failed_count)
#
ipsec: Fix setting the hi-sequence number for decrypt Type: fix two problems; 1 - just because anti-reply is not enabled doesn't mean the high sequence number should not be used. - fix, there needs to be some means to detect a wrapped packet, so we use a window size of 2^30. 2 - The SA object was used as a scratch pad for the high-sequence number used during decryption. That means that once the batch has been processed the high-sequence number used is lost. This means it is not possible to distinguish this case: if (seq < IPSEC_SA_ANTI_REPLAY_WINDOW_LOWER_BOUND (tl)) { ... if (post_decrypt) { if (hi_seq_used == sa->seq_hi) /* the high sequence number used to succesfully decrypt this * packet is the same as the last-sequnence number of the SA. * that means this packet did not cause a wrap. * this packet is thus out of window and should be dropped */ return 1; else /* The packet decrypted with a different high sequence number * to the SA, that means it is the wrap packet and should be * accepted */ return 0; } - fix: don't use the SA as a scratch pad, use the 'packet_data' - the same place that is used as the scratch pad for the low sequence number. other consequences: - An SA doesn't have seq and last_seq, it has only seq; the sequence numnber of the last packet tx'd or rx'd. - there's 64bits of space available on the SA's first cache line. move the AES CTR mode IV there. - test the ESN/AR combinations to catch the bugs this fixes. This doubles the amount of tests, but without AR on they only run for 2 seconds. In the AR tests, the time taken to wait for packets that won't arrive is dropped from 1 to 0.2 seconds thus reducing the runtime of these tests from 10-15 to about 5 sceonds. Signed-off-by: Neale Ranns <neale@graphiant.com> Change-Id: Iaac78905289a272dc01930d70decd8109cf5e7a5
2021-06-28 13:31:28 +00:00
# but if we move the window forward to case B, then we can wrap
# again
#
p.scapy_tra_sa.seq_num = 0x100000000 + anti_replay_window_size + 0x555
pkt = Ether(
src=self.tra_if.remote_mac, dst=self.tra_if.local_mac
) / p.scapy_tra_sa.encrypt(
IP(src=self.tra_if.remote_ip4, dst=self.tra_if.local_ip4) / ICMP(),
seq_num=p.scapy_tra_sa.seq_num,
)
rx = self.send_and_expect(self.tra_if, [pkt], self.tra_if)
decrypted = p.vpp_tra_sa.decrypt(rx[0][IP])
p.scapy_tra_sa.seq_num = 0x200000444
pkt = Ether(
src=self.tra_if.remote_mac, dst=self.tra_if.local_mac
) / p.scapy_tra_sa.encrypt(
IP(src=self.tra_if.remote_ip4, dst=self.tra_if.local_ip4) / ICMP(),
seq_num=0x200000444,
)
rx = self.send_and_expect(self.tra_if, [pkt], self.tra_if)
decrypted = p.vpp_tra_sa.decrypt(rx[0][IP])
else:
#
# without ESN TX sequence numbers can't wrap and packets are
# dropped from here on out.
#
ipsec: Fix setting the hi-sequence number for decrypt Type: fix two problems; 1 - just because anti-reply is not enabled doesn't mean the high sequence number should not be used. - fix, there needs to be some means to detect a wrapped packet, so we use a window size of 2^30. 2 - The SA object was used as a scratch pad for the high-sequence number used during decryption. That means that once the batch has been processed the high-sequence number used is lost. This means it is not possible to distinguish this case: if (seq < IPSEC_SA_ANTI_REPLAY_WINDOW_LOWER_BOUND (tl)) { ... if (post_decrypt) { if (hi_seq_used == sa->seq_hi) /* the high sequence number used to succesfully decrypt this * packet is the same as the last-sequnence number of the SA. * that means this packet did not cause a wrap. * this packet is thus out of window and should be dropped */ return 1; else /* The packet decrypted with a different high sequence number * to the SA, that means it is the wrap packet and should be * accepted */ return 0; } - fix: don't use the SA as a scratch pad, use the 'packet_data' - the same place that is used as the scratch pad for the low sequence number. other consequences: - An SA doesn't have seq and last_seq, it has only seq; the sequence numnber of the last packet tx'd or rx'd. - there's 64bits of space available on the SA's first cache line. move the AES CTR mode IV there. - test the ESN/AR combinations to catch the bugs this fixes. This doubles the amount of tests, but without AR on they only run for 2 seconds. In the AR tests, the time taken to wait for packets that won't arrive is dropped from 1 to 0.2 seconds thus reducing the runtime of these tests from 10-15 to about 5 sceonds. Signed-off-by: Neale Ranns <neale@graphiant.com> Change-Id: Iaac78905289a272dc01930d70decd8109cf5e7a5
2021-06-28 13:31:28 +00:00
self.send_and_assert_no_replies(self.tra_if, pkts, timeout=0.2)
seq_cycle_count += len(pkts)
self.assert_error_counter_equal(seq_cycle_node_name, seq_cycle_count)
err = p.tra_sa_out.get_err("seq_cycled") + initial_sa_node_cycled_diff
self.assertEqual(err, seq_cycle_count)
# move the security-associations seq number on to the last we used
self.vapi.cli("test ipsec sa %d seq 0x15f" % p.scapy_tra_sa_id)
p.scapy_tra_sa.seq_num = 351
p.vpp_tra_sa.seq_num = 351
def verify_tra_lost(self):
p = self.params[socket.AF_INET]
esn_en = p.vpp_tra_sa.esn_en
#
# send packets with seq numbers 1->34
# this means the window size is still in Case B (see RFC4303
# Appendix A)
#
# for reasons i haven't investigated Scapy won't create a packet with
# seq_num=0
#
pkts = [
(
Ether(src=self.tra_if.remote_mac, dst=self.tra_if.local_mac)
/ p.scapy_tra_sa.encrypt(
IP(src=self.tra_if.remote_ip4, dst=self.tra_if.local_ip4) / ICMP(),
seq_num=seq,
)
)
for seq in range(1, 3)
]
self.send_and_expect(self.tra_if, pkts, self.tra_if)
self.assertEqual(p.tra_sa_in.get_err("lost"), 0)
# skip a sequence number
pkts = [
(
Ether(src=self.tra_if.remote_mac, dst=self.tra_if.local_mac)
/ p.scapy_tra_sa.encrypt(
IP(src=self.tra_if.remote_ip4, dst=self.tra_if.local_ip4) / ICMP(),
seq_num=seq,
)
)
for seq in range(4, 6)
]
self.send_and_expect(self.tra_if, pkts, self.tra_if)
self.assertEqual(p.tra_sa_in.get_err("lost"), 0)
# the lost packet are counted untill we get up past the first
# sizeof(replay_window) packets
pkts = [
(
Ether(src=self.tra_if.remote_mac, dst=self.tra_if.local_mac)
/ p.scapy_tra_sa.encrypt(
IP(src=self.tra_if.remote_ip4, dst=self.tra_if.local_ip4) / ICMP(),
seq_num=seq,
)
)
for seq in range(6, 100)
]
self.send_and_expect(self.tra_if, pkts, self.tra_if)
self.assertEqual(p.tra_sa_in.get_err("lost"), 1)
# lost of holes in the sequence
pkts = [
(
Ether(src=self.tra_if.remote_mac, dst=self.tra_if.local_mac)
/ p.scapy_tra_sa.encrypt(
IP(src=self.tra_if.remote_ip4, dst=self.tra_if.local_ip4) / ICMP(),
seq_num=seq,
)
)
for seq in range(100, 200, 2)
]
self.send_and_expect(self.tra_if, pkts, self.tra_if, n_rx=50)
pkts = [
(
Ether(src=self.tra_if.remote_mac, dst=self.tra_if.local_mac)
/ p.scapy_tra_sa.encrypt(
IP(src=self.tra_if.remote_ip4, dst=self.tra_if.local_ip4) / ICMP(),
seq_num=seq,
)
)
for seq in range(200, 300)
]
self.send_and_expect(self.tra_if, pkts, self.tra_if)
self.assertEqual(p.tra_sa_in.get_err("lost"), 51)
# a big hole in the seq number space
pkts = [
(
Ether(src=self.tra_if.remote_mac, dst=self.tra_if.local_mac)
/ p.scapy_tra_sa.encrypt(
IP(src=self.tra_if.remote_ip4, dst=self.tra_if.local_ip4) / ICMP(),
seq_num=seq,
)
)
for seq in range(400, 500)
]
self.send_and_expect(self.tra_if, pkts, self.tra_if)
self.assertEqual(p.tra_sa_in.get_err("lost"), 151)
def verify_tra_basic4(self, count=1, payload_size=54):
"""ipsec v4 transport basic test"""
self.vapi.cli("clear errors")
self.vapi.cli("clear ipsec sa")
try:
p = self.params[socket.AF_INET]
send_pkts = self.gen_encrypt_pkts(
p,
p.scapy_tra_sa,
self.tra_if,
src=self.tra_if.remote_ip4,
dst=self.tra_if.local_ip4,
count=count,
payload_size=payload_size,
)
recv_pkts = self.send_and_expect(self.tra_if, send_pkts, self.tra_if)
for rx in recv_pkts:
self.assertEqual(len(rx) - len(Ether()), rx[IP].len)
self.assert_packet_checksums_valid(rx)
try:
decrypted = p.vpp_tra_sa.decrypt(rx[IP])
self.assert_packet_checksums_valid(decrypted)
except:
self.logger.debug(ppp("Unexpected packet:", rx))
raise
finally:
self.logger.info(self.vapi.ppcli("show error"))
self.logger.info(self.vapi.ppcli("show ipsec all"))
pkts = p.tra_sa_in.get_stats()["packets"]
self.assertEqual(
pkts, count, "incorrect SA in counts: expected %d != %d" % (count, pkts)
)
pkts = p.tra_sa_out.get_stats()["packets"]
self.assertEqual(
pkts, count, "incorrect SA out counts: expected %d != %d" % (count, pkts)
)
self.assertEqual(p.tra_sa_out.get_err("lost"), 0)
self.assertEqual(p.tra_sa_in.get_err("lost"), 0)
self.assert_packet_counter_equal(self.tra4_encrypt_node_name, count)
self.assert_packet_counter_equal(self.tra4_decrypt_node_name[0], count)
def _verify_tra_anti_replay_algorithm_esn(self):
def seq_num(seqh, seql):
return (seqh << 32) | (seql & 0xFFFF_FFFF)
p = self.params[socket.AF_INET]
anti_replay_window_size = p.anti_replay_window_size
seq_cycle_node_name = "/err/%s/seq_cycled" % self.tra4_encrypt_node_name
replay_count = self.get_replay_counts(p)
hash_failed_count = self.get_hash_failed_counts(p)
seq_cycle_count = self.statistics.get_err_counter(seq_cycle_node_name)
if ESP == self.encryption_type:
undersize_node_name = "/err/%s/runt" % self.tra4_decrypt_node_name[0]
undersize_count = self.statistics.get_err_counter(undersize_node_name)
# reset the TX SA to avoid conflict with left configuration
self.vapi.cli(f"test ipsec sa {p.vpp_tra_sa_id} seq 0x0")
"""
RFC 4303 Appendix A2. Case A
|: new Th marker
a-i: possible seq num received
+: Bl, Tl, Bl', Tl'
[BT]l(sign) = [BT]l (sign) 2^32 mod 2^32 (Th inc/dec-remented by 1)
Th - 1 Th Th + 1
--|--a--+---b---+-c--|--d--+---e---+-f--|--g--+---h---+--i-|--
========= ========= =========
Bl- Tl- Bl Tl Bl+ Tl+
Case A implies Tl >= W - 1
"""
Th = 1
Tl = anti_replay_window_size + 40
Bl = Tl - anti_replay_window_size + 1
# move VPP's RX AR window to Case A
self.vapi.cli(f"test ipsec sa {p.scapy_tra_sa_id} seq {seq_num(Th, Tl):#x}")
p.scapy_tra_sa.seq_num = seq_num(Th, Tl)
"""
case a: Seql < Bl
- pre-crypto check: algorithm predicts that the packet wrap the window
-> Seqh = Th + 1
- integrity check: should fail
- post-crypto check: ...
"""
pkts = [
(
Ether(src=self.tra_if.remote_mac, dst=self.tra_if.local_mac)
/ p.scapy_tra_sa.encrypt(
IP(src=self.tra_if.remote_ip4, dst=self.tra_if.local_ip4) / ICMP(),
seq_num=seq,
)
)
for seq in range(seq_num(Th - 1, Bl - 20), seq_num(Th - 1, Bl - 5))
]
# out-of-window packets fail integrity check
self.send_and_assert_no_replies(self.tra_if, pkts, timeout=0.2)
hash_failed_count += len(pkts)
self.assertEqual(self.get_hash_failed_counts(p), hash_failed_count)
"""
case b: Bl <= Seql <= Tl
- pre-crypto check: algorithm predicts that the packet is in the window
-> Seqh = Th
-> check for a replayed packet with Seql
- integrity check: should fail
- post-crypto check: ...
"""
pkts = [
(
Ether(src=self.tra_if.remote_mac, dst=self.tra_if.local_mac)
/ p.scapy_tra_sa.encrypt(
IP(src=self.tra_if.remote_ip4, dst=self.tra_if.local_ip4) / ICMP(),
seq_num=seq,
)
)
for seq in range(seq_num(Th, Tl - 10), seq_num(Th, Tl - 5))
]
self.send_and_expect(self.tra_if, pkts, self.tra_if)
p.scapy_tra_sa.seq_num = seq_num(Th - 1, Tl)
pkts = [
(
Ether(src=self.tra_if.remote_mac, dst=self.tra_if.local_mac)
/ p.scapy_tra_sa.encrypt(
IP(src=self.tra_if.remote_ip4, dst=self.tra_if.local_ip4) / ICMP(),
seq_num=seq,
)
)
for seq in range(seq_num(Th - 1, Tl - 35), seq_num(Th - 1, Tl - 5))
]
self.send_and_assert_no_replies(self.tra_if, pkts, timeout=0.2)
# some packets are rejected by the pre-crypto check
replay_count += 5
self.assertEqual(self.get_replay_counts(p), replay_count)
# out-of-window packets fail integrity check
hash_failed_count += len(pkts) - 5
self.assertEqual(self.get_hash_failed_counts(p), hash_failed_count)
"""
case c: Seql > Tl
- pre-crypto check: algorithm predicts that the packet does not wrap the window
-> Seqh = Th
- integrity check: should fail
- post-crypto check: ...
"""
pkts = [
(
Ether(src=self.tra_if.remote_mac, dst=self.tra_if.local_mac)
/ p.scapy_tra_sa.encrypt(
IP(src=self.tra_if.remote_ip4, dst=self.tra_if.local_ip4) / ICMP(),
seq_num=seq,
)
)
for seq in range(seq_num(Th - 1, Tl + 5), seq_num(Th - 1, Tl + 20))
]
self.send_and_assert_no_replies(self.tra_if, pkts, timeout=0.2)
# out-of-window packets fail integrity check
hash_failed_count += len(pkts)
self.assertEqual(self.get_hash_failed_counts(p), hash_failed_count)
"""
case d: Seql < Bl
- pre-crypto check: algorithm predicts that the packet wrap the window
-> Seqh = Th + 1
- integrity check: should fail
- post-crypto check: ...
"""
p.scapy_tra_sa.seq_num = seq_num(Th, Tl)
pkts = [
(
Ether(src=self.tra_if.remote_mac, dst=self.tra_if.local_mac)
/ p.scapy_tra_sa.encrypt(
IP(src=self.tra_if.remote_ip4, dst=self.tra_if.local_ip4) / ICMP(),
seq_num=seq,
)
)
for seq in range(seq_num(Th, Bl - 20), seq_num(Th, Bl - 5))
]
# out-of-window packets fail integrity check
self.send_and_assert_no_replies(self.tra_if, pkts, timeout=0.2)
hash_failed_count += len(pkts)
self.assertEqual(self.get_hash_failed_counts(p), hash_failed_count)
"""
case e: Bl <= Seql <= Tl
- pre-crypto check: algorithm predicts that the packet is in the window
-> Seqh = Th
-> check for a replayed packet with Seql
- integrity check: should pass
- post-crypto check: should pass
-> Seql is marked in the AR window
"""
pkts = [
(
Ether(src=self.tra_if.remote_mac, dst=self.tra_if.local_mac)
/ p.scapy_tra_sa.encrypt(
IP(src=self.tra_if.remote_ip4, dst=self.tra_if.local_ip4) / ICMP(),
seq_num=seq,
)
)
for seq in range(seq_num(Th, Bl + 10), seq_num(Th, Bl + 30))
]
self.send_and_expect(self.tra_if, pkts, self.tra_if)
"""
case f: Seql > Tl
- pre-crypto check: algorithm predicts that the packet does not wrap the window
-> Seqh = Th
- integrity check: should pass
- post-crypto check: should pass
-> AR window shift (the window stays Case A)
-> Seql is marked in the AR window
"""
pkts = [
(
Ether(src=self.tra_if.remote_mac, dst=self.tra_if.local_mac)
/ p.scapy_tra_sa.encrypt(
IP(src=self.tra_if.remote_ip4, dst=self.tra_if.local_ip4) / ICMP(),
seq_num=seq,
)
)
for seq in range(seq_num(Th, Tl + 50), seq_num(Th, Tl + 60))
]
self.send_and_expect(self.tra_if, pkts, self.tra_if)
"""
case g: Seql < Bl
- pre-crypto check: algorithm predicts that the packet wrap the window
-> Seqh = Th + 1
- integrity check: should pass
- post-crypto check: should pass
-> AR window shift (may set the window in Case B)
-> Seql is marked in the AR window
"""
p.scapy_tra_sa.seq_num = seq_num(Th + 1, Tl)
pkts = [
(
Ether(src=self.tra_if.remote_mac, dst=self.tra_if.local_mac)
/ p.scapy_tra_sa.encrypt(
IP(src=self.tra_if.remote_ip4, dst=self.tra_if.local_ip4) / ICMP(),
seq_num=seq,
)
)
# set the window in Case B (the minimum window size is 64
# so we are sure to overlap)
for seq in range(seq_num(Th + 1, 10), seq_num(Th + 1, 20))
]
self.send_and_expect(self.tra_if, pkts, self.tra_if)
# reset the VPP's RX AR window to Case A
Th = 1
Tl = 2 * anti_replay_window_size + 40
Bl = Tl - anti_replay_window_size + 1
self.vapi.cli(f"test ipsec sa {p.scapy_tra_sa_id} seq {seq_num(Th, Tl):#x}")
p.scapy_tra_sa.seq_num = seq_num(Th + 1, Tl)
pkts = [
(
Ether(src=self.tra_if.remote_mac, dst=self.tra_if.local_mac)
/ p.scapy_tra_sa.encrypt(
IP(src=self.tra_if.remote_ip4, dst=self.tra_if.local_ip4) / ICMP(),
seq_num=seq,
)
)
# the AR will stay in Case A
for seq in range(
seq_num(Th + 1, anti_replay_window_size + 10),
seq_num(Th + 1, anti_replay_window_size + 20),
)
]
self.send_and_expect(self.tra_if, pkts, self.tra_if)
"""
case h: Bl <= Seql <= Tl
- pre-crypto check: algorithm predicts that the packet is in the window
-> Seqh = Th
-> check for a replayed packet with Seql
- integrity check: the wrap is not detected, should fail
- post-crypto check: ...
"""
Th += 1
Tl = anti_replay_window_size + 20
Bl = Tl - anti_replay_window_size + 1
p.scapy_tra_sa.seq_num = seq_num(Th + 1, Tl)
pkts = [
(
Ether(src=self.tra_if.remote_mac, dst=self.tra_if.local_mac)
/ p.scapy_tra_sa.encrypt(
IP(src=self.tra_if.remote_ip4, dst=self.tra_if.local_ip4) / ICMP(),
seq_num=seq,
)
)
for seq in range(seq_num(Th + 1, Tl - 20), seq_num(Th + 1, Tl - 5))
]
self.send_and_assert_no_replies(self.tra_if, pkts, timeout=0.2)
# some packets are rejected by the pre-crypto check
replay_count += 5
self.assertEqual(self.get_replay_counts(p), replay_count)
# out-of-window packets fail integrity check
hash_failed_count += len(pkts) - 5
self.assertEqual(self.get_hash_failed_counts(p), hash_failed_count)
"""
case i: Seql > Tl
- pre-crypto check: algorithm predicts that the packet does not wrap the window
-> Seqh = Th
- integrity check: the wrap is not detected, shoud fail
- post-crypto check: ...
"""
pkts = [
(
Ether(src=self.tra_if.remote_mac, dst=self.tra_if.local_mac)
/ p.scapy_tra_sa.encrypt(
IP(src=self.tra_if.remote_ip4, dst=self.tra_if.local_ip4) / ICMP(),
seq_num=seq,
)
)
for seq in range(seq_num(Th + 1, Tl + 5), seq_num(Th + 1, Tl + 15))
]
# out-of-window packets fail integrity check
self.send_and_assert_no_replies(self.tra_if, pkts, timeout=0.2)
hash_failed_count += len(pkts)
self.assertEqual(self.get_hash_failed_counts(p), hash_failed_count)
"""
RFC 4303 Appendix A2. Case B
Th - 1 Th Th + 1
----|-a-+-----b----+--c--|-d-+----e-----+--f--|-g-+--h---
========= =========== ===========
Tl- Bl Tl Bl+ Tl+
Case B implies Tl < W - 1
"""
# reset the VPP's RX AR window to Case B
Th = 2
Tl = 30 # minimum window size of 64, we are sure to overlap
Bl = (Tl - anti_replay_window_size + 1) % (1 << 32)
self.vapi.cli(f"test ipsec sa {p.scapy_tra_sa_id} seq {seq_num(Th, Tl):#x}")
p.scapy_tra_sa.seq_num = seq_num(Th, Tl)
"""
case a: Seql <= Tl < Bl
- pre-crypto check: algorithm predicts that the packet is in the window
-> Seqh = Th
-> check for replayed packet
- integrity check: should fail
- post-crypto check: ...
"""
pkts = [
(
Ether(src=self.tra_if.remote_mac, dst=self.tra_if.local_mac)
/ p.scapy_tra_sa.encrypt(
IP(src=self.tra_if.remote_ip4, dst=self.tra_if.local_ip4) / ICMP(),
seq_num=seq,
)
)
for seq in range(seq_num(Th, 5), seq_num(Th, 10))
]
self.send_and_expect(self.tra_if, pkts, self.tra_if)
p.scapy_tra_sa.seq_num = seq_num(Th - 1, Tl)
pkts = [
(
Ether(src=self.tra_if.remote_mac, dst=self.tra_if.local_mac)
/ p.scapy_tra_sa.encrypt(
IP(src=self.tra_if.remote_ip4, dst=self.tra_if.local_ip4) / ICMP(),
seq_num=seq,
)
)
for seq in range(seq_num(Th - 1, 0), seq_num(Th - 1, 15))
]
self.send_and_assert_no_replies(self.tra_if, pkts, timeout=0.2)
# some packets are rejected by the pre-crypto check
replay_count += 5
self.assertEqual(self.get_replay_counts(p), replay_count)
# out-of-window packets fail integrity check
hash_failed_count += len(pkts) - 5
self.assertEqual(self.get_hash_failed_counts(p), hash_failed_count)
"""
case b: Tl < Seql < Bl
- pre-crypto check: algorithm predicts that the packet will shift the window
-> Seqh = Th
- integrity check: should fail
- post-crypto check: ...
"""
pkts = [
(
Ether(src=self.tra_if.remote_mac, dst=self.tra_if.local_mac)
/ p.scapy_tra_sa.encrypt(
IP(src=self.tra_if.remote_ip4, dst=self.tra_if.local_ip4) / ICMP(),
seq_num=seq,
)
)
for seq in range(seq_num(Th - 1, Tl + 10), seq_num(Th - 1, Tl + 20))
]
self.send_and_assert_no_replies(self.tra_if, pkts, timeout=0.2)
# out-of-window packets fail integrity check
hash_failed_count += len(pkts)
self.assertEqual(self.get_hash_failed_counts(p), hash_failed_count)
"""
case c: Tl < Bl <= Seql
- pre-crypto check: algorithm predicts that the packet is in the window
-> Seqh = Th - 1
-> check for a replayed packet with Seql
- integrity check: should pass
- post-crypto check: should pass
-> Seql is marked in the AR window
"""
pkts = [
(
Ether(src=self.tra_if.remote_mac, dst=self.tra_if.local_mac)
/ p.scapy_tra_sa.encrypt(
IP(src=self.tra_if.remote_ip4, dst=self.tra_if.local_ip4) / ICMP(),
seq_num=seq,
)
)
for seq in range(seq_num(Th - 1, Bl + 10), seq_num(Th - 1, Bl + 20))
]
self.send_and_expect(self.tra_if, pkts, self.tra_if)
"""
case d: Seql <= Tl < Bl
- pre-crypto check: algorithm predicts that the packet is the window
-> Seqh = Th
-> check for replayed packet
- integrity check: should pass
- post-crypto check: should pass
-> Seql is marked in the AR window
"""
p.scapy_tra_sa.seq_num = seq_num(Th, Tl)
pkts = [
(
Ether(src=self.tra_if.remote_mac, dst=self.tra_if.local_mac)
/ p.scapy_tra_sa.encrypt(
IP(src=self.tra_if.remote_ip4, dst=self.tra_if.local_ip4) / ICMP(),
seq_num=seq,
)
)
for seq in range(seq_num(Th, 15), seq_num(Th, 25))
]
self.send_and_expect(self.tra_if, pkts, self.tra_if)
"""
case e: Tl < Seql < Bl
- pre-crypto check: algorithm predicts that the packet is in the window
-> Seqh = Th
-> check for a replayed packet with Seql
- integrity check: should pass
- post-crypto check: should pass
-> AR window shift (may set the window in Case A)
-> Seql is marked in the AR window
"""
pkts = [
(
Ether(src=self.tra_if.remote_mac, dst=self.tra_if.local_mac)
/ p.scapy_tra_sa.encrypt(
IP(src=self.tra_if.remote_ip4, dst=self.tra_if.local_ip4) / ICMP(),
seq_num=seq,
)
)
for seq in range(seq_num(Th, Tl + 5), seq_num(Th, Tl + 15))
]
# the window stays in Case B
self.send_and_expect(self.tra_if, pkts, self.tra_if)
pkts = [
(
Ether(src=self.tra_if.remote_mac, dst=self.tra_if.local_mac)
/ p.scapy_tra_sa.encrypt(
IP(src=self.tra_if.remote_ip4, dst=self.tra_if.local_ip4) / ICMP(),
seq_num=seq,
)
)
for seq in range(
seq_num(Th, Tl + anti_replay_window_size + 5),
seq_num(Th, Tl + anti_replay_window_size + 15),
)
]
# the window moves to Case A
self.send_and_expect(self.tra_if, pkts, self.tra_if)
# reset the VPP's RX AR window to Case B
Th = 2
Tl = 30 # minimum window size of 64, we are sure to overlap
Bl = (Tl - anti_replay_window_size + 1) % (1 << 32)
self.vapi.cli(f"test ipsec sa {p.scapy_tra_sa_id} seq {seq_num(Th, Tl):#x}")
p.scapy_tra_sa.seq_num = seq_num(Th, Tl)
"""
case f: Tl < Bl <= Seql
- pre-crypto check: algorithm predicts that the packet is in the previous window
-> Seqh = Th - 1
-> check for a replayed packet with Seql
- integrity check: should fail
- post-crypto check: ...
"""
pkts = [
(
Ether(src=self.tra_if.remote_mac, dst=self.tra_if.local_mac)
/ p.scapy_tra_sa.encrypt(
IP(src=self.tra_if.remote_ip4, dst=self.tra_if.local_ip4) / ICMP(),
seq_num=seq,
)
)
for seq in range(seq_num(Th, Bl + 10), seq_num(Th, Bl + 20))
]
self.send_and_assert_no_replies(self.tra_if, pkts, timeout=0.2)
# out-of-window packets fail integrity check
hash_failed_count += len(pkts)
self.assertEqual(self.get_hash_failed_counts(p), hash_failed_count)
"""
case g: Seql <= Tl < Bl
- pre-crypto check: algorithm predicts that the packet is the window
-> Seqh = Th
-> check for replayed packet
- integrity check: should fail
- post-crypto check: ...
"""
pkts = [
(
Ether(src=self.tra_if.remote_mac, dst=self.tra_if.local_mac)
/ p.scapy_tra_sa.encrypt(
IP(src=self.tra_if.remote_ip4, dst=self.tra_if.local_ip4) / ICMP(),
seq_num=seq,
)
)
for seq in range(seq_num(Th, 10), seq_num(Th, 15))
]
self.send_and_expect(self.tra_if, pkts, self.tra_if)
p.scapy_tra_sa.seq_num = seq_num(Th + 1, Tl)
pkts = [
(
Ether(src=self.tra_if.remote_mac, dst=self.tra_if.local_mac)
/ p.scapy_tra_sa.encrypt(
IP(src=self.tra_if.remote_ip4, dst=self.tra_if.local_ip4) / ICMP(),
seq_num=seq,
)
)
for seq in range(seq_num(Th + 1, 0), seq_num(Th + 1, 15))
]
self.send_and_assert_no_replies(self.tra_if, pkts, timeout=0.2)
# some packets are rejected by the pre-crypto check
replay_count += 5
self.assertEqual(self.get_replay_counts(p), replay_count)
# out-of-window packets fail integrity check
hash_failed_count += len(pkts) - 5
self.assertEqual(self.get_hash_failed_counts(p), hash_failed_count)
"""
case h: Tl < Seql < Bl
- pre-crypto check: algorithm predicts that the packet will shift the window
-> Seqh = Th
- integrity check: should fail
- post-crypto check: ...
"""
pkts = [
(
Ether(src=self.tra_if.remote_mac, dst=self.tra_if.local_mac)
/ p.scapy_tra_sa.encrypt(
IP(src=self.tra_if.remote_ip4, dst=self.tra_if.local_ip4) / ICMP(),
seq_num=seq,
)
)
for seq in range(seq_num(Th + 1, Tl + 10), seq_num(Th + 1, Tl + 20))
]
self.send_and_assert_no_replies(self.tra_if, pkts, timeout=0.2)
# out-of-window packets fail integrity check
hash_failed_count += len(pkts)
self.assertEqual(self.get_hash_failed_counts(p), hash_failed_count)
def _verify_tra_anti_replay_algorithm_no_esn(self):
def seq_num(seql):
return seql & 0xFFFF_FFFF
p = self.params[socket.AF_INET]
anti_replay_window_size = p.anti_replay_window_size
seq_cycle_node_name = "/err/%s/seq_cycled" % self.tra4_encrypt_node_name
replay_count = self.get_replay_counts(p)
hash_failed_count = self.get_hash_failed_counts(p)
seq_cycle_count = self.statistics.get_err_counter(seq_cycle_node_name)
if ESP == self.encryption_type:
undersize_node_name = "/err/%s/runt" % self.tra4_decrypt_node_name[0]
undersize_count = self.statistics.get_err_counter(undersize_node_name)
# reset the TX SA to avoid conflict with left configuration
self.vapi.cli(f"test ipsec sa {p.vpp_tra_sa_id} seq 0x0")
"""
RFC 4303 Appendix A2. Case A
a-c: possible seq num received
+: Bl, Tl
|--a--+---b---+-c--|
=========
Bl Tl
No ESN implies Th = 0
Case A implies Tl >= W - 1
"""
Tl = anti_replay_window_size + 40
Bl = Tl - anti_replay_window_size + 1
# move VPP's RX AR window to Case A
self.vapi.cli(f"test ipsec sa {p.scapy_tra_sa_id} seq {seq_num(Tl):#x}")
p.scapy_tra_sa.seq_num = seq_num(Tl)
"""
case a: Seql < Bl
- pre-crypto check: algorithm predicts that the packet is out of window
-> packet should be dropped
- integrity check: ...
- post-crypto check: ...
"""
pkts = [
(
Ether(src=self.tra_if.remote_mac, dst=self.tra_if.local_mac)
/ p.scapy_tra_sa.encrypt(
IP(src=self.tra_if.remote_ip4, dst=self.tra_if.local_ip4) / ICMP(),
seq_num=seq,
)
)
for seq in range(seq_num(Bl - 20), seq_num(Bl - 5))
]
# out-of-window packets
self.send_and_assert_no_replies(self.tra_if, pkts, timeout=0.2)
replay_count += len(pkts)
self.assertEqual(self.get_replay_counts(p), replay_count)
"""
case b: Bl <= Seql <= Tl
- pre-crypto check: algorithm predicts that the packet is in the window
-> check for a replayed packet with Seql
- integrity check: should pass
- post-crypto check:
-> check for a replayed packet with Seql
"""
pkts = [
(
Ether(src=self.tra_if.remote_mac, dst=self.tra_if.local_mac)
/ p.scapy_tra_sa.encrypt(
IP(src=self.tra_if.remote_ip4, dst=self.tra_if.local_ip4) / ICMP(),
seq_num=seq,
)
)
for seq in range(seq_num(Tl - 50), seq_num(Tl - 30))
]
self.send_and_expect(self.tra_if, pkts, self.tra_if)
pkts = [
(
Ether(src=self.tra_if.remote_mac, dst=self.tra_if.local_mac)
/ p.scapy_tra_sa.encrypt(
IP(src=self.tra_if.remote_ip4, dst=self.tra_if.local_ip4) / ICMP(),
seq_num=seq,
)
)
for seq in range(seq_num(Tl - 35), seq_num(Tl - 30))
]
self.send_and_assert_no_replies(self.tra_if, pkts, timeout=0.2)
# replayed packets
replay_count += 5
self.assertEqual(self.get_replay_counts(p), replay_count)
"""
case c: Seql > Tl
- pre-crypto check: algorithm predicts that the packet will shift the window
- integrity check: should pass
- post-crypto check: should pass
-> AR window is shifted
"""
pkts = [
(
Ether(src=self.tra_if.remote_mac, dst=self.tra_if.local_mac)
/ p.scapy_tra_sa.encrypt(
IP(src=self.tra_if.remote_ip4, dst=self.tra_if.local_ip4) / ICMP(),
seq_num=seq,
)
)
for seq in range(seq_num(Tl + 5), seq_num(Tl + 20))
]
self.send_and_expect(self.tra_if, pkts, self.tra_if)
"""
RFC 4303 Appendix A2. Case B
|-a-----+------b-----|
=========
Tl
Case B implies Tl < W - 1
"""
# reset the VPP's RX AR window to Case B
Tl = 30 # minimum window size of 64, we are sure to overlap
Bl = seq_num(Tl - anti_replay_window_size + 1)
self.vapi.cli(f"test ipsec sa {p.scapy_tra_sa_id} seq {seq_num(Tl):#x}")
"""
case a: Seql <= Tl < Bl
- pre-crypto check: algorithm predicts that the packet is in the window
-> check for replayed packet
- integrity check: should fail
- post-crypto check: ...
"""
pkts = [
(
Ether(src=self.tra_if.remote_mac, dst=self.tra_if.local_mac)
/ p.scapy_tra_sa.encrypt(
IP(src=self.tra_if.remote_ip4, dst=self.tra_if.local_ip4) / ICMP(),
seq_num=seq,
)
)
for seq in range(seq_num(5), seq_num(10))
]
self.send_and_expect(self.tra_if, pkts, self.tra_if)
"""
case b: Tl < Seql < Bl
- pre-crypto check: algorithm predicts that the packet will shift the window
- integrity check: should pass
- post-crypto check: should pass
-> AR window is shifted
"""
pkts = [
(
Ether(src=self.tra_if.remote_mac, dst=self.tra_if.local_mac)
/ p.scapy_tra_sa.encrypt(
IP(src=self.tra_if.remote_ip4, dst=self.tra_if.local_ip4) / ICMP(),
seq_num=seq,
)
)
for seq in range(seq_num(-50), seq_num(-20))
]
self.send_and_expect(self.tra_if, pkts, self.tra_if)
def verify_tra_anti_replay_algorithm(self):
if self.params[socket.AF_INET].vpp_tra_sa.esn_en:
self._verify_tra_anti_replay_algorithm_esn()
else:
self._verify_tra_anti_replay_algorithm_no_esn()
@unittest.skipIf(
"ping" in config.excluded_plugins, "Exclude tests requiring Ping plugin"
)
class IpsecTra4Tests(IpsecTra4):
"""UT test methods for Transport v4"""
def test_tra_anti_replay(self):
"""ipsec v4 transport anti-replay test"""
self.verify_tra_anti_replay()
def test_tra_anti_replay_algorithm(self):
"""ipsec v4 transport anti-replay algorithm test"""
self.verify_tra_anti_replay_algorithm()
def test_tra_lost(self):
"""ipsec v4 transport lost packet test"""
self.verify_tra_lost()
def test_tra_basic(self, count=1):
"""ipsec v4 transport basic test"""
self.verify_tra_basic4(count=1)
def test_tra_burst(self):
"""ipsec v4 transport burst test"""
self.verify_tra_basic4(count=257)
class IpsecTra6(object):
"""verify methods for Transport v6"""
def verify_tra_basic6(self, count=1, payload_size=54):
self.vapi.cli("clear errors")
self.vapi.cli("clear ipsec sa")
try:
p = self.params[socket.AF_INET6]
send_pkts = self.gen_encrypt_pkts6(
p,
p.scapy_tra_sa,
self.tra_if,
src=self.tra_if.remote_ip6,
dst=self.tra_if.local_ip6,
count=count,
payload_size=payload_size,
)
recv_pkts = self.send_and_expect(self.tra_if, send_pkts, self.tra_if)
for rx in recv_pkts:
self.assertEqual(len(rx) - len(Ether()) - len(IPv6()), rx[IPv6].plen)
try:
decrypted = p.vpp_tra_sa.decrypt(rx[IPv6])
self.assert_packet_checksums_valid(decrypted)
except:
self.logger.debug(ppp("Unexpected packet:", rx))
raise
finally:
self.logger.info(self.vapi.ppcli("show error"))
self.logger.info(self.vapi.ppcli("show ipsec all"))
pkts = p.tra_sa_in.get_stats()["packets"]
self.assertEqual(
pkts, count, "incorrect SA in counts: expected %d != %d" % (count, pkts)
)
pkts = p.tra_sa_out.get_stats()["packets"]
self.assertEqual(
pkts, count, "incorrect SA out counts: expected %d != %d" % (count, pkts)
)
self.assert_packet_counter_equal(self.tra6_encrypt_node_name, count)
self.assert_packet_counter_equal(self.tra6_decrypt_node_name[0], count)
def gen_encrypt_pkts_ext_hdrs6(
self, sa, sw_intf, src, dst, count=1, payload_size=54
):
return [
Ether(src=sw_intf.remote_mac, dst=sw_intf.local_mac)
/ sa.encrypt(
IPv6(src=src, dst=dst)
/ ICMPv6EchoRequest(id=0, seq=1, data="X" * payload_size)
)
for i in range(count)
]
def gen_pkts_ext_hdrs6(self, sw_intf, src, dst, count=1, payload_size=54):
return [
Ether(src=sw_intf.remote_mac, dst=sw_intf.local_mac)
/ IPv6(src=src, dst=dst)
/ IPv6ExtHdrHopByHop()
/ IPv6ExtHdrFragment(id=2, offset=0)
/ Raw(b"\xff" * 200)
for i in range(count)
]
def verify_tra_encrypted6(self, p, sa, rxs):
decrypted = []
for rx in rxs:
self.assert_packet_checksums_valid(rx)
try:
decrypt_pkt = p.vpp_tra_sa.decrypt(rx[IPv6])
decrypted.append(decrypt_pkt)
self.assert_equal(decrypt_pkt.src, self.tra_if.local_ip6)
self.assert_equal(decrypt_pkt.dst, self.tra_if.remote_ip6)
except:
self.logger.debug(ppp("Unexpected packet:", rx))
try:
self.logger.debug(ppp("Decrypted packet:", decrypt_pkt))
except:
pass
raise
return decrypted
def verify_tra_66_ext_hdrs(self, p):
count = 63
#
# check we can decrypt with options
#
tx = self.gen_encrypt_pkts_ext_hdrs6(
p.scapy_tra_sa,
self.tra_if,
src=self.tra_if.remote_ip6,
dst=self.tra_if.local_ip6,
count=count,
)
self.send_and_expect(self.tra_if, tx, self.tra_if)
#
# injecting a packet from ourselves to be routed of box is a hack
# but it matches an outbout policy, alors je ne regrette rien
#
# one extension before ESP
tx = (
Ether(src=self.pg2.remote_mac, dst=self.pg2.local_mac)
/ IPv6(src=self.tra_if.local_ip6, dst=self.tra_if.remote_ip6)
/ IPv6ExtHdrFragment(id=2, offset=0)
/ Raw(b"\xff" * 200)
)
rxs = self.send_and_expect(self.pg2, [tx], self.tra_if)
dcs = self.verify_tra_encrypted6(p, p.vpp_tra_sa, rxs)
for dc in dcs:
# for reasons i'm not going to investigate scapy does not
# created the correct headers after decrypt. but reparsing
# the ipv6 packet fixes it
dc = IPv6(raw(dc[IPv6]))
self.assert_equal(dc[IPv6ExtHdrFragment].id, 2)
# two extensions before ESP
tx = (
Ether(src=self.pg2.remote_mac, dst=self.pg2.local_mac)
/ IPv6(src=self.tra_if.local_ip6, dst=self.tra_if.remote_ip6)
/ IPv6ExtHdrHopByHop()
/ IPv6ExtHdrFragment(id=2, offset=0)
/ Raw(b"\xff" * 200)
)
rxs = self.send_and_expect(self.pg2, [tx], self.tra_if)
dcs = self.verify_tra_encrypted6(p, p.vpp_tra_sa, rxs)
for dc in dcs:
dc = IPv6(raw(dc[IPv6]))
self.assertTrue(dc[IPv6ExtHdrHopByHop])
self.assert_equal(dc[IPv6ExtHdrFragment].id, 2)
# two extensions before ESP, one after
tx = (
Ether(src=self.pg2.remote_mac, dst=self.pg2.local_mac)
/ IPv6(src=self.tra_if.local_ip6, dst=self.tra_if.remote_ip6)
/ IPv6ExtHdrHopByHop()
/ IPv6ExtHdrFragment(id=2, offset=0)
/ IPv6ExtHdrDestOpt()
/ Raw(b"\xff" * 200)
)
rxs = self.send_and_expect(self.pg2, [tx], self.tra_if)
dcs = self.verify_tra_encrypted6(p, p.vpp_tra_sa, rxs)
for dc in dcs:
dc = IPv6(raw(dc[IPv6]))
self.assertTrue(dc[IPv6ExtHdrDestOpt])
self.assertTrue(dc[IPv6ExtHdrHopByHop])
self.assert_equal(dc[IPv6ExtHdrFragment].id, 2)
@unittest.skipIf(
"ping" in config.excluded_plugins, "Exclude tests requiring Ping plugin"
)
class IpsecTra6Tests(IpsecTra6):
"""UT test methods for Transport v6"""
def test_tra_basic6(self):
"""ipsec v6 transport basic test"""
self.verify_tra_basic6(count=1)
def test_tra_burst6(self):
"""ipsec v6 transport burst test"""
self.verify_tra_basic6(count=257)
class IpsecTra6ExtTests(IpsecTra6):
def test_tra_ext_hdrs_66(self):
"""ipsec 6o6 tra extension headers test"""
self.verify_tra_66_ext_hdrs(self.params[socket.AF_INET6])
class IpsecTra46Tests(IpsecTra4Tests, IpsecTra6Tests):
"""UT test methods for Transport v6 and v4"""
pass
class IpsecTun4(object):
"""verify methods for Tunnel v4"""
def verify_counters4(self, p, count, n_frags=None, worker=None):
if not n_frags:
n_frags = count
if hasattr(p, "spd_policy_in_any"):
pkts = p.spd_policy_in_any.get_stats(worker)["packets"]
self.assertEqual(
pkts,
count,
"incorrect SPD any policy: expected %d != %d" % (count, pkts),
)
if hasattr(p, "tun_sa_in"):
pkts = p.tun_sa_in.get_stats(worker)["packets"]
self.assertEqual(
pkts, count, "incorrect SA in counts: expected %d != %d" % (count, pkts)
)
pkts = p.tun_sa_out.get_stats(worker)["packets"]
self.assertEqual(
pkts,
n_frags,
"incorrect SA out counts: expected %d != %d" % (count, pkts),
)
self.assert_packet_counter_equal(self.tun4_encrypt_node_name, n_frags)
self.assert_packet_counter_equal(self.tun4_decrypt_node_name[0], count)
def verify_decrypted(self, p, rxs):
for rx in rxs:
self.assert_equal(rx[IP].src, p.remote_tun_if_host)
self.assert_equal(rx[IP].dst, self.pg1.remote_ip4)
self.assert_packet_checksums_valid(rx)
ipsec: fix padding/alignment for native IPsec encryption Not all ESP crypto algorithms require padding/alignment to be the same as AES block/IV size. CCM, CTR and GCM all have no padding/alignment requirements, and the RFCs indicate that no padding (beyond ESPs 4 octet alignment requirement) should be used unless TFC (traffic flow confidentiality) has been requested. CTR: https://tools.ietf.org/html/rfc3686#section-3.2 GCM: https://tools.ietf.org/html/rfc4106#section-3.2 CCM: https://tools.ietf.org/html/rfc4309#section-3.2 - VPP is incorrectly using the IV/AES block size to pad CTR and GCM. These modes do not require padding (beyond ESPs 4 octet requirement), as a result packets will have unnecessary padding, which will waste bandwidth at least and possibly fail certain network configurations that have finely tuned MTU configurations at worst. Fix this as well as changing the field names from ".*block_size" to ".*block_align" to better represent their actual (and only) use. Rename "block_sz" in esp_encrypt to "esp_align" and set it correctly as well. test: ipsec: Add unit-test to test for RFC correct padding/alignment test: patch scapy to not incorrectly pad ccm, ctr, gcm modes as well - Scapy is also incorrectly using the AES block size of 16 to pad CCM, CTR, and GCM cipher modes. A bug report has been opened with the and acknowledged with the upstream scapy project as well: https://github.com/secdev/scapy/issues/2322 Ticket: VPP-1928 Type: fix Signed-off-by: Christian Hopps <chopps@labn.net> Change-Id: Iaa4d6a325a2e99fdcb2c375a3395bcfe7947770e
2019-11-03 07:02:15 -05:00
def verify_esp_padding(self, sa, esp_payload, decrypt_pkt):
align = sa.crypt_algo.block_size
if align < 4:
align = 4
exp_len = (len(decrypt_pkt) + 2 + (align - 1)) & ~(align - 1)
exp_len += sa.crypt_algo.iv_size
exp_len += sa.crypt_algo.icv_size or sa.auth_algo.icv_size
self.assertEqual(exp_len, len(esp_payload))
def verify_encrypted(self, p, sa, rxs):
decrypt_pkts = []
for rx in rxs:
if p.nat_header:
self.assertEqual(rx[UDP].dport, p.nat_header.dport)
self.assert_packet_checksums_valid(rx)
self.assertEqual(len(rx) - len(Ether()), rx[IP].len)
try:
ipsec: fix padding/alignment for native IPsec encryption Not all ESP crypto algorithms require padding/alignment to be the same as AES block/IV size. CCM, CTR and GCM all have no padding/alignment requirements, and the RFCs indicate that no padding (beyond ESPs 4 octet alignment requirement) should be used unless TFC (traffic flow confidentiality) has been requested. CTR: https://tools.ietf.org/html/rfc3686#section-3.2 GCM: https://tools.ietf.org/html/rfc4106#section-3.2 CCM: https://tools.ietf.org/html/rfc4309#section-3.2 - VPP is incorrectly using the IV/AES block size to pad CTR and GCM. These modes do not require padding (beyond ESPs 4 octet requirement), as a result packets will have unnecessary padding, which will waste bandwidth at least and possibly fail certain network configurations that have finely tuned MTU configurations at worst. Fix this as well as changing the field names from ".*block_size" to ".*block_align" to better represent their actual (and only) use. Rename "block_sz" in esp_encrypt to "esp_align" and set it correctly as well. test: ipsec: Add unit-test to test for RFC correct padding/alignment test: patch scapy to not incorrectly pad ccm, ctr, gcm modes as well - Scapy is also incorrectly using the AES block size of 16 to pad CCM, CTR, and GCM cipher modes. A bug report has been opened with the and acknowledged with the upstream scapy project as well: https://github.com/secdev/scapy/issues/2322 Ticket: VPP-1928 Type: fix Signed-off-by: Christian Hopps <chopps@labn.net> Change-Id: Iaa4d6a325a2e99fdcb2c375a3395bcfe7947770e
2019-11-03 07:02:15 -05:00
rx_ip = rx[IP]
decrypt_pkt = p.vpp_tun_sa.decrypt(rx_ip)
if not decrypt_pkt.haslayer(IP):
decrypt_pkt = IP(decrypt_pkt[Raw].load)
ipsec: fix padding/alignment for native IPsec encryption Not all ESP crypto algorithms require padding/alignment to be the same as AES block/IV size. CCM, CTR and GCM all have no padding/alignment requirements, and the RFCs indicate that no padding (beyond ESPs 4 octet alignment requirement) should be used unless TFC (traffic flow confidentiality) has been requested. CTR: https://tools.ietf.org/html/rfc3686#section-3.2 GCM: https://tools.ietf.org/html/rfc4106#section-3.2 CCM: https://tools.ietf.org/html/rfc4309#section-3.2 - VPP is incorrectly using the IV/AES block size to pad CTR and GCM. These modes do not require padding (beyond ESPs 4 octet requirement), as a result packets will have unnecessary padding, which will waste bandwidth at least and possibly fail certain network configurations that have finely tuned MTU configurations at worst. Fix this as well as changing the field names from ".*block_size" to ".*block_align" to better represent their actual (and only) use. Rename "block_sz" in esp_encrypt to "esp_align" and set it correctly as well. test: ipsec: Add unit-test to test for RFC correct padding/alignment test: patch scapy to not incorrectly pad ccm, ctr, gcm modes as well - Scapy is also incorrectly using the AES block size of 16 to pad CCM, CTR, and GCM cipher modes. A bug report has been opened with the and acknowledged with the upstream scapy project as well: https://github.com/secdev/scapy/issues/2322 Ticket: VPP-1928 Type: fix Signed-off-by: Christian Hopps <chopps@labn.net> Change-Id: Iaa4d6a325a2e99fdcb2c375a3395bcfe7947770e
2019-11-03 07:02:15 -05:00
if rx_ip.proto == socket.IPPROTO_ESP:
self.verify_esp_padding(sa, rx_ip[ESP].data, decrypt_pkt)
decrypt_pkts.append(decrypt_pkt)
self.assert_equal(decrypt_pkt.src, self.pg1.remote_ip4)
self.assert_equal(decrypt_pkt.dst, p.remote_tun_if_host)
except:
self.logger.debug(ppp("Unexpected packet:", rx))
try:
self.logger.debug(ppp("Decrypted packet:", decrypt_pkt))
except:
pass
raise
pkts = reassemble4(decrypt_pkts)
for pkt in pkts:
self.assert_packet_checksums_valid(pkt)
def verify_tun_44(self, p, count=1, payload_size=64, n_rx=None):
self.vapi.cli("clear errors")
self.vapi.cli("clear ipsec counters")
self.vapi.cli("clear ipsec sa")
if not n_rx:
n_rx = count
try:
send_pkts = self.gen_encrypt_pkts(
p,
p.scapy_tun_sa,
self.tun_if,
src=p.remote_tun_if_host,
dst=self.pg1.remote_ip4,
count=count,
payload_size=payload_size,
)
recv_pkts = self.send_and_expect(self.tun_if, send_pkts, self.pg1)
self.verify_decrypted(p, recv_pkts)
send_pkts = self.gen_pkts(
self.pg1,
src=self.pg1.remote_ip4,
dst=p.remote_tun_if_host,
count=count,
payload_size=payload_size,
)
recv_pkts = self.send_and_expect(self.pg1, send_pkts, self.tun_if, n_rx)
self.verify_encrypted(p, p.vpp_tun_sa, recv_pkts)
for rx in recv_pkts:
self.assertEqual(rx[IP].src, p.tun_src)
self.assertEqual(rx[IP].dst, p.tun_dst)
finally:
self.logger.info(self.vapi.ppcli("show error"))
self.logger.info(self.vapi.ppcli("show ipsec all"))
self.logger.info(self.vapi.ppcli("show ipsec sa 0"))
self.logger.info(self.vapi.ppcli("show ipsec sa 4"))
self.verify_counters4(p, count, n_rx)
def verify_tun_dropped_44(self, p, count=1, payload_size=64, n_rx=None):
self.vapi.cli("clear errors")
if not n_rx:
n_rx = count
try:
send_pkts = self.gen_encrypt_pkts(
p,
p.scapy_tun_sa,
self.tun_if,
src=p.remote_tun_if_host,
dst=self.pg1.remote_ip4,
count=count,
)
self.send_and_assert_no_replies(self.tun_if, send_pkts)
send_pkts = self.gen_pkts(
self.pg1,
src=self.pg1.remote_ip4,
dst=p.remote_tun_if_host,
count=count,
payload_size=payload_size,
)
self.send_and_assert_no_replies(self.pg1, send_pkts)
finally:
self.logger.info(self.vapi.ppcli("show error"))
self.logger.info(self.vapi.ppcli("show ipsec all"))
def verify_tun_reass_44(self, p):
self.vapi.cli("clear errors")
self.vapi.ip_reassembly_enable_disable(
sw_if_index=self.tun_if.sw_if_index, enable_ip4=True
)
try:
send_pkts = self.gen_encrypt_pkts(
p,
p.scapy_tun_sa,
self.tun_if,
src=p.remote_tun_if_host,
dst=self.pg1.remote_ip4,
payload_size=1900,
count=1,
)
send_pkts = fragment_rfc791(send_pkts[0], 1400)
recv_pkts = self.send_and_expect(self.tun_if, send_pkts, self.pg1, n_rx=1)
self.verify_decrypted(p, recv_pkts)
send_pkts = self.gen_pkts(
self.pg1, src=self.pg1.remote_ip4, dst=p.remote_tun_if_host, count=1
)
recv_pkts = self.send_and_expect(self.pg1, send_pkts, self.tun_if)
self.verify_encrypted(p, p.vpp_tun_sa, recv_pkts)
finally:
self.logger.info(self.vapi.ppcli("show error"))
self.logger.info(self.vapi.ppcli("show ipsec all"))
self.verify_counters4(p, 1, 1)
self.vapi.ip_reassembly_enable_disable(
sw_if_index=self.tun_if.sw_if_index, enable_ip4=False
)
def verify_tun_64(self, p, count=1):
self.vapi.cli("clear errors")
self.vapi.cli("clear ipsec sa")
try:
send_pkts = self.gen_encrypt_pkts6(
p,
p.scapy_tun_sa,
self.tun_if,
src=p.remote_tun_if_host6,
dst=self.pg1.remote_ip6,
count=count,
)
recv_pkts = self.send_and_expect(self.tun_if, send_pkts, self.pg1)
for recv_pkt in recv_pkts:
self.assert_equal(recv_pkt[IPv6].src, p.remote_tun_if_host6)
self.assert_equal(recv_pkt[IPv6].dst, self.pg1.remote_ip6)
self.assert_packet_checksums_valid(recv_pkt)
send_pkts = self.gen_pkts6(
p,
self.pg1,
src=self.pg1.remote_ip6,
dst=p.remote_tun_if_host6,
count=count,
)
recv_pkts = self.send_and_expect(self.pg1, send_pkts, self.tun_if)
for recv_pkt in recv_pkts:
try:
decrypt_pkt = p.vpp_tun_sa.decrypt(recv_pkt[IP])
if not decrypt_pkt.haslayer(IPv6):
decrypt_pkt = IPv6(decrypt_pkt[Raw].load)
self.assert_equal(decrypt_pkt.src, self.pg1.remote_ip6)
self.assert_equal(decrypt_pkt.dst, p.remote_tun_if_host6)
self.assert_packet_checksums_valid(decrypt_pkt)
except:
self.logger.error(ppp("Unexpected packet:", recv_pkt))
try:
self.logger.debug(ppp("Decrypted packet:", decrypt_pkt))
except:
pass
raise
finally:
self.logger.info(self.vapi.ppcli("show error"))
self.logger.info(self.vapi.ppcli("show ipsec all"))
self.verify_counters4(p, count)
def verify_keepalive(self, p):
# the sizeof Raw is calculated to pad to the minimum ehternet
# frame size of 64 btyes
pkt = (
Ether(src=self.tun_if.remote_mac, dst=self.tun_if.local_mac)
/ IP(src=p.remote_tun_if_host, dst=self.tun_if.local_ip4)
/ UDP(sport=333, dport=4500)
/ Raw(b"\xff")
/ Padding(0 * 21)
)
self.send_and_assert_no_replies(self.tun_if, pkt * 31)
self.assert_error_counter_equal(
"/err/%s/nat_keepalive" % self.tun4_input_node, 31
)
pkt = (
Ether(src=self.tun_if.remote_mac, dst=self.tun_if.local_mac)
/ IP(src=p.remote_tun_if_host, dst=self.tun_if.local_ip4)
/ UDP(sport=333, dport=4500)
/ Raw(b"\xfe")
)
self.send_and_assert_no_replies(self.tun_if, pkt * 31)
self.assert_error_counter_equal("/err/%s/too_short" % self.tun4_input_node, 31)
pkt = (
Ether(src=self.tun_if.remote_mac, dst=self.tun_if.local_mac)
/ IP(src=p.remote_tun_if_host, dst=self.tun_if.local_ip4)
/ UDP(sport=333, dport=4500)
/ Raw(b"\xfe")
/ Padding(0 * 21)
)
self.send_and_assert_no_replies(self.tun_if, pkt * 31)
self.assert_error_counter_equal("/err/%s/too_short" % self.tun4_input_node, 62)
class IpsecTun4Tests(IpsecTun4):
"""UT test methods for Tunnel v4"""
def test_tun_basic44(self):
"""ipsec 4o4 tunnel basic test"""
self.verify_tun_44(self.params[socket.AF_INET], count=1)
self.tun_if.admin_down()
self.tun_if.resolve_arp()
self.tun_if.admin_up()
self.verify_tun_44(self.params[socket.AF_INET], count=1)
def test_tun_reass_basic44(self):
"""ipsec 4o4 tunnel basic reassembly test"""
self.verify_tun_reass_44(self.params[socket.AF_INET])
def test_tun_burst44(self):
"""ipsec 4o4 tunnel burst test"""
self.verify_tun_44(self.params[socket.AF_INET], count=127)
class IpsecTun6(object):
"""verify methods for Tunnel v6"""
def verify_counters6(self, p_in, p_out, count, worker=None):
if hasattr(p_in, "tun_sa_in"):
pkts = p_in.tun_sa_in.get_stats(worker)["packets"]
self.assertEqual(
pkts, count, "incorrect SA in counts: expected %d != %d" % (count, pkts)
)
if hasattr(p_out, "tun_sa_out"):
pkts = p_out.tun_sa_out.get_stats(worker)["packets"]
self.assertEqual(
pkts,
count,
"incorrect SA out counts: expected %d != %d" % (count, pkts),
)
self.assert_packet_counter_equal(self.tun6_encrypt_node_name, count)
self.assert_packet_counter_equal(self.tun6_decrypt_node_name[0], count)
def verify_decrypted6(self, p, rxs):
for rx in rxs:
self.assert_equal(rx[IPv6].src, p.remote_tun_if_host)
self.assert_equal(rx[IPv6].dst, self.pg1.remote_ip6)
self.assert_packet_checksums_valid(rx)
def verify_encrypted6(self, p, sa, rxs):
for rx in rxs:
self.assert_packet_checksums_valid(rx)
self.assertEqual(len(rx) - len(Ether()) - len(IPv6()), rx[IPv6].plen)
self.assert_equal(rx[IPv6].hlim, p.outer_hop_limit)
if p.outer_flow_label:
self.assert_equal(rx[IPv6].fl, p.outer_flow_label)
try:
decrypt_pkt = p.vpp_tun_sa.decrypt(rx[IPv6])
if not decrypt_pkt.haslayer(IPv6):
decrypt_pkt = IPv6(decrypt_pkt[Raw].load)
self.assert_packet_checksums_valid(decrypt_pkt)
self.assert_equal(decrypt_pkt.src, self.pg1.remote_ip6)
self.assert_equal(decrypt_pkt.dst, p.remote_tun_if_host)
self.assert_equal(decrypt_pkt.hlim, p.inner_hop_limit - 1)
self.assert_equal(decrypt_pkt.fl, p.inner_flow_label)
except:
self.logger.debug(ppp("Unexpected packet:", rx))
try:
self.logger.debug(ppp("Decrypted packet:", decrypt_pkt))
except:
pass
raise
ipsec: IPSec interface correct drop w/ no protection Type: improvement When an IPSec interface is first constructed, the end node of the feature arc is not changed, which means it is interface-output. This means that traffic directed into adjacencies on the link, that do not have protection (w/ an SA), drop like this: ... 00:00:01:111710: ip4-midchain tx_sw_if_index 4 dpo-idx 24 : ipv4 via 0.0.0.0 ipsec0: mtu:9000 next:6 flags:[] stacked-on: [@1]: dpo-drop ip4 flow hash: 0x00000000 00000000: 4500005c000100003f01cb8cac100202010101010800ecf40000000058585858 00000020: 58585858585858585858585858585858585858585858585858585858 00:00:01:111829: local0-output ipsec0 00000000: 4500005c000100003f01cb8cac100202010101010800ecf40000000058585858 00000020: 5858585858585858585858585858585858585858585858585858585858585858 00000040: 58585858585858585858585858585858585858585858585858585858c2cf08c0 00000060: 2a2c103cd0126bd8b03c4ec20ce2bd02dd77b3e3a4f49664 00:00:01:112017: error-drop rx:pg1 00:00:01:112034: drop local0-output: interface is down although that's a drop, no packets should go to local0, and we want all IPvX packets to go through ipX-drop. This change sets the interface's end-arc node to the appropriate drop node when the interface is created, and when the last protection is removed. The resulting drop is: ... 00:00:01:111504: ip4-midchain tx_sw_if_index 4 dpo-idx 24 : ipv4 via 0.0.0.0 ipsec0: mtu:9000 next:0 flags:[] stacked-on: [@1]: dpo-drop ip4 flow hash: 0x00000000 00000000: 4500005c000100003f01cb8cac100202010101010800ecf40000000058585858 00000020: 58585858585858585858585858585858585858585858585858585858 00:00:01:111533: ip4-drop ICMP: 172.16.2.2 -> 1.1.1.1 tos 0x00, ttl 63, length 92, checksum 0xcb8c dscp CS0 ecn NON_ECN fragment id 0x0001 ICMP echo_request checksum 0xecf4 id 0 00:00:01:111620: error-drop rx:pg1 00:00:01:111640: drop null-node: blackholed packets Signed-off-by: Neale Ranns <neale@graphiant.com> Change-Id: I7e7de23c541d9f1210a05e6984a688f1f821a155
2022-01-10 10:38:43 +00:00
def verify_drop_tun_tx_66(self, p_in, count=1, payload_size=64):
self.vapi.cli("clear errors")
self.vapi.cli("clear ipsec sa")
send_pkts = self.gen_pkts6(
p_in,
self.pg1,
src=self.pg1.remote_ip6,
dst=p_in.remote_tun_if_host,
count=count,
payload_size=payload_size,
)
ipsec: IPSec interface correct drop w/ no protection Type: improvement When an IPSec interface is first constructed, the end node of the feature arc is not changed, which means it is interface-output. This means that traffic directed into adjacencies on the link, that do not have protection (w/ an SA), drop like this: ... 00:00:01:111710: ip4-midchain tx_sw_if_index 4 dpo-idx 24 : ipv4 via 0.0.0.0 ipsec0: mtu:9000 next:6 flags:[] stacked-on: [@1]: dpo-drop ip4 flow hash: 0x00000000 00000000: 4500005c000100003f01cb8cac100202010101010800ecf40000000058585858 00000020: 58585858585858585858585858585858585858585858585858585858 00:00:01:111829: local0-output ipsec0 00000000: 4500005c000100003f01cb8cac100202010101010800ecf40000000058585858 00000020: 5858585858585858585858585858585858585858585858585858585858585858 00000040: 58585858585858585858585858585858585858585858585858585858c2cf08c0 00000060: 2a2c103cd0126bd8b03c4ec20ce2bd02dd77b3e3a4f49664 00:00:01:112017: error-drop rx:pg1 00:00:01:112034: drop local0-output: interface is down although that's a drop, no packets should go to local0, and we want all IPvX packets to go through ipX-drop. This change sets the interface's end-arc node to the appropriate drop node when the interface is created, and when the last protection is removed. The resulting drop is: ... 00:00:01:111504: ip4-midchain tx_sw_if_index 4 dpo-idx 24 : ipv4 via 0.0.0.0 ipsec0: mtu:9000 next:0 flags:[] stacked-on: [@1]: dpo-drop ip4 flow hash: 0x00000000 00000000: 4500005c000100003f01cb8cac100202010101010800ecf40000000058585858 00000020: 58585858585858585858585858585858585858585858585858585858 00:00:01:111533: ip4-drop ICMP: 172.16.2.2 -> 1.1.1.1 tos 0x00, ttl 63, length 92, checksum 0xcb8c dscp CS0 ecn NON_ECN fragment id 0x0001 ICMP echo_request checksum 0xecf4 id 0 00:00:01:111620: error-drop rx:pg1 00:00:01:111640: drop null-node: blackholed packets Signed-off-by: Neale Ranns <neale@graphiant.com> Change-Id: I7e7de23c541d9f1210a05e6984a688f1f821a155
2022-01-10 10:38:43 +00:00
self.send_and_assert_no_replies(self.tun_if, send_pkts)
self.logger.info(self.vapi.cli("sh punt stats"))
def verify_drop_tun_rx_66(self, p_in, count=1, payload_size=64):
self.vapi.cli("clear errors")
self.vapi.cli("clear ipsec sa")
send_pkts = self.gen_encrypt_pkts6(
p_in,
p_in.scapy_tun_sa,
self.tun_if,
src=p_in.remote_tun_if_host,
dst=self.pg1.remote_ip6,
count=count,
)
self.send_and_assert_no_replies(self.tun_if, send_pkts)
ipsec: IPSec interface correct drop w/ no protection Type: improvement When an IPSec interface is first constructed, the end node of the feature arc is not changed, which means it is interface-output. This means that traffic directed into adjacencies on the link, that do not have protection (w/ an SA), drop like this: ... 00:00:01:111710: ip4-midchain tx_sw_if_index 4 dpo-idx 24 : ipv4 via 0.0.0.0 ipsec0: mtu:9000 next:6 flags:[] stacked-on: [@1]: dpo-drop ip4 flow hash: 0x00000000 00000000: 4500005c000100003f01cb8cac100202010101010800ecf40000000058585858 00000020: 58585858585858585858585858585858585858585858585858585858 00:00:01:111829: local0-output ipsec0 00000000: 4500005c000100003f01cb8cac100202010101010800ecf40000000058585858 00000020: 5858585858585858585858585858585858585858585858585858585858585858 00000040: 58585858585858585858585858585858585858585858585858585858c2cf08c0 00000060: 2a2c103cd0126bd8b03c4ec20ce2bd02dd77b3e3a4f49664 00:00:01:112017: error-drop rx:pg1 00:00:01:112034: drop local0-output: interface is down although that's a drop, no packets should go to local0, and we want all IPvX packets to go through ipX-drop. This change sets the interface's end-arc node to the appropriate drop node when the interface is created, and when the last protection is removed. The resulting drop is: ... 00:00:01:111504: ip4-midchain tx_sw_if_index 4 dpo-idx 24 : ipv4 via 0.0.0.0 ipsec0: mtu:9000 next:0 flags:[] stacked-on: [@1]: dpo-drop ip4 flow hash: 0x00000000 00000000: 4500005c000100003f01cb8cac100202010101010800ecf40000000058585858 00000020: 58585858585858585858585858585858585858585858585858585858 00:00:01:111533: ip4-drop ICMP: 172.16.2.2 -> 1.1.1.1 tos 0x00, ttl 63, length 92, checksum 0xcb8c dscp CS0 ecn NON_ECN fragment id 0x0001 ICMP echo_request checksum 0xecf4 id 0 00:00:01:111620: error-drop rx:pg1 00:00:01:111640: drop null-node: blackholed packets Signed-off-by: Neale Ranns <neale@graphiant.com> Change-Id: I7e7de23c541d9f1210a05e6984a688f1f821a155
2022-01-10 10:38:43 +00:00
def verify_drop_tun_66(self, p_in, count=1, payload_size=64):
self.verify_drop_tun_tx_66(p_in, count=count, payload_size=payload_size)
self.verify_drop_tun_rx_66(p_in, count=count, payload_size=payload_size)
def verify_tun_66(self, p_in, p_out=None, count=1, payload_size=64):
self.vapi.cli("clear errors")
self.vapi.cli("clear ipsec sa")
if not p_out:
p_out = p_in
try:
send_pkts = self.gen_encrypt_pkts6(
p_in,
p_in.scapy_tun_sa,
self.tun_if,
src=p_in.remote_tun_if_host,
dst=self.pg1.remote_ip6,
count=count,
payload_size=payload_size,
)
recv_pkts = self.send_and_expect(self.tun_if, send_pkts, self.pg1)
self.verify_decrypted6(p_in, recv_pkts)
send_pkts = self.gen_pkts6(
p_in,
self.pg1,
src=self.pg1.remote_ip6,
dst=p_out.remote_tun_if_host,
count=count,
payload_size=payload_size,
)
recv_pkts = self.send_and_expect(self.pg1, send_pkts, self.tun_if)
self.verify_encrypted6(p_out, p_out.vpp_tun_sa, recv_pkts)
for rx in recv_pkts:
self.assertEqual(rx[IPv6].src, p_out.tun_src)
self.assertEqual(rx[IPv6].dst, p_out.tun_dst)
finally:
self.logger.info(self.vapi.ppcli("show error"))
self.logger.info(self.vapi.ppcli("show ipsec all"))
self.verify_counters6(p_in, p_out, count)
def verify_tun_reass_66(self, p):
self.vapi.cli("clear errors")
self.vapi.ip_reassembly_enable_disable(
sw_if_index=self.tun_if.sw_if_index, enable_ip6=True
)
try:
send_pkts = self.gen_encrypt_pkts6(
p,
p.scapy_tun_sa,
self.tun_if,
src=p.remote_tun_if_host,
dst=self.pg1.remote_ip6,
count=1,
payload_size=1850,
)
send_pkts = fragment_rfc8200(send_pkts[0], 1, 1400, self.logger)
recv_pkts = self.send_and_expect(self.tun_if, send_pkts, self.pg1, n_rx=1)
self.verify_decrypted6(p, recv_pkts)
send_pkts = self.gen_pkts6(
p,
self.pg1,
src=self.pg1.remote_ip6,
dst=p.remote_tun_if_host,
count=1,
payload_size=64,
)
recv_pkts = self.send_and_expect(self.pg1, send_pkts, self.tun_if)
self.verify_encrypted6(p, p.vpp_tun_sa, recv_pkts)
finally:
self.logger.info(self.vapi.ppcli("show error"))
self.logger.info(self.vapi.ppcli("show ipsec all"))
self.verify_counters6(p, p, 1)
self.vapi.ip_reassembly_enable_disable(
sw_if_index=self.tun_if.sw_if_index, enable_ip6=False
)
def verify_tun_46(self, p, count=1):
"""ipsec 4o6 tunnel basic test"""
self.vapi.cli("clear errors")
self.vapi.cli("clear ipsec sa")
try:
send_pkts = self.gen_encrypt_pkts(
p,
p.scapy_tun_sa,
self.tun_if,
src=p.remote_tun_if_host4,
dst=self.pg1.remote_ip4,
count=count,
)
recv_pkts = self.send_and_expect(self.tun_if, send_pkts, self.pg1)
for recv_pkt in recv_pkts:
self.assert_equal(recv_pkt[IP].src, p.remote_tun_if_host4)
self.assert_equal(recv_pkt[IP].dst, self.pg1.remote_ip4)
self.assert_packet_checksums_valid(recv_pkt)
send_pkts = self.gen_pkts(
self.pg1,
src=self.pg1.remote_ip4,
dst=p.remote_tun_if_host4,
count=count,
)
recv_pkts = self.send_and_expect(self.pg1, send_pkts, self.tun_if)
for recv_pkt in recv_pkts:
try:
decrypt_pkt = p.vpp_tun_sa.decrypt(recv_pkt[IPv6])
if not decrypt_pkt.haslayer(IP):
decrypt_pkt = IP(decrypt_pkt[Raw].load)
self.assert_equal(decrypt_pkt.src, self.pg1.remote_ip4)
self.assert_equal(decrypt_pkt.dst, p.remote_tun_if_host4)
self.assert_packet_checksums_valid(decrypt_pkt)
except:
self.logger.debug(ppp("Unexpected packet:", recv_pkt))
try:
self.logger.debug(ppp("Decrypted packet:", decrypt_pkt))
except:
pass
raise
finally:
self.logger.info(self.vapi.ppcli("show error"))
self.logger.info(self.vapi.ppcli("show ipsec all"))
self.verify_counters6(p, p, count)
def verify_keepalive(self, p):
# the sizeof Raw is calculated to pad to the minimum ehternet
# frame size of 64 btyes
pkt = (
Ether(src=self.tun_if.remote_mac, dst=self.tun_if.local_mac)
/ IPv6(src=p.remote_tun_if_host, dst=self.tun_if.local_ip6)
/ UDP(sport=333, dport=4500)
/ Raw(b"\xff")
/ Padding(0 * 1)
)
self.send_and_assert_no_replies(self.tun_if, pkt * 31)
self.assert_error_counter_equal(
"/err/%s/nat_keepalive" % self.tun6_input_node, 31
)
pkt = (
Ether(src=self.tun_if.remote_mac, dst=self.tun_if.local_mac)
/ IPv6(src=p.remote_tun_if_host, dst=self.tun_if.local_ip6)
/ UDP(sport=333, dport=4500)
/ Raw(b"\xfe")
)
self.send_and_assert_no_replies(self.tun_if, pkt * 31)
self.assert_error_counter_equal("/err/%s/too_short" % self.tun6_input_node, 31)
pkt = (
Ether(src=self.tun_if.remote_mac, dst=self.tun_if.local_mac)
/ IPv6(src=p.remote_tun_if_host, dst=self.tun_if.local_ip6)
/ UDP(sport=333, dport=4500)
/ Raw(b"\xfe")
/ Padding(0 * 21)
)
self.send_and_assert_no_replies(self.tun_if, pkt * 31)
self.assert_error_counter_equal("/err/%s/too_short" % self.tun6_input_node, 62)
class IpsecTun6Tests(IpsecTun6):
"""UT test methods for Tunnel v6"""
def test_tun_basic66(self):
"""ipsec 6o6 tunnel basic test"""
self.verify_tun_66(self.params[socket.AF_INET6], count=1)
def test_tun_reass_basic66(self):
"""ipsec 6o6 tunnel basic reassembly test"""
self.verify_tun_reass_66(self.params[socket.AF_INET6])
def test_tun_burst66(self):
"""ipsec 6o6 tunnel burst test"""
self.verify_tun_66(self.params[socket.AF_INET6], count=257)
class IpsecTun6HandoffTests(IpsecTun6):
"""UT test methods for Tunnel v6 with multiple workers"""
vpp_worker_count = 2
def test_tun_handoff_66(self):
"""ipsec 6o6 tunnel worker hand-off test"""
self.vapi.cli("clear errors")
self.vapi.cli("clear ipsec sa")
N_PKTS = 15
p = self.params[socket.AF_INET6]
# inject alternately on worker 0 and 1. all counts on the SA
# should be against worker 0
for worker in [0, 1, 0, 1]:
send_pkts = self.gen_encrypt_pkts6(
p,
p.scapy_tun_sa,
self.tun_if,
src=p.remote_tun_if_host,
dst=self.pg1.remote_ip6,
count=N_PKTS,
)
recv_pkts = self.send_and_expect(
self.tun_if, send_pkts, self.pg1, worker=worker
)
self.verify_decrypted6(p, recv_pkts)
send_pkts = self.gen_pkts6(
p,
self.pg1,
src=self.pg1.remote_ip6,
dst=p.remote_tun_if_host,
count=N_PKTS,
)
recv_pkts = self.send_and_expect(
self.pg1, send_pkts, self.tun_if, worker=worker
)
self.verify_encrypted6(p, p.vpp_tun_sa, recv_pkts)
# all counts against the first worker that was used
self.verify_counters6(p, p, 4 * N_PKTS, worker=0)
class IpsecTun4HandoffTests(IpsecTun4):
"""UT test methods for Tunnel v4 with multiple workers"""
vpp_worker_count = 2
def test_tun_handooff_44(self):
"""ipsec 4o4 tunnel worker hand-off test"""
self.vapi.cli("clear errors")
self.vapi.cli("clear ipsec sa")
N_PKTS = 15
p = self.params[socket.AF_INET]
# inject alternately on worker 0 and 1. all counts on the SA
# should be against worker 0
for worker in [0, 1, 0, 1]:
send_pkts = self.gen_encrypt_pkts(
p,
p.scapy_tun_sa,
self.tun_if,
src=p.remote_tun_if_host,
dst=self.pg1.remote_ip4,
count=N_PKTS,
)
recv_pkts = self.send_and_expect(
self.tun_if, send_pkts, self.pg1, worker=worker
)
self.verify_decrypted(p, recv_pkts)
send_pkts = self.gen_pkts(
self.pg1,
src=self.pg1.remote_ip4,
dst=p.remote_tun_if_host,
count=N_PKTS,
)
recv_pkts = self.send_and_expect(
self.pg1, send_pkts, self.tun_if, worker=worker
)
self.verify_encrypted(p, p.vpp_tun_sa, recv_pkts)
# all counts against the first worker that was used
self.verify_counters4(p, 4 * N_PKTS, worker=0)
class IpsecTun46Tests(IpsecTun4Tests, IpsecTun6Tests):
"""UT test methods for Tunnel v6 & v4"""
pass
class IPSecIPv4Fwd(VppTestCase):
"""Test IPSec by capturing and verifying IPv4 forwarded pkts"""
ipsec: Performance improvement of ipsec4_output_node using flow cache Adding flow cache support to improve outbound IPv4/IPSec SPD lookup performance. Details about flow cache: Mechanism: 1. First packet of a flow will undergo linear search in SPD table. Once a policy match is found, a new entry will be added into the flow cache. From 2nd packet onwards, the policy lookup will happen in flow cache. 2. The flow cache is implemented using bihash without collision handling. This will avoid the logic to age out or recycle the old flows in flow cache. Whenever a collision occurs, old entry will be overwritten by the new entry. Worst case is when all the 256 packets in a batch result in collision and fall back to linear search. Average and best case will be O(1). 3. The size of flow cache is fixed and decided based on the number of flows to be supported. The default is set to 1 million flows. This can be made as a configurable option as a next step. 4. Whenever a SPD rule is added/deleted by the control plane, the flow cache entries will be completely deleted (reset) in the control plane. The assumption here is that SPD rule add/del is not a frequent operation from control plane. Flow cache reset is done, by putting the data plane in fall back mode, to bypass flow cache and do linear search till the SPD rule add/delete operation is complete. Once the rule is successfully added/deleted, the data plane will be allowed to make use of the flow cache. The flow cache will be reset only after flushing out the inflight packets from all the worker cores using vlib_worker_wait_one_loop(). Details about bihash usage: 1. A new bihash template (16_8) is added to support IPv4 5 tuple. BIHASH_KVP_PER_PAGE and BIHASH_KVP_AT_BUCKET_LEVEL are set to 1 in the new template. It means only one KVP is supported per bucket. 2. Collision handling is avoided by calling BV (clib_bihash_add_or_overwrite_stale) function. Through the stale callback function pointer, the KVP entry will be overwritten during collision. 3. Flow cache reset is done using BV (clib_bihash_foreach_key_value_pair) function. Through the callback function pointer, the KVP value is reset to ~0ULL. MRR performance numbers with 1 core, 1 ESP Tunnel, null-encrypt, 64B for different SPD policy matching indices: SPD Policy index : 1 10 100 1000 Throughput : MPPS/MPPS MPPS/MPPS MPPS/MPPS KPPS/MPPS (Baseline/Optimized) ARM Neoverse N1 : 5.2/4.84 4.55/4.84 2.11/4.84 329.5/4.84 ARM TX2 : 2.81/2.6 2.51/2.6 1.27/2.6 176.62/2.6 INTEL SKX : 4.93/4.48 4.29/4.46 2.05/4.48 336.79/4.47 Next Steps: Following can be made as a configurable option through startup conf at IPSec level: 1. Enable/Disable Flow cache. 2. Bihash configuration like number of buckets and memory size. 3. Dual/Quad loop unroll can be applied around bihash to further improve the performance. 4. The same flow cache logic can be applied for IPv6 as well as in IPSec inbound direction. A deeper and wider flow cache using bihash_40_8 can replace existing bihash_16_8, to make it common for both IPv4 and IPv6 in both outbound and inbound directions. Following changes are made based on the review comments: 1. ON/OFF flow cache through startup conf. Default: OFF 2. Flow cache stale entry detection using epoch counter. 3. Avoid host order endianness conversion during flow cache lookup. 4. Move IPSec startup conf to a common file. 5. Added SPD flow cache unit test case 6. Replaced bihash with vectors to implement flow cache. 7. ipsec_add_del_policy API is not mpsafe. Cleaned up inflight packets check in control plane. Type: improvement Signed-off-by: mgovind <govindarajan.Mohandoss@arm.com> Signed-off-by: Zachary Leaf <zachary.leaf@arm.com> Tested-by: Jieqiang Wang <jieqiang.wang@arm.com> Change-Id: I62b4d6625fbc6caf292427a5d2046aa5672b2006
2021-03-19 19:20:49 +00:00
@classmethod
def setUpConstants(cls):
super(IPSecIPv4Fwd, cls).setUpConstants()
ipsec: Performance improvement of ipsec4_output_node using flow cache Adding flow cache support to improve outbound IPv4/IPSec SPD lookup performance. Details about flow cache: Mechanism: 1. First packet of a flow will undergo linear search in SPD table. Once a policy match is found, a new entry will be added into the flow cache. From 2nd packet onwards, the policy lookup will happen in flow cache. 2. The flow cache is implemented using bihash without collision handling. This will avoid the logic to age out or recycle the old flows in flow cache. Whenever a collision occurs, old entry will be overwritten by the new entry. Worst case is when all the 256 packets in a batch result in collision and fall back to linear search. Average and best case will be O(1). 3. The size of flow cache is fixed and decided based on the number of flows to be supported. The default is set to 1 million flows. This can be made as a configurable option as a next step. 4. Whenever a SPD rule is added/deleted by the control plane, the flow cache entries will be completely deleted (reset) in the control plane. The assumption here is that SPD rule add/del is not a frequent operation from control plane. Flow cache reset is done, by putting the data plane in fall back mode, to bypass flow cache and do linear search till the SPD rule add/delete operation is complete. Once the rule is successfully added/deleted, the data plane will be allowed to make use of the flow cache. The flow cache will be reset only after flushing out the inflight packets from all the worker cores using vlib_worker_wait_one_loop(). Details about bihash usage: 1. A new bihash template (16_8) is added to support IPv4 5 tuple. BIHASH_KVP_PER_PAGE and BIHASH_KVP_AT_BUCKET_LEVEL are set to 1 in the new template. It means only one KVP is supported per bucket. 2. Collision handling is avoided by calling BV (clib_bihash_add_or_overwrite_stale) function. Through the stale callback function pointer, the KVP entry will be overwritten during collision. 3. Flow cache reset is done using BV (clib_bihash_foreach_key_value_pair) function. Through the callback function pointer, the KVP value is reset to ~0ULL. MRR performance numbers with 1 core, 1 ESP Tunnel, null-encrypt, 64B for different SPD policy matching indices: SPD Policy index : 1 10 100 1000 Throughput : MPPS/MPPS MPPS/MPPS MPPS/MPPS KPPS/MPPS (Baseline/Optimized) ARM Neoverse N1 : 5.2/4.84 4.55/4.84 2.11/4.84 329.5/4.84 ARM TX2 : 2.81/2.6 2.51/2.6 1.27/2.6 176.62/2.6 INTEL SKX : 4.93/4.48 4.29/4.46 2.05/4.48 336.79/4.47 Next Steps: Following can be made as a configurable option through startup conf at IPSec level: 1. Enable/Disable Flow cache. 2. Bihash configuration like number of buckets and memory size. 3. Dual/Quad loop unroll can be applied around bihash to further improve the performance. 4. The same flow cache logic can be applied for IPv6 as well as in IPSec inbound direction. A deeper and wider flow cache using bihash_40_8 can replace existing bihash_16_8, to make it common for both IPv4 and IPv6 in both outbound and inbound directions. Following changes are made based on the review comments: 1. ON/OFF flow cache through startup conf. Default: OFF 2. Flow cache stale entry detection using epoch counter. 3. Avoid host order endianness conversion during flow cache lookup. 4. Move IPSec startup conf to a common file. 5. Added SPD flow cache unit test case 6. Replaced bihash with vectors to implement flow cache. 7. ipsec_add_del_policy API is not mpsafe. Cleaned up inflight packets check in control plane. Type: improvement Signed-off-by: mgovind <govindarajan.Mohandoss@arm.com> Signed-off-by: Zachary Leaf <zachary.leaf@arm.com> Tested-by: Jieqiang Wang <jieqiang.wang@arm.com> Change-Id: I62b4d6625fbc6caf292427a5d2046aa5672b2006
2021-03-19 19:20:49 +00:00
def setUp(self):
super(IPSecIPv4Fwd, self).setUp()
ipsec: Performance improvement of ipsec4_output_node using flow cache Adding flow cache support to improve outbound IPv4/IPSec SPD lookup performance. Details about flow cache: Mechanism: 1. First packet of a flow will undergo linear search in SPD table. Once a policy match is found, a new entry will be added into the flow cache. From 2nd packet onwards, the policy lookup will happen in flow cache. 2. The flow cache is implemented using bihash without collision handling. This will avoid the logic to age out or recycle the old flows in flow cache. Whenever a collision occurs, old entry will be overwritten by the new entry. Worst case is when all the 256 packets in a batch result in collision and fall back to linear search. Average and best case will be O(1). 3. The size of flow cache is fixed and decided based on the number of flows to be supported. The default is set to 1 million flows. This can be made as a configurable option as a next step. 4. Whenever a SPD rule is added/deleted by the control plane, the flow cache entries will be completely deleted (reset) in the control plane. The assumption here is that SPD rule add/del is not a frequent operation from control plane. Flow cache reset is done, by putting the data plane in fall back mode, to bypass flow cache and do linear search till the SPD rule add/delete operation is complete. Once the rule is successfully added/deleted, the data plane will be allowed to make use of the flow cache. The flow cache will be reset only after flushing out the inflight packets from all the worker cores using vlib_worker_wait_one_loop(). Details about bihash usage: 1. A new bihash template (16_8) is added to support IPv4 5 tuple. BIHASH_KVP_PER_PAGE and BIHASH_KVP_AT_BUCKET_LEVEL are set to 1 in the new template. It means only one KVP is supported per bucket. 2. Collision handling is avoided by calling BV (clib_bihash_add_or_overwrite_stale) function. Through the stale callback function pointer, the KVP entry will be overwritten during collision. 3. Flow cache reset is done using BV (clib_bihash_foreach_key_value_pair) function. Through the callback function pointer, the KVP value is reset to ~0ULL. MRR performance numbers with 1 core, 1 ESP Tunnel, null-encrypt, 64B for different SPD policy matching indices: SPD Policy index : 1 10 100 1000 Throughput : MPPS/MPPS MPPS/MPPS MPPS/MPPS KPPS/MPPS (Baseline/Optimized) ARM Neoverse N1 : 5.2/4.84 4.55/4.84 2.11/4.84 329.5/4.84 ARM TX2 : 2.81/2.6 2.51/2.6 1.27/2.6 176.62/2.6 INTEL SKX : 4.93/4.48 4.29/4.46 2.05/4.48 336.79/4.47 Next Steps: Following can be made as a configurable option through startup conf at IPSec level: 1. Enable/Disable Flow cache. 2. Bihash configuration like number of buckets and memory size. 3. Dual/Quad loop unroll can be applied around bihash to further improve the performance. 4. The same flow cache logic can be applied for IPv6 as well as in IPSec inbound direction. A deeper and wider flow cache using bihash_40_8 can replace existing bihash_16_8, to make it common for both IPv4 and IPv6 in both outbound and inbound directions. Following changes are made based on the review comments: 1. ON/OFF flow cache through startup conf. Default: OFF 2. Flow cache stale entry detection using epoch counter. 3. Avoid host order endianness conversion during flow cache lookup. 4. Move IPSec startup conf to a common file. 5. Added SPD flow cache unit test case 6. Replaced bihash with vectors to implement flow cache. 7. ipsec_add_del_policy API is not mpsafe. Cleaned up inflight packets check in control plane. Type: improvement Signed-off-by: mgovind <govindarajan.Mohandoss@arm.com> Signed-off-by: Zachary Leaf <zachary.leaf@arm.com> Tested-by: Jieqiang Wang <jieqiang.wang@arm.com> Change-Id: I62b4d6625fbc6caf292427a5d2046aa5672b2006
2021-03-19 19:20:49 +00:00
# store SPD objects so we can remove configs on tear down
self.spd_objs = []
self.spd_policies = []
def tearDown(self):
# remove SPD policies
for obj in self.spd_policies:
obj.remove_vpp_config()
self.spd_policies = []
# remove SPD items (interface bindings first, then SPD)
for obj in reversed(self.spd_objs):
obj.remove_vpp_config()
self.spd_objs = []
# close down pg intfs
for pg in self.pg_interfaces:
pg.unconfig_ip4()
pg.admin_down()
super(IPSecIPv4Fwd, self).tearDown()
ipsec: Performance improvement of ipsec4_output_node using flow cache Adding flow cache support to improve outbound IPv4/IPSec SPD lookup performance. Details about flow cache: Mechanism: 1. First packet of a flow will undergo linear search in SPD table. Once a policy match is found, a new entry will be added into the flow cache. From 2nd packet onwards, the policy lookup will happen in flow cache. 2. The flow cache is implemented using bihash without collision handling. This will avoid the logic to age out or recycle the old flows in flow cache. Whenever a collision occurs, old entry will be overwritten by the new entry. Worst case is when all the 256 packets in a batch result in collision and fall back to linear search. Average and best case will be O(1). 3. The size of flow cache is fixed and decided based on the number of flows to be supported. The default is set to 1 million flows. This can be made as a configurable option as a next step. 4. Whenever a SPD rule is added/deleted by the control plane, the flow cache entries will be completely deleted (reset) in the control plane. The assumption here is that SPD rule add/del is not a frequent operation from control plane. Flow cache reset is done, by putting the data plane in fall back mode, to bypass flow cache and do linear search till the SPD rule add/delete operation is complete. Once the rule is successfully added/deleted, the data plane will be allowed to make use of the flow cache. The flow cache will be reset only after flushing out the inflight packets from all the worker cores using vlib_worker_wait_one_loop(). Details about bihash usage: 1. A new bihash template (16_8) is added to support IPv4 5 tuple. BIHASH_KVP_PER_PAGE and BIHASH_KVP_AT_BUCKET_LEVEL are set to 1 in the new template. It means only one KVP is supported per bucket. 2. Collision handling is avoided by calling BV (clib_bihash_add_or_overwrite_stale) function. Through the stale callback function pointer, the KVP entry will be overwritten during collision. 3. Flow cache reset is done using BV (clib_bihash_foreach_key_value_pair) function. Through the callback function pointer, the KVP value is reset to ~0ULL. MRR performance numbers with 1 core, 1 ESP Tunnel, null-encrypt, 64B for different SPD policy matching indices: SPD Policy index : 1 10 100 1000 Throughput : MPPS/MPPS MPPS/MPPS MPPS/MPPS KPPS/MPPS (Baseline/Optimized) ARM Neoverse N1 : 5.2/4.84 4.55/4.84 2.11/4.84 329.5/4.84 ARM TX2 : 2.81/2.6 2.51/2.6 1.27/2.6 176.62/2.6 INTEL SKX : 4.93/4.48 4.29/4.46 2.05/4.48 336.79/4.47 Next Steps: Following can be made as a configurable option through startup conf at IPSec level: 1. Enable/Disable Flow cache. 2. Bihash configuration like number of buckets and memory size. 3. Dual/Quad loop unroll can be applied around bihash to further improve the performance. 4. The same flow cache logic can be applied for IPv6 as well as in IPSec inbound direction. A deeper and wider flow cache using bihash_40_8 can replace existing bihash_16_8, to make it common for both IPv4 and IPv6 in both outbound and inbound directions. Following changes are made based on the review comments: 1. ON/OFF flow cache through startup conf. Default: OFF 2. Flow cache stale entry detection using epoch counter. 3. Avoid host order endianness conversion during flow cache lookup. 4. Move IPSec startup conf to a common file. 5. Added SPD flow cache unit test case 6. Replaced bihash with vectors to implement flow cache. 7. ipsec_add_del_policy API is not mpsafe. Cleaned up inflight packets check in control plane. Type: improvement Signed-off-by: mgovind <govindarajan.Mohandoss@arm.com> Signed-off-by: Zachary Leaf <zachary.leaf@arm.com> Tested-by: Jieqiang Wang <jieqiang.wang@arm.com> Change-Id: I62b4d6625fbc6caf292427a5d2046aa5672b2006
2021-03-19 19:20:49 +00:00
def create_interfaces(self, num_ifs=2):
# create interfaces pg0 ... pg<num_ifs>
self.create_pg_interfaces(range(num_ifs))
for pg in self.pg_interfaces:
# put the interface up
pg.admin_up()
# configure IPv4 address on the interface
pg.config_ip4()
# resolve ARP, so that we know VPP MAC
pg.resolve_arp()
self.logger.info(self.vapi.ppcli("show int addr"))
def spd_create_and_intf_add(self, spd_id, pg_list):
spd = VppIpsecSpd(self, spd_id)
spd.add_vpp_config()
self.spd_objs.append(spd)
for pg in pg_list:
spdItf = VppIpsecSpdItfBinding(self, spd, pg)
spdItf.add_vpp_config()
self.spd_objs.append(spdItf)
def get_policy(self, policy_type):
e = VppEnum.vl_api_ipsec_spd_action_t
if policy_type == "protect":
return e.IPSEC_API_SPD_ACTION_PROTECT
elif policy_type == "bypass":
return e.IPSEC_API_SPD_ACTION_BYPASS
elif policy_type == "discard":
return e.IPSEC_API_SPD_ACTION_DISCARD
else:
raise Exception("Invalid policy type: %s", policy_type)
def spd_add_rem_policy(
self,
spd_id,
src_if,
dst_if,
proto,
is_out,
priority,
policy_type,
remove=False,
all_ips=False,
ip_range=False,
local_ip_start=ip_address("0.0.0.0"),
local_ip_stop=ip_address("255.255.255.255"),
remote_ip_start=ip_address("0.0.0.0"),
remote_ip_stop=ip_address("255.255.255.255"),
remote_port_start=0,
remote_port_stop=65535,
local_port_start=0,
local_port_stop=65535,
):
ipsec: Performance improvement of ipsec4_output_node using flow cache Adding flow cache support to improve outbound IPv4/IPSec SPD lookup performance. Details about flow cache: Mechanism: 1. First packet of a flow will undergo linear search in SPD table. Once a policy match is found, a new entry will be added into the flow cache. From 2nd packet onwards, the policy lookup will happen in flow cache. 2. The flow cache is implemented using bihash without collision handling. This will avoid the logic to age out or recycle the old flows in flow cache. Whenever a collision occurs, old entry will be overwritten by the new entry. Worst case is when all the 256 packets in a batch result in collision and fall back to linear search. Average and best case will be O(1). 3. The size of flow cache is fixed and decided based on the number of flows to be supported. The default is set to 1 million flows. This can be made as a configurable option as a next step. 4. Whenever a SPD rule is added/deleted by the control plane, the flow cache entries will be completely deleted (reset) in the control plane. The assumption here is that SPD rule add/del is not a frequent operation from control plane. Flow cache reset is done, by putting the data plane in fall back mode, to bypass flow cache and do linear search till the SPD rule add/delete operation is complete. Once the rule is successfully added/deleted, the data plane will be allowed to make use of the flow cache. The flow cache will be reset only after flushing out the inflight packets from all the worker cores using vlib_worker_wait_one_loop(). Details about bihash usage: 1. A new bihash template (16_8) is added to support IPv4 5 tuple. BIHASH_KVP_PER_PAGE and BIHASH_KVP_AT_BUCKET_LEVEL are set to 1 in the new template. It means only one KVP is supported per bucket. 2. Collision handling is avoided by calling BV (clib_bihash_add_or_overwrite_stale) function. Through the stale callback function pointer, the KVP entry will be overwritten during collision. 3. Flow cache reset is done using BV (clib_bihash_foreach_key_value_pair) function. Through the callback function pointer, the KVP value is reset to ~0ULL. MRR performance numbers with 1 core, 1 ESP Tunnel, null-encrypt, 64B for different SPD policy matching indices: SPD Policy index : 1 10 100 1000 Throughput : MPPS/MPPS MPPS/MPPS MPPS/MPPS KPPS/MPPS (Baseline/Optimized) ARM Neoverse N1 : 5.2/4.84 4.55/4.84 2.11/4.84 329.5/4.84 ARM TX2 : 2.81/2.6 2.51/2.6 1.27/2.6 176.62/2.6 INTEL SKX : 4.93/4.48 4.29/4.46 2.05/4.48 336.79/4.47 Next Steps: Following can be made as a configurable option through startup conf at IPSec level: 1. Enable/Disable Flow cache. 2. Bihash configuration like number of buckets and memory size. 3. Dual/Quad loop unroll can be applied around bihash to further improve the performance. 4. The same flow cache logic can be applied for IPv6 as well as in IPSec inbound direction. A deeper and wider flow cache using bihash_40_8 can replace existing bihash_16_8, to make it common for both IPv4 and IPv6 in both outbound and inbound directions. Following changes are made based on the review comments: 1. ON/OFF flow cache through startup conf. Default: OFF 2. Flow cache stale entry detection using epoch counter. 3. Avoid host order endianness conversion during flow cache lookup. 4. Move IPSec startup conf to a common file. 5. Added SPD flow cache unit test case 6. Replaced bihash with vectors to implement flow cache. 7. ipsec_add_del_policy API is not mpsafe. Cleaned up inflight packets check in control plane. Type: improvement Signed-off-by: mgovind <govindarajan.Mohandoss@arm.com> Signed-off-by: Zachary Leaf <zachary.leaf@arm.com> Tested-by: Jieqiang Wang <jieqiang.wang@arm.com> Change-Id: I62b4d6625fbc6caf292427a5d2046aa5672b2006
2021-03-19 19:20:49 +00:00
spd = VppIpsecSpd(self, spd_id)
if all_ips:
src_range_low = ip_address("0.0.0.0")
src_range_high = ip_address("255.255.255.255")
dst_range_low = ip_address("0.0.0.0")
dst_range_high = ip_address("255.255.255.255")
elif ip_range:
src_range_low = local_ip_start
src_range_high = local_ip_stop
dst_range_low = remote_ip_start
dst_range_high = remote_ip_stop
ipsec: Performance improvement of ipsec4_output_node using flow cache Adding flow cache support to improve outbound IPv4/IPSec SPD lookup performance. Details about flow cache: Mechanism: 1. First packet of a flow will undergo linear search in SPD table. Once a policy match is found, a new entry will be added into the flow cache. From 2nd packet onwards, the policy lookup will happen in flow cache. 2. The flow cache is implemented using bihash without collision handling. This will avoid the logic to age out or recycle the old flows in flow cache. Whenever a collision occurs, old entry will be overwritten by the new entry. Worst case is when all the 256 packets in a batch result in collision and fall back to linear search. Average and best case will be O(1). 3. The size of flow cache is fixed and decided based on the number of flows to be supported. The default is set to 1 million flows. This can be made as a configurable option as a next step. 4. Whenever a SPD rule is added/deleted by the control plane, the flow cache entries will be completely deleted (reset) in the control plane. The assumption here is that SPD rule add/del is not a frequent operation from control plane. Flow cache reset is done, by putting the data plane in fall back mode, to bypass flow cache and do linear search till the SPD rule add/delete operation is complete. Once the rule is successfully added/deleted, the data plane will be allowed to make use of the flow cache. The flow cache will be reset only after flushing out the inflight packets from all the worker cores using vlib_worker_wait_one_loop(). Details about bihash usage: 1. A new bihash template (16_8) is added to support IPv4 5 tuple. BIHASH_KVP_PER_PAGE and BIHASH_KVP_AT_BUCKET_LEVEL are set to 1 in the new template. It means only one KVP is supported per bucket. 2. Collision handling is avoided by calling BV (clib_bihash_add_or_overwrite_stale) function. Through the stale callback function pointer, the KVP entry will be overwritten during collision. 3. Flow cache reset is done using BV (clib_bihash_foreach_key_value_pair) function. Through the callback function pointer, the KVP value is reset to ~0ULL. MRR performance numbers with 1 core, 1 ESP Tunnel, null-encrypt, 64B for different SPD policy matching indices: SPD Policy index : 1 10 100 1000 Throughput : MPPS/MPPS MPPS/MPPS MPPS/MPPS KPPS/MPPS (Baseline/Optimized) ARM Neoverse N1 : 5.2/4.84 4.55/4.84 2.11/4.84 329.5/4.84 ARM TX2 : 2.81/2.6 2.51/2.6 1.27/2.6 176.62/2.6 INTEL SKX : 4.93/4.48 4.29/4.46 2.05/4.48 336.79/4.47 Next Steps: Following can be made as a configurable option through startup conf at IPSec level: 1. Enable/Disable Flow cache. 2. Bihash configuration like number of buckets and memory size. 3. Dual/Quad loop unroll can be applied around bihash to further improve the performance. 4. The same flow cache logic can be applied for IPv6 as well as in IPSec inbound direction. A deeper and wider flow cache using bihash_40_8 can replace existing bihash_16_8, to make it common for both IPv4 and IPv6 in both outbound and inbound directions. Following changes are made based on the review comments: 1. ON/OFF flow cache through startup conf. Default: OFF 2. Flow cache stale entry detection using epoch counter. 3. Avoid host order endianness conversion during flow cache lookup. 4. Move IPSec startup conf to a common file. 5. Added SPD flow cache unit test case 6. Replaced bihash with vectors to implement flow cache. 7. ipsec_add_del_policy API is not mpsafe. Cleaned up inflight packets check in control plane. Type: improvement Signed-off-by: mgovind <govindarajan.Mohandoss@arm.com> Signed-off-by: Zachary Leaf <zachary.leaf@arm.com> Tested-by: Jieqiang Wang <jieqiang.wang@arm.com> Change-Id: I62b4d6625fbc6caf292427a5d2046aa5672b2006
2021-03-19 19:20:49 +00:00
else:
src_range_low = src_if.remote_ip4
src_range_high = src_if.remote_ip4
dst_range_low = dst_if.remote_ip4
dst_range_high = dst_if.remote_ip4
spdEntry = VppIpsecSpdEntry(
self,
spd,
0,
src_range_low,
src_range_high,
dst_range_low,
dst_range_high,
proto,
priority=priority,
policy=self.get_policy(policy_type),
is_outbound=is_out,
remote_port_start=remote_port_start,
remote_port_stop=remote_port_stop,
local_port_start=local_port_start,
local_port_stop=local_port_stop,
)
if remove is False:
ipsec: Performance improvement of ipsec4_output_node using flow cache Adding flow cache support to improve outbound IPv4/IPSec SPD lookup performance. Details about flow cache: Mechanism: 1. First packet of a flow will undergo linear search in SPD table. Once a policy match is found, a new entry will be added into the flow cache. From 2nd packet onwards, the policy lookup will happen in flow cache. 2. The flow cache is implemented using bihash without collision handling. This will avoid the logic to age out or recycle the old flows in flow cache. Whenever a collision occurs, old entry will be overwritten by the new entry. Worst case is when all the 256 packets in a batch result in collision and fall back to linear search. Average and best case will be O(1). 3. The size of flow cache is fixed and decided based on the number of flows to be supported. The default is set to 1 million flows. This can be made as a configurable option as a next step. 4. Whenever a SPD rule is added/deleted by the control plane, the flow cache entries will be completely deleted (reset) in the control plane. The assumption here is that SPD rule add/del is not a frequent operation from control plane. Flow cache reset is done, by putting the data plane in fall back mode, to bypass flow cache and do linear search till the SPD rule add/delete operation is complete. Once the rule is successfully added/deleted, the data plane will be allowed to make use of the flow cache. The flow cache will be reset only after flushing out the inflight packets from all the worker cores using vlib_worker_wait_one_loop(). Details about bihash usage: 1. A new bihash template (16_8) is added to support IPv4 5 tuple. BIHASH_KVP_PER_PAGE and BIHASH_KVP_AT_BUCKET_LEVEL are set to 1 in the new template. It means only one KVP is supported per bucket. 2. Collision handling is avoided by calling BV (clib_bihash_add_or_overwrite_stale) function. Through the stale callback function pointer, the KVP entry will be overwritten during collision. 3. Flow cache reset is done using BV (clib_bihash_foreach_key_value_pair) function. Through the callback function pointer, the KVP value is reset to ~0ULL. MRR performance numbers with 1 core, 1 ESP Tunnel, null-encrypt, 64B for different SPD policy matching indices: SPD Policy index : 1 10 100 1000 Throughput : MPPS/MPPS MPPS/MPPS MPPS/MPPS KPPS/MPPS (Baseline/Optimized) ARM Neoverse N1 : 5.2/4.84 4.55/4.84 2.11/4.84 329.5/4.84 ARM TX2 : 2.81/2.6 2.51/2.6 1.27/2.6 176.62/2.6 INTEL SKX : 4.93/4.48 4.29/4.46 2.05/4.48 336.79/4.47 Next Steps: Following can be made as a configurable option through startup conf at IPSec level: 1. Enable/Disable Flow cache. 2. Bihash configuration like number of buckets and memory size. 3. Dual/Quad loop unroll can be applied around bihash to further improve the performance. 4. The same flow cache logic can be applied for IPv6 as well as in IPSec inbound direction. A deeper and wider flow cache using bihash_40_8 can replace existing bihash_16_8, to make it common for both IPv4 and IPv6 in both outbound and inbound directions. Following changes are made based on the review comments: 1. ON/OFF flow cache through startup conf. Default: OFF 2. Flow cache stale entry detection using epoch counter. 3. Avoid host order endianness conversion during flow cache lookup. 4. Move IPSec startup conf to a common file. 5. Added SPD flow cache unit test case 6. Replaced bihash with vectors to implement flow cache. 7. ipsec_add_del_policy API is not mpsafe. Cleaned up inflight packets check in control plane. Type: improvement Signed-off-by: mgovind <govindarajan.Mohandoss@arm.com> Signed-off-by: Zachary Leaf <zachary.leaf@arm.com> Tested-by: Jieqiang Wang <jieqiang.wang@arm.com> Change-Id: I62b4d6625fbc6caf292427a5d2046aa5672b2006
2021-03-19 19:20:49 +00:00
spdEntry.add_vpp_config()
self.spd_policies.append(spdEntry)
else:
spdEntry.remove_vpp_config()
self.spd_policies.remove(spdEntry)
self.logger.info(self.vapi.ppcli("show ipsec all"))
return spdEntry
def create_stream(
self, src_if, dst_if, pkt_count, src_prt=1234, dst_prt=5678, proto="UDP"
):
ipsec: Performance improvement of ipsec4_output_node using flow cache Adding flow cache support to improve outbound IPv4/IPSec SPD lookup performance. Details about flow cache: Mechanism: 1. First packet of a flow will undergo linear search in SPD table. Once a policy match is found, a new entry will be added into the flow cache. From 2nd packet onwards, the policy lookup will happen in flow cache. 2. The flow cache is implemented using bihash without collision handling. This will avoid the logic to age out or recycle the old flows in flow cache. Whenever a collision occurs, old entry will be overwritten by the new entry. Worst case is when all the 256 packets in a batch result in collision and fall back to linear search. Average and best case will be O(1). 3. The size of flow cache is fixed and decided based on the number of flows to be supported. The default is set to 1 million flows. This can be made as a configurable option as a next step. 4. Whenever a SPD rule is added/deleted by the control plane, the flow cache entries will be completely deleted (reset) in the control plane. The assumption here is that SPD rule add/del is not a frequent operation from control plane. Flow cache reset is done, by putting the data plane in fall back mode, to bypass flow cache and do linear search till the SPD rule add/delete operation is complete. Once the rule is successfully added/deleted, the data plane will be allowed to make use of the flow cache. The flow cache will be reset only after flushing out the inflight packets from all the worker cores using vlib_worker_wait_one_loop(). Details about bihash usage: 1. A new bihash template (16_8) is added to support IPv4 5 tuple. BIHASH_KVP_PER_PAGE and BIHASH_KVP_AT_BUCKET_LEVEL are set to 1 in the new template. It means only one KVP is supported per bucket. 2. Collision handling is avoided by calling BV (clib_bihash_add_or_overwrite_stale) function. Through the stale callback function pointer, the KVP entry will be overwritten during collision. 3. Flow cache reset is done using BV (clib_bihash_foreach_key_value_pair) function. Through the callback function pointer, the KVP value is reset to ~0ULL. MRR performance numbers with 1 core, 1 ESP Tunnel, null-encrypt, 64B for different SPD policy matching indices: SPD Policy index : 1 10 100 1000 Throughput : MPPS/MPPS MPPS/MPPS MPPS/MPPS KPPS/MPPS (Baseline/Optimized) ARM Neoverse N1 : 5.2/4.84 4.55/4.84 2.11/4.84 329.5/4.84 ARM TX2 : 2.81/2.6 2.51/2.6 1.27/2.6 176.62/2.6 INTEL SKX : 4.93/4.48 4.29/4.46 2.05/4.48 336.79/4.47 Next Steps: Following can be made as a configurable option through startup conf at IPSec level: 1. Enable/Disable Flow cache. 2. Bihash configuration like number of buckets and memory size. 3. Dual/Quad loop unroll can be applied around bihash to further improve the performance. 4. The same flow cache logic can be applied for IPv6 as well as in IPSec inbound direction. A deeper and wider flow cache using bihash_40_8 can replace existing bihash_16_8, to make it common for both IPv4 and IPv6 in both outbound and inbound directions. Following changes are made based on the review comments: 1. ON/OFF flow cache through startup conf. Default: OFF 2. Flow cache stale entry detection using epoch counter. 3. Avoid host order endianness conversion during flow cache lookup. 4. Move IPSec startup conf to a common file. 5. Added SPD flow cache unit test case 6. Replaced bihash with vectors to implement flow cache. 7. ipsec_add_del_policy API is not mpsafe. Cleaned up inflight packets check in control plane. Type: improvement Signed-off-by: mgovind <govindarajan.Mohandoss@arm.com> Signed-off-by: Zachary Leaf <zachary.leaf@arm.com> Tested-by: Jieqiang Wang <jieqiang.wang@arm.com> Change-Id: I62b4d6625fbc6caf292427a5d2046aa5672b2006
2021-03-19 19:20:49 +00:00
packets = []
# create SA
sa = SecurityAssociation(
ESP,
spi=1000,
crypt_algo="AES-CBC",
crypt_key=b"JPjyOWBeVEQiMe7h",
auth_algo="HMAC-SHA1-96",
auth_key=b"C91KUR9GYMm5GfkEvNjX",
tunnel_header=IP(src=src_if.remote_ip4, dst=dst_if.remote_ip4),
nat_t_header=UDP(sport=src_prt, dport=dst_prt),
)
ipsec: Performance improvement of ipsec4_output_node using flow cache Adding flow cache support to improve outbound IPv4/IPSec SPD lookup performance. Details about flow cache: Mechanism: 1. First packet of a flow will undergo linear search in SPD table. Once a policy match is found, a new entry will be added into the flow cache. From 2nd packet onwards, the policy lookup will happen in flow cache. 2. The flow cache is implemented using bihash without collision handling. This will avoid the logic to age out or recycle the old flows in flow cache. Whenever a collision occurs, old entry will be overwritten by the new entry. Worst case is when all the 256 packets in a batch result in collision and fall back to linear search. Average and best case will be O(1). 3. The size of flow cache is fixed and decided based on the number of flows to be supported. The default is set to 1 million flows. This can be made as a configurable option as a next step. 4. Whenever a SPD rule is added/deleted by the control plane, the flow cache entries will be completely deleted (reset) in the control plane. The assumption here is that SPD rule add/del is not a frequent operation from control plane. Flow cache reset is done, by putting the data plane in fall back mode, to bypass flow cache and do linear search till the SPD rule add/delete operation is complete. Once the rule is successfully added/deleted, the data plane will be allowed to make use of the flow cache. The flow cache will be reset only after flushing out the inflight packets from all the worker cores using vlib_worker_wait_one_loop(). Details about bihash usage: 1. A new bihash template (16_8) is added to support IPv4 5 tuple. BIHASH_KVP_PER_PAGE and BIHASH_KVP_AT_BUCKET_LEVEL are set to 1 in the new template. It means only one KVP is supported per bucket. 2. Collision handling is avoided by calling BV (clib_bihash_add_or_overwrite_stale) function. Through the stale callback function pointer, the KVP entry will be overwritten during collision. 3. Flow cache reset is done using BV (clib_bihash_foreach_key_value_pair) function. Through the callback function pointer, the KVP value is reset to ~0ULL. MRR performance numbers with 1 core, 1 ESP Tunnel, null-encrypt, 64B for different SPD policy matching indices: SPD Policy index : 1 10 100 1000 Throughput : MPPS/MPPS MPPS/MPPS MPPS/MPPS KPPS/MPPS (Baseline/Optimized) ARM Neoverse N1 : 5.2/4.84 4.55/4.84 2.11/4.84 329.5/4.84 ARM TX2 : 2.81/2.6 2.51/2.6 1.27/2.6 176.62/2.6 INTEL SKX : 4.93/4.48 4.29/4.46 2.05/4.48 336.79/4.47 Next Steps: Following can be made as a configurable option through startup conf at IPSec level: 1. Enable/Disable Flow cache. 2. Bihash configuration like number of buckets and memory size. 3. Dual/Quad loop unroll can be applied around bihash to further improve the performance. 4. The same flow cache logic can be applied for IPv6 as well as in IPSec inbound direction. A deeper and wider flow cache using bihash_40_8 can replace existing bihash_16_8, to make it common for both IPv4 and IPv6 in both outbound and inbound directions. Following changes are made based on the review comments: 1. ON/OFF flow cache through startup conf. Default: OFF 2. Flow cache stale entry detection using epoch counter. 3. Avoid host order endianness conversion during flow cache lookup. 4. Move IPSec startup conf to a common file. 5. Added SPD flow cache unit test case 6. Replaced bihash with vectors to implement flow cache. 7. ipsec_add_del_policy API is not mpsafe. Cleaned up inflight packets check in control plane. Type: improvement Signed-off-by: mgovind <govindarajan.Mohandoss@arm.com> Signed-off-by: Zachary Leaf <zachary.leaf@arm.com> Tested-by: Jieqiang Wang <jieqiang.wang@arm.com> Change-Id: I62b4d6625fbc6caf292427a5d2046aa5672b2006
2021-03-19 19:20:49 +00:00
for i in range(pkt_count):
# create packet info stored in the test case instance
info = self.create_packet_info(src_if, dst_if)
# convert the info into packet payload
payload = self.info_to_payload(info)
# create the packet itself
p = []
if proto == "UDP-ESP":
p = Ether(dst=src_if.local_mac, src=src_if.remote_mac) / sa.encrypt(
IP(src=src_if.remote_ip4, dst=dst_if.remote_ip4)
/ UDP(sport=src_prt, dport=dst_prt)
/ Raw(payload)
)
elif proto == "UDP":
p = (
Ether(dst=src_if.local_mac, src=src_if.remote_mac)
/ IP(src=src_if.remote_ip4, dst=dst_if.remote_ip4)
/ UDP(sport=src_prt, dport=dst_prt)
/ Raw(payload)
)
elif proto == "TCP":
p = (
Ether(dst=src_if.local_mac, src=src_if.remote_mac)
/ IP(src=src_if.remote_ip4, dst=dst_if.remote_ip4)
/ TCP(sport=src_prt, dport=dst_prt)
/ Raw(payload)
)
ipsec: Performance improvement of ipsec4_output_node using flow cache Adding flow cache support to improve outbound IPv4/IPSec SPD lookup performance. Details about flow cache: Mechanism: 1. First packet of a flow will undergo linear search in SPD table. Once a policy match is found, a new entry will be added into the flow cache. From 2nd packet onwards, the policy lookup will happen in flow cache. 2. The flow cache is implemented using bihash without collision handling. This will avoid the logic to age out or recycle the old flows in flow cache. Whenever a collision occurs, old entry will be overwritten by the new entry. Worst case is when all the 256 packets in a batch result in collision and fall back to linear search. Average and best case will be O(1). 3. The size of flow cache is fixed and decided based on the number of flows to be supported. The default is set to 1 million flows. This can be made as a configurable option as a next step. 4. Whenever a SPD rule is added/deleted by the control plane, the flow cache entries will be completely deleted (reset) in the control plane. The assumption here is that SPD rule add/del is not a frequent operation from control plane. Flow cache reset is done, by putting the data plane in fall back mode, to bypass flow cache and do linear search till the SPD rule add/delete operation is complete. Once the rule is successfully added/deleted, the data plane will be allowed to make use of the flow cache. The flow cache will be reset only after flushing out the inflight packets from all the worker cores using vlib_worker_wait_one_loop(). Details about bihash usage: 1. A new bihash template (16_8) is added to support IPv4 5 tuple. BIHASH_KVP_PER_PAGE and BIHASH_KVP_AT_BUCKET_LEVEL are set to 1 in the new template. It means only one KVP is supported per bucket. 2. Collision handling is avoided by calling BV (clib_bihash_add_or_overwrite_stale) function. Through the stale callback function pointer, the KVP entry will be overwritten during collision. 3. Flow cache reset is done using BV (clib_bihash_foreach_key_value_pair) function. Through the callback function pointer, the KVP value is reset to ~0ULL. MRR performance numbers with 1 core, 1 ESP Tunnel, null-encrypt, 64B for different SPD policy matching indices: SPD Policy index : 1 10 100 1000 Throughput : MPPS/MPPS MPPS/MPPS MPPS/MPPS KPPS/MPPS (Baseline/Optimized) ARM Neoverse N1 : 5.2/4.84 4.55/4.84 2.11/4.84 329.5/4.84 ARM TX2 : 2.81/2.6 2.51/2.6 1.27/2.6 176.62/2.6 INTEL SKX : 4.93/4.48 4.29/4.46 2.05/4.48 336.79/4.47 Next Steps: Following can be made as a configurable option through startup conf at IPSec level: 1. Enable/Disable Flow cache. 2. Bihash configuration like number of buckets and memory size. 3. Dual/Quad loop unroll can be applied around bihash to further improve the performance. 4. The same flow cache logic can be applied for IPv6 as well as in IPSec inbound direction. A deeper and wider flow cache using bihash_40_8 can replace existing bihash_16_8, to make it common for both IPv4 and IPv6 in both outbound and inbound directions. Following changes are made based on the review comments: 1. ON/OFF flow cache through startup conf. Default: OFF 2. Flow cache stale entry detection using epoch counter. 3. Avoid host order endianness conversion during flow cache lookup. 4. Move IPSec startup conf to a common file. 5. Added SPD flow cache unit test case 6. Replaced bihash with vectors to implement flow cache. 7. ipsec_add_del_policy API is not mpsafe. Cleaned up inflight packets check in control plane. Type: improvement Signed-off-by: mgovind <govindarajan.Mohandoss@arm.com> Signed-off-by: Zachary Leaf <zachary.leaf@arm.com> Tested-by: Jieqiang Wang <jieqiang.wang@arm.com> Change-Id: I62b4d6625fbc6caf292427a5d2046aa5672b2006
2021-03-19 19:20:49 +00:00
# store a copy of the packet in the packet info
info.data = p.copy()
# append the packet to the list
packets.append(p)
# return the created packet list
return packets
def verify_capture(self, src_if, dst_if, capture):
packet_info = None
for packet in capture:
try:
ip = packet[IP]
udp = packet[UDP]
# convert the payload to packet info object
payload_info = self.payload_to_info(packet)
# make sure the indexes match
self.assert_equal(
payload_info.src, src_if.sw_if_index, "source sw_if_index"
)
self.assert_equal(
payload_info.dst, dst_if.sw_if_index, "destination sw_if_index"
)
ipsec: Performance improvement of ipsec4_output_node using flow cache Adding flow cache support to improve outbound IPv4/IPSec SPD lookup performance. Details about flow cache: Mechanism: 1. First packet of a flow will undergo linear search in SPD table. Once a policy match is found, a new entry will be added into the flow cache. From 2nd packet onwards, the policy lookup will happen in flow cache. 2. The flow cache is implemented using bihash without collision handling. This will avoid the logic to age out or recycle the old flows in flow cache. Whenever a collision occurs, old entry will be overwritten by the new entry. Worst case is when all the 256 packets in a batch result in collision and fall back to linear search. Average and best case will be O(1). 3. The size of flow cache is fixed and decided based on the number of flows to be supported. The default is set to 1 million flows. This can be made as a configurable option as a next step. 4. Whenever a SPD rule is added/deleted by the control plane, the flow cache entries will be completely deleted (reset) in the control plane. The assumption here is that SPD rule add/del is not a frequent operation from control plane. Flow cache reset is done, by putting the data plane in fall back mode, to bypass flow cache and do linear search till the SPD rule add/delete operation is complete. Once the rule is successfully added/deleted, the data plane will be allowed to make use of the flow cache. The flow cache will be reset only after flushing out the inflight packets from all the worker cores using vlib_worker_wait_one_loop(). Details about bihash usage: 1. A new bihash template (16_8) is added to support IPv4 5 tuple. BIHASH_KVP_PER_PAGE and BIHASH_KVP_AT_BUCKET_LEVEL are set to 1 in the new template. It means only one KVP is supported per bucket. 2. Collision handling is avoided by calling BV (clib_bihash_add_or_overwrite_stale) function. Through the stale callback function pointer, the KVP entry will be overwritten during collision. 3. Flow cache reset is done using BV (clib_bihash_foreach_key_value_pair) function. Through the callback function pointer, the KVP value is reset to ~0ULL. MRR performance numbers with 1 core, 1 ESP Tunnel, null-encrypt, 64B for different SPD policy matching indices: SPD Policy index : 1 10 100 1000 Throughput : MPPS/MPPS MPPS/MPPS MPPS/MPPS KPPS/MPPS (Baseline/Optimized) ARM Neoverse N1 : 5.2/4.84 4.55/4.84 2.11/4.84 329.5/4.84 ARM TX2 : 2.81/2.6 2.51/2.6 1.27/2.6 176.62/2.6 INTEL SKX : 4.93/4.48 4.29/4.46 2.05/4.48 336.79/4.47 Next Steps: Following can be made as a configurable option through startup conf at IPSec level: 1. Enable/Disable Flow cache. 2. Bihash configuration like number of buckets and memory size. 3. Dual/Quad loop unroll can be applied around bihash to further improve the performance. 4. The same flow cache logic can be applied for IPv6 as well as in IPSec inbound direction. A deeper and wider flow cache using bihash_40_8 can replace existing bihash_16_8, to make it common for both IPv4 and IPv6 in both outbound and inbound directions. Following changes are made based on the review comments: 1. ON/OFF flow cache through startup conf. Default: OFF 2. Flow cache stale entry detection using epoch counter. 3. Avoid host order endianness conversion during flow cache lookup. 4. Move IPSec startup conf to a common file. 5. Added SPD flow cache unit test case 6. Replaced bihash with vectors to implement flow cache. 7. ipsec_add_del_policy API is not mpsafe. Cleaned up inflight packets check in control plane. Type: improvement Signed-off-by: mgovind <govindarajan.Mohandoss@arm.com> Signed-off-by: Zachary Leaf <zachary.leaf@arm.com> Tested-by: Jieqiang Wang <jieqiang.wang@arm.com> Change-Id: I62b4d6625fbc6caf292427a5d2046aa5672b2006
2021-03-19 19:20:49 +00:00
packet_info = self.get_next_packet_info_for_interface2(
src_if.sw_if_index, dst_if.sw_if_index, packet_info
)
ipsec: Performance improvement of ipsec4_output_node using flow cache Adding flow cache support to improve outbound IPv4/IPSec SPD lookup performance. Details about flow cache: Mechanism: 1. First packet of a flow will undergo linear search in SPD table. Once a policy match is found, a new entry will be added into the flow cache. From 2nd packet onwards, the policy lookup will happen in flow cache. 2. The flow cache is implemented using bihash without collision handling. This will avoid the logic to age out or recycle the old flows in flow cache. Whenever a collision occurs, old entry will be overwritten by the new entry. Worst case is when all the 256 packets in a batch result in collision and fall back to linear search. Average and best case will be O(1). 3. The size of flow cache is fixed and decided based on the number of flows to be supported. The default is set to 1 million flows. This can be made as a configurable option as a next step. 4. Whenever a SPD rule is added/deleted by the control plane, the flow cache entries will be completely deleted (reset) in the control plane. The assumption here is that SPD rule add/del is not a frequent operation from control plane. Flow cache reset is done, by putting the data plane in fall back mode, to bypass flow cache and do linear search till the SPD rule add/delete operation is complete. Once the rule is successfully added/deleted, the data plane will be allowed to make use of the flow cache. The flow cache will be reset only after flushing out the inflight packets from all the worker cores using vlib_worker_wait_one_loop(). Details about bihash usage: 1. A new bihash template (16_8) is added to support IPv4 5 tuple. BIHASH_KVP_PER_PAGE and BIHASH_KVP_AT_BUCKET_LEVEL are set to 1 in the new template. It means only one KVP is supported per bucket. 2. Collision handling is avoided by calling BV (clib_bihash_add_or_overwrite_stale) function. Through the stale callback function pointer, the KVP entry will be overwritten during collision. 3. Flow cache reset is done using BV (clib_bihash_foreach_key_value_pair) function. Through the callback function pointer, the KVP value is reset to ~0ULL. MRR performance numbers with 1 core, 1 ESP Tunnel, null-encrypt, 64B for different SPD policy matching indices: SPD Policy index : 1 10 100 1000 Throughput : MPPS/MPPS MPPS/MPPS MPPS/MPPS KPPS/MPPS (Baseline/Optimized) ARM Neoverse N1 : 5.2/4.84 4.55/4.84 2.11/4.84 329.5/4.84 ARM TX2 : 2.81/2.6 2.51/2.6 1.27/2.6 176.62/2.6 INTEL SKX : 4.93/4.48 4.29/4.46 2.05/4.48 336.79/4.47 Next Steps: Following can be made as a configurable option through startup conf at IPSec level: 1. Enable/Disable Flow cache. 2. Bihash configuration like number of buckets and memory size. 3. Dual/Quad loop unroll can be applied around bihash to further improve the performance. 4. The same flow cache logic can be applied for IPv6 as well as in IPSec inbound direction. A deeper and wider flow cache using bihash_40_8 can replace existing bihash_16_8, to make it common for both IPv4 and IPv6 in both outbound and inbound directions. Following changes are made based on the review comments: 1. ON/OFF flow cache through startup conf. Default: OFF 2. Flow cache stale entry detection using epoch counter. 3. Avoid host order endianness conversion during flow cache lookup. 4. Move IPSec startup conf to a common file. 5. Added SPD flow cache unit test case 6. Replaced bihash with vectors to implement flow cache. 7. ipsec_add_del_policy API is not mpsafe. Cleaned up inflight packets check in control plane. Type: improvement Signed-off-by: mgovind <govindarajan.Mohandoss@arm.com> Signed-off-by: Zachary Leaf <zachary.leaf@arm.com> Tested-by: Jieqiang Wang <jieqiang.wang@arm.com> Change-Id: I62b4d6625fbc6caf292427a5d2046aa5672b2006
2021-03-19 19:20:49 +00:00
# make sure we didn't run out of saved packets
self.assertIsNotNone(packet_info)
self.assert_equal(
payload_info.index, packet_info.index, "packet info index"
)
ipsec: Performance improvement of ipsec4_output_node using flow cache Adding flow cache support to improve outbound IPv4/IPSec SPD lookup performance. Details about flow cache: Mechanism: 1. First packet of a flow will undergo linear search in SPD table. Once a policy match is found, a new entry will be added into the flow cache. From 2nd packet onwards, the policy lookup will happen in flow cache. 2. The flow cache is implemented using bihash without collision handling. This will avoid the logic to age out or recycle the old flows in flow cache. Whenever a collision occurs, old entry will be overwritten by the new entry. Worst case is when all the 256 packets in a batch result in collision and fall back to linear search. Average and best case will be O(1). 3. The size of flow cache is fixed and decided based on the number of flows to be supported. The default is set to 1 million flows. This can be made as a configurable option as a next step. 4. Whenever a SPD rule is added/deleted by the control plane, the flow cache entries will be completely deleted (reset) in the control plane. The assumption here is that SPD rule add/del is not a frequent operation from control plane. Flow cache reset is done, by putting the data plane in fall back mode, to bypass flow cache and do linear search till the SPD rule add/delete operation is complete. Once the rule is successfully added/deleted, the data plane will be allowed to make use of the flow cache. The flow cache will be reset only after flushing out the inflight packets from all the worker cores using vlib_worker_wait_one_loop(). Details about bihash usage: 1. A new bihash template (16_8) is added to support IPv4 5 tuple. BIHASH_KVP_PER_PAGE and BIHASH_KVP_AT_BUCKET_LEVEL are set to 1 in the new template. It means only one KVP is supported per bucket. 2. Collision handling is avoided by calling BV (clib_bihash_add_or_overwrite_stale) function. Through the stale callback function pointer, the KVP entry will be overwritten during collision. 3. Flow cache reset is done using BV (clib_bihash_foreach_key_value_pair) function. Through the callback function pointer, the KVP value is reset to ~0ULL. MRR performance numbers with 1 core, 1 ESP Tunnel, null-encrypt, 64B for different SPD policy matching indices: SPD Policy index : 1 10 100 1000 Throughput : MPPS/MPPS MPPS/MPPS MPPS/MPPS KPPS/MPPS (Baseline/Optimized) ARM Neoverse N1 : 5.2/4.84 4.55/4.84 2.11/4.84 329.5/4.84 ARM TX2 : 2.81/2.6 2.51/2.6 1.27/2.6 176.62/2.6 INTEL SKX : 4.93/4.48 4.29/4.46 2.05/4.48 336.79/4.47 Next Steps: Following can be made as a configurable option through startup conf at IPSec level: 1. Enable/Disable Flow cache. 2. Bihash configuration like number of buckets and memory size. 3. Dual/Quad loop unroll can be applied around bihash to further improve the performance. 4. The same flow cache logic can be applied for IPv6 as well as in IPSec inbound direction. A deeper and wider flow cache using bihash_40_8 can replace existing bihash_16_8, to make it common for both IPv4 and IPv6 in both outbound and inbound directions. Following changes are made based on the review comments: 1. ON/OFF flow cache through startup conf. Default: OFF 2. Flow cache stale entry detection using epoch counter. 3. Avoid host order endianness conversion during flow cache lookup. 4. Move IPSec startup conf to a common file. 5. Added SPD flow cache unit test case 6. Replaced bihash with vectors to implement flow cache. 7. ipsec_add_del_policy API is not mpsafe. Cleaned up inflight packets check in control plane. Type: improvement Signed-off-by: mgovind <govindarajan.Mohandoss@arm.com> Signed-off-by: Zachary Leaf <zachary.leaf@arm.com> Tested-by: Jieqiang Wang <jieqiang.wang@arm.com> Change-Id: I62b4d6625fbc6caf292427a5d2046aa5672b2006
2021-03-19 19:20:49 +00:00
saved_packet = packet_info.data # fetch the saved packet
# assert the values match
self.assert_equal(ip.src, saved_packet[IP].src, "IP source address")
ipsec: Performance improvement of ipsec4_output_node using flow cache Adding flow cache support to improve outbound IPv4/IPSec SPD lookup performance. Details about flow cache: Mechanism: 1. First packet of a flow will undergo linear search in SPD table. Once a policy match is found, a new entry will be added into the flow cache. From 2nd packet onwards, the policy lookup will happen in flow cache. 2. The flow cache is implemented using bihash without collision handling. This will avoid the logic to age out or recycle the old flows in flow cache. Whenever a collision occurs, old entry will be overwritten by the new entry. Worst case is when all the 256 packets in a batch result in collision and fall back to linear search. Average and best case will be O(1). 3. The size of flow cache is fixed and decided based on the number of flows to be supported. The default is set to 1 million flows. This can be made as a configurable option as a next step. 4. Whenever a SPD rule is added/deleted by the control plane, the flow cache entries will be completely deleted (reset) in the control plane. The assumption here is that SPD rule add/del is not a frequent operation from control plane. Flow cache reset is done, by putting the data plane in fall back mode, to bypass flow cache and do linear search till the SPD rule add/delete operation is complete. Once the rule is successfully added/deleted, the data plane will be allowed to make use of the flow cache. The flow cache will be reset only after flushing out the inflight packets from all the worker cores using vlib_worker_wait_one_loop(). Details about bihash usage: 1. A new bihash template (16_8) is added to support IPv4 5 tuple. BIHASH_KVP_PER_PAGE and BIHASH_KVP_AT_BUCKET_LEVEL are set to 1 in the new template. It means only one KVP is supported per bucket. 2. Collision handling is avoided by calling BV (clib_bihash_add_or_overwrite_stale) function. Through the stale callback function pointer, the KVP entry will be overwritten during collision. 3. Flow cache reset is done using BV (clib_bihash_foreach_key_value_pair) function. Through the callback function pointer, the KVP value is reset to ~0ULL. MRR performance numbers with 1 core, 1 ESP Tunnel, null-encrypt, 64B for different SPD policy matching indices: SPD Policy index : 1 10 100 1000 Throughput : MPPS/MPPS MPPS/MPPS MPPS/MPPS KPPS/MPPS (Baseline/Optimized) ARM Neoverse N1 : 5.2/4.84 4.55/4.84 2.11/4.84 329.5/4.84 ARM TX2 : 2.81/2.6 2.51/2.6 1.27/2.6 176.62/2.6 INTEL SKX : 4.93/4.48 4.29/4.46 2.05/4.48 336.79/4.47 Next Steps: Following can be made as a configurable option through startup conf at IPSec level: 1. Enable/Disable Flow cache. 2. Bihash configuration like number of buckets and memory size. 3. Dual/Quad loop unroll can be applied around bihash to further improve the performance. 4. The same flow cache logic can be applied for IPv6 as well as in IPSec inbound direction. A deeper and wider flow cache using bihash_40_8 can replace existing bihash_16_8, to make it common for both IPv4 and IPv6 in both outbound and inbound directions. Following changes are made based on the review comments: 1. ON/OFF flow cache through startup conf. Default: OFF 2. Flow cache stale entry detection using epoch counter. 3. Avoid host order endianness conversion during flow cache lookup. 4. Move IPSec startup conf to a common file. 5. Added SPD flow cache unit test case 6. Replaced bihash with vectors to implement flow cache. 7. ipsec_add_del_policy API is not mpsafe. Cleaned up inflight packets check in control plane. Type: improvement Signed-off-by: mgovind <govindarajan.Mohandoss@arm.com> Signed-off-by: Zachary Leaf <zachary.leaf@arm.com> Tested-by: Jieqiang Wang <jieqiang.wang@arm.com> Change-Id: I62b4d6625fbc6caf292427a5d2046aa5672b2006
2021-03-19 19:20:49 +00:00
# ... more assertions here
self.assert_equal(udp.sport, saved_packet[UDP].sport, "UDP source port")
ipsec: Performance improvement of ipsec4_output_node using flow cache Adding flow cache support to improve outbound IPv4/IPSec SPD lookup performance. Details about flow cache: Mechanism: 1. First packet of a flow will undergo linear search in SPD table. Once a policy match is found, a new entry will be added into the flow cache. From 2nd packet onwards, the policy lookup will happen in flow cache. 2. The flow cache is implemented using bihash without collision handling. This will avoid the logic to age out or recycle the old flows in flow cache. Whenever a collision occurs, old entry will be overwritten by the new entry. Worst case is when all the 256 packets in a batch result in collision and fall back to linear search. Average and best case will be O(1). 3. The size of flow cache is fixed and decided based on the number of flows to be supported. The default is set to 1 million flows. This can be made as a configurable option as a next step. 4. Whenever a SPD rule is added/deleted by the control plane, the flow cache entries will be completely deleted (reset) in the control plane. The assumption here is that SPD rule add/del is not a frequent operation from control plane. Flow cache reset is done, by putting the data plane in fall back mode, to bypass flow cache and do linear search till the SPD rule add/delete operation is complete. Once the rule is successfully added/deleted, the data plane will be allowed to make use of the flow cache. The flow cache will be reset only after flushing out the inflight packets from all the worker cores using vlib_worker_wait_one_loop(). Details about bihash usage: 1. A new bihash template (16_8) is added to support IPv4 5 tuple. BIHASH_KVP_PER_PAGE and BIHASH_KVP_AT_BUCKET_LEVEL are set to 1 in the new template. It means only one KVP is supported per bucket. 2. Collision handling is avoided by calling BV (clib_bihash_add_or_overwrite_stale) function. Through the stale callback function pointer, the KVP entry will be overwritten during collision. 3. Flow cache reset is done using BV (clib_bihash_foreach_key_value_pair) function. Through the callback function pointer, the KVP value is reset to ~0ULL. MRR performance numbers with 1 core, 1 ESP Tunnel, null-encrypt, 64B for different SPD policy matching indices: SPD Policy index : 1 10 100 1000 Throughput : MPPS/MPPS MPPS/MPPS MPPS/MPPS KPPS/MPPS (Baseline/Optimized) ARM Neoverse N1 : 5.2/4.84 4.55/4.84 2.11/4.84 329.5/4.84 ARM TX2 : 2.81/2.6 2.51/2.6 1.27/2.6 176.62/2.6 INTEL SKX : 4.93/4.48 4.29/4.46 2.05/4.48 336.79/4.47 Next Steps: Following can be made as a configurable option through startup conf at IPSec level: 1. Enable/Disable Flow cache. 2. Bihash configuration like number of buckets and memory size. 3. Dual/Quad loop unroll can be applied around bihash to further improve the performance. 4. The same flow cache logic can be applied for IPv6 as well as in IPSec inbound direction. A deeper and wider flow cache using bihash_40_8 can replace existing bihash_16_8, to make it common for both IPv4 and IPv6 in both outbound and inbound directions. Following changes are made based on the review comments: 1. ON/OFF flow cache through startup conf. Default: OFF 2. Flow cache stale entry detection using epoch counter. 3. Avoid host order endianness conversion during flow cache lookup. 4. Move IPSec startup conf to a common file. 5. Added SPD flow cache unit test case 6. Replaced bihash with vectors to implement flow cache. 7. ipsec_add_del_policy API is not mpsafe. Cleaned up inflight packets check in control plane. Type: improvement Signed-off-by: mgovind <govindarajan.Mohandoss@arm.com> Signed-off-by: Zachary Leaf <zachary.leaf@arm.com> Tested-by: Jieqiang Wang <jieqiang.wang@arm.com> Change-Id: I62b4d6625fbc6caf292427a5d2046aa5672b2006
2021-03-19 19:20:49 +00:00
except Exception as e:
self.logger.error(ppp("Unexpected or invalid packet:", packet))
ipsec: Performance improvement of ipsec4_output_node using flow cache Adding flow cache support to improve outbound IPv4/IPSec SPD lookup performance. Details about flow cache: Mechanism: 1. First packet of a flow will undergo linear search in SPD table. Once a policy match is found, a new entry will be added into the flow cache. From 2nd packet onwards, the policy lookup will happen in flow cache. 2. The flow cache is implemented using bihash without collision handling. This will avoid the logic to age out or recycle the old flows in flow cache. Whenever a collision occurs, old entry will be overwritten by the new entry. Worst case is when all the 256 packets in a batch result in collision and fall back to linear search. Average and best case will be O(1). 3. The size of flow cache is fixed and decided based on the number of flows to be supported. The default is set to 1 million flows. This can be made as a configurable option as a next step. 4. Whenever a SPD rule is added/deleted by the control plane, the flow cache entries will be completely deleted (reset) in the control plane. The assumption here is that SPD rule add/del is not a frequent operation from control plane. Flow cache reset is done, by putting the data plane in fall back mode, to bypass flow cache and do linear search till the SPD rule add/delete operation is complete. Once the rule is successfully added/deleted, the data plane will be allowed to make use of the flow cache. The flow cache will be reset only after flushing out the inflight packets from all the worker cores using vlib_worker_wait_one_loop(). Details about bihash usage: 1. A new bihash template (16_8) is added to support IPv4 5 tuple. BIHASH_KVP_PER_PAGE and BIHASH_KVP_AT_BUCKET_LEVEL are set to 1 in the new template. It means only one KVP is supported per bucket. 2. Collision handling is avoided by calling BV (clib_bihash_add_or_overwrite_stale) function. Through the stale callback function pointer, the KVP entry will be overwritten during collision. 3. Flow cache reset is done using BV (clib_bihash_foreach_key_value_pair) function. Through the callback function pointer, the KVP value is reset to ~0ULL. MRR performance numbers with 1 core, 1 ESP Tunnel, null-encrypt, 64B for different SPD policy matching indices: SPD Policy index : 1 10 100 1000 Throughput : MPPS/MPPS MPPS/MPPS MPPS/MPPS KPPS/MPPS (Baseline/Optimized) ARM Neoverse N1 : 5.2/4.84 4.55/4.84 2.11/4.84 329.5/4.84 ARM TX2 : 2.81/2.6 2.51/2.6 1.27/2.6 176.62/2.6 INTEL SKX : 4.93/4.48 4.29/4.46 2.05/4.48 336.79/4.47 Next Steps: Following can be made as a configurable option through startup conf at IPSec level: 1. Enable/Disable Flow cache. 2. Bihash configuration like number of buckets and memory size. 3. Dual/Quad loop unroll can be applied around bihash to further improve the performance. 4. The same flow cache logic can be applied for IPv6 as well as in IPSec inbound direction. A deeper and wider flow cache using bihash_40_8 can replace existing bihash_16_8, to make it common for both IPv4 and IPv6 in both outbound and inbound directions. Following changes are made based on the review comments: 1. ON/OFF flow cache through startup conf. Default: OFF 2. Flow cache stale entry detection using epoch counter. 3. Avoid host order endianness conversion during flow cache lookup. 4. Move IPSec startup conf to a common file. 5. Added SPD flow cache unit test case 6. Replaced bihash with vectors to implement flow cache. 7. ipsec_add_del_policy API is not mpsafe. Cleaned up inflight packets check in control plane. Type: improvement Signed-off-by: mgovind <govindarajan.Mohandoss@arm.com> Signed-off-by: Zachary Leaf <zachary.leaf@arm.com> Tested-by: Jieqiang Wang <jieqiang.wang@arm.com> Change-Id: I62b4d6625fbc6caf292427a5d2046aa5672b2006
2021-03-19 19:20:49 +00:00
raise
remaining_packet = self.get_next_packet_info_for_interface2(
src_if.sw_if_index, dst_if.sw_if_index, packet_info
)
self.assertIsNone(
remaining_packet,
"Interface %s: Packet expected from interface "
"%s didn't arrive" % (dst_if.name, src_if.name),
)
ipsec: Performance improvement of ipsec4_output_node using flow cache Adding flow cache support to improve outbound IPv4/IPSec SPD lookup performance. Details about flow cache: Mechanism: 1. First packet of a flow will undergo linear search in SPD table. Once a policy match is found, a new entry will be added into the flow cache. From 2nd packet onwards, the policy lookup will happen in flow cache. 2. The flow cache is implemented using bihash without collision handling. This will avoid the logic to age out or recycle the old flows in flow cache. Whenever a collision occurs, old entry will be overwritten by the new entry. Worst case is when all the 256 packets in a batch result in collision and fall back to linear search. Average and best case will be O(1). 3. The size of flow cache is fixed and decided based on the number of flows to be supported. The default is set to 1 million flows. This can be made as a configurable option as a next step. 4. Whenever a SPD rule is added/deleted by the control plane, the flow cache entries will be completely deleted (reset) in the control plane. The assumption here is that SPD rule add/del is not a frequent operation from control plane. Flow cache reset is done, by putting the data plane in fall back mode, to bypass flow cache and do linear search till the SPD rule add/delete operation is complete. Once the rule is successfully added/deleted, the data plane will be allowed to make use of the flow cache. The flow cache will be reset only after flushing out the inflight packets from all the worker cores using vlib_worker_wait_one_loop(). Details about bihash usage: 1. A new bihash template (16_8) is added to support IPv4 5 tuple. BIHASH_KVP_PER_PAGE and BIHASH_KVP_AT_BUCKET_LEVEL are set to 1 in the new template. It means only one KVP is supported per bucket. 2. Collision handling is avoided by calling BV (clib_bihash_add_or_overwrite_stale) function. Through the stale callback function pointer, the KVP entry will be overwritten during collision. 3. Flow cache reset is done using BV (clib_bihash_foreach_key_value_pair) function. Through the callback function pointer, the KVP value is reset to ~0ULL. MRR performance numbers with 1 core, 1 ESP Tunnel, null-encrypt, 64B for different SPD policy matching indices: SPD Policy index : 1 10 100 1000 Throughput : MPPS/MPPS MPPS/MPPS MPPS/MPPS KPPS/MPPS (Baseline/Optimized) ARM Neoverse N1 : 5.2/4.84 4.55/4.84 2.11/4.84 329.5/4.84 ARM TX2 : 2.81/2.6 2.51/2.6 1.27/2.6 176.62/2.6 INTEL SKX : 4.93/4.48 4.29/4.46 2.05/4.48 336.79/4.47 Next Steps: Following can be made as a configurable option through startup conf at IPSec level: 1. Enable/Disable Flow cache. 2. Bihash configuration like number of buckets and memory size. 3. Dual/Quad loop unroll can be applied around bihash to further improve the performance. 4. The same flow cache logic can be applied for IPv6 as well as in IPSec inbound direction. A deeper and wider flow cache using bihash_40_8 can replace existing bihash_16_8, to make it common for both IPv4 and IPv6 in both outbound and inbound directions. Following changes are made based on the review comments: 1. ON/OFF flow cache through startup conf. Default: OFF 2. Flow cache stale entry detection using epoch counter. 3. Avoid host order endianness conversion during flow cache lookup. 4. Move IPSec startup conf to a common file. 5. Added SPD flow cache unit test case 6. Replaced bihash with vectors to implement flow cache. 7. ipsec_add_del_policy API is not mpsafe. Cleaned up inflight packets check in control plane. Type: improvement Signed-off-by: mgovind <govindarajan.Mohandoss@arm.com> Signed-off-by: Zachary Leaf <zachary.leaf@arm.com> Tested-by: Jieqiang Wang <jieqiang.wang@arm.com> Change-Id: I62b4d6625fbc6caf292427a5d2046aa5672b2006
2021-03-19 19:20:49 +00:00
def verify_policy_match(self, pkt_count, spdEntry):
self.logger.info("XXXX %s %s", str(spdEntry), str(spdEntry.get_stats()))
matched_pkts = spdEntry.get_stats().get("packets")
self.logger.info("Policy %s matched: %d pkts", str(spdEntry), matched_pkts)
ipsec: Performance improvement of ipsec4_output_node using flow cache Adding flow cache support to improve outbound IPv4/IPSec SPD lookup performance. Details about flow cache: Mechanism: 1. First packet of a flow will undergo linear search in SPD table. Once a policy match is found, a new entry will be added into the flow cache. From 2nd packet onwards, the policy lookup will happen in flow cache. 2. The flow cache is implemented using bihash without collision handling. This will avoid the logic to age out or recycle the old flows in flow cache. Whenever a collision occurs, old entry will be overwritten by the new entry. Worst case is when all the 256 packets in a batch result in collision and fall back to linear search. Average and best case will be O(1). 3. The size of flow cache is fixed and decided based on the number of flows to be supported. The default is set to 1 million flows. This can be made as a configurable option as a next step. 4. Whenever a SPD rule is added/deleted by the control plane, the flow cache entries will be completely deleted (reset) in the control plane. The assumption here is that SPD rule add/del is not a frequent operation from control plane. Flow cache reset is done, by putting the data plane in fall back mode, to bypass flow cache and do linear search till the SPD rule add/delete operation is complete. Once the rule is successfully added/deleted, the data plane will be allowed to make use of the flow cache. The flow cache will be reset only after flushing out the inflight packets from all the worker cores using vlib_worker_wait_one_loop(). Details about bihash usage: 1. A new bihash template (16_8) is added to support IPv4 5 tuple. BIHASH_KVP_PER_PAGE and BIHASH_KVP_AT_BUCKET_LEVEL are set to 1 in the new template. It means only one KVP is supported per bucket. 2. Collision handling is avoided by calling BV (clib_bihash_add_or_overwrite_stale) function. Through the stale callback function pointer, the KVP entry will be overwritten during collision. 3. Flow cache reset is done using BV (clib_bihash_foreach_key_value_pair) function. Through the callback function pointer, the KVP value is reset to ~0ULL. MRR performance numbers with 1 core, 1 ESP Tunnel, null-encrypt, 64B for different SPD policy matching indices: SPD Policy index : 1 10 100 1000 Throughput : MPPS/MPPS MPPS/MPPS MPPS/MPPS KPPS/MPPS (Baseline/Optimized) ARM Neoverse N1 : 5.2/4.84 4.55/4.84 2.11/4.84 329.5/4.84 ARM TX2 : 2.81/2.6 2.51/2.6 1.27/2.6 176.62/2.6 INTEL SKX : 4.93/4.48 4.29/4.46 2.05/4.48 336.79/4.47 Next Steps: Following can be made as a configurable option through startup conf at IPSec level: 1. Enable/Disable Flow cache. 2. Bihash configuration like number of buckets and memory size. 3. Dual/Quad loop unroll can be applied around bihash to further improve the performance. 4. The same flow cache logic can be applied for IPv6 as well as in IPSec inbound direction. A deeper and wider flow cache using bihash_40_8 can replace existing bihash_16_8, to make it common for both IPv4 and IPv6 in both outbound and inbound directions. Following changes are made based on the review comments: 1. ON/OFF flow cache through startup conf. Default: OFF 2. Flow cache stale entry detection using epoch counter. 3. Avoid host order endianness conversion during flow cache lookup. 4. Move IPSec startup conf to a common file. 5. Added SPD flow cache unit test case 6. Replaced bihash with vectors to implement flow cache. 7. ipsec_add_del_policy API is not mpsafe. Cleaned up inflight packets check in control plane. Type: improvement Signed-off-by: mgovind <govindarajan.Mohandoss@arm.com> Signed-off-by: Zachary Leaf <zachary.leaf@arm.com> Tested-by: Jieqiang Wang <jieqiang.wang@arm.com> Change-Id: I62b4d6625fbc6caf292427a5d2046aa5672b2006
2021-03-19 19:20:49 +00:00
self.assert_equal(pkt_count, matched_pkts)
# Method verify_l3_l4_capture() will verify network and transport layer
# fields of the packet sa.encrypt() gives interface number garbadge.
# thus interface validation get failed (scapy bug?). However our intent
# is to verify IP layer and above and that is covered.
def verify_l3_l4_capture(
self, src_if, dst_if, capture, tcp_port_in=1234, udp_port_in=5678
):
for packet in capture:
try:
self.assert_packet_checksums_valid(packet)
self.assert_equal(
packet[IP].src,
src_if.remote_ip4,
"decrypted packet source address",
)
self.assert_equal(
packet[IP].dst,
dst_if.remote_ip4,
"decrypted packet destination address",
)
if packet.haslayer(TCP):
self.assertFalse(
packet.haslayer(UDP),
"unexpected UDP header in decrypted packet",
)
elif packet.haslayer(UDP):
if packet[UDP].payload:
self.assertFalse(
packet[UDP][1].haslayer(UDP),
"unexpected UDP header in decrypted packet",
)
else:
self.assertFalse(
packet.haslayer(UDP),
"unexpected UDP header in decrypted packet",
)
self.assert_equal(
packet[ICMP].id, self.icmp_id_in, "decrypted packet ICMP ID"
)
except Exception:
self.logger.error(ppp("Unexpected or invalid plain packet:", packet))
raise
class SpdFlowCacheTemplate(IPSecIPv4Fwd):
@classmethod
def setUpConstants(cls):
super(SpdFlowCacheTemplate, cls).setUpConstants()
# Override this method with required cmdline parameters e.g.
# cls.vpp_cmdline.extend(["ipsec", "{",
# "ipv4-outbound-spd-flow-cache on",
# "}"])
# cls.logger.info("VPP modified cmdline is %s" % " "
# .join(cls.vpp_cmdline))
def setUp(self):
super(SpdFlowCacheTemplate, self).setUp()
def tearDown(self):
super(SpdFlowCacheTemplate, self).tearDown()
ipsec: perf improvement of ipsec4_input_node using flow cache Adding flow cache support to improve inbound IPv4/IPSec Security Policy Database (SPD) lookup performance. By enabling the flow cache in startup conf, this replaces a linear O(N) SPD search, with an O(1) hash table search. This patch is the ipsec4_input_node counterpart to https://gerrit.fd.io/r/c/vpp/+/31694, and shares much of the same code, theory and mechanism of action. Details about the flow cache: Mechanism: 1. First packet of a flow will undergo linear search in SPD table. Once a policy match is found, a new entry will be added into the flow cache. From 2nd packet onwards, the policy lookup will happen in flow cache. 2. The flow cache is implemented using a hash table without collision handling. This will avoid the logic to age out or recycle the old flows in flow cache. Whenever a collision occurs, the old entry will be overwritten by the new entry. Worst case is when all the 256 packets in a batch result in collision, falling back to linear search. Average and best case will be O(1). 3. The size of flow cache is fixed and decided based on the number of flows to be supported. The default is set to 1 million flows, but is configurable by a startup.conf option. 4. Whenever a SPD rule is added/deleted by the control plane, all current flow cache entries will be invalidated. As the SPD API is not mp-safe, the data plane will wait for the control plane operation to complete. Cache invalidation is via an epoch counter that is incremented on policy add/del and stored with each entry in the flow cache. If the epoch counter in the flow cache does not match the current count, the entry is considered stale, and we fall back to linear search. The following configurable options are available through startup conf under the ipsec{} entry: 1. ipv4-inbound-spd-flow-cache on/off - enable SPD flow cache (default off) 2. ipv4-inbound-spd-hash-buckets %d - set number of hash buckets (default 4,194,304: ~1 million flows with 25% load factor) Performance with 1 core, 1 ESP Tunnel, null-decrypt then bypass, 94B (null encrypted packet) for different SPD policy matching indices: SPD Policy index : 2 10 100 1000 Throughput : Mbps/Mbps Mbps/Mbps Mbps/Mbps Mbps/Mbps (Baseline/Optimized) ARM TX2 : 300/290 230/290 70/290 8.5/290 Type: improvement Signed-off-by: Zachary Leaf <zachary.leaf@arm.com> Signed-off-by: mgovind <govindarajan.Mohandoss@arm.com> Tested-by: Jieqiang Wang <jieqiang.wang@arm.com> Change-Id: I8be2ad4715accbb335c38cd933904119db75827b
2021-06-25 08:11:15 -05:00
def get_spd_flow_cache_entries(self, outbound):
"""'show ipsec spd' output:
ipsec: perf improvement of ipsec4_input_node using flow cache Adding flow cache support to improve inbound IPv4/IPSec Security Policy Database (SPD) lookup performance. By enabling the flow cache in startup conf, this replaces a linear O(N) SPD search, with an O(1) hash table search. This patch is the ipsec4_input_node counterpart to https://gerrit.fd.io/r/c/vpp/+/31694, and shares much of the same code, theory and mechanism of action. Details about the flow cache: Mechanism: 1. First packet of a flow will undergo linear search in SPD table. Once a policy match is found, a new entry will be added into the flow cache. From 2nd packet onwards, the policy lookup will happen in flow cache. 2. The flow cache is implemented using a hash table without collision handling. This will avoid the logic to age out or recycle the old flows in flow cache. Whenever a collision occurs, the old entry will be overwritten by the new entry. Worst case is when all the 256 packets in a batch result in collision, falling back to linear search. Average and best case will be O(1). 3. The size of flow cache is fixed and decided based on the number of flows to be supported. The default is set to 1 million flows, but is configurable by a startup.conf option. 4. Whenever a SPD rule is added/deleted by the control plane, all current flow cache entries will be invalidated. As the SPD API is not mp-safe, the data plane will wait for the control plane operation to complete. Cache invalidation is via an epoch counter that is incremented on policy add/del and stored with each entry in the flow cache. If the epoch counter in the flow cache does not match the current count, the entry is considered stale, and we fall back to linear search. The following configurable options are available through startup conf under the ipsec{} entry: 1. ipv4-inbound-spd-flow-cache on/off - enable SPD flow cache (default off) 2. ipv4-inbound-spd-hash-buckets %d - set number of hash buckets (default 4,194,304: ~1 million flows with 25% load factor) Performance with 1 core, 1 ESP Tunnel, null-decrypt then bypass, 94B (null encrypted packet) for different SPD policy matching indices: SPD Policy index : 2 10 100 1000 Throughput : Mbps/Mbps Mbps/Mbps Mbps/Mbps Mbps/Mbps (Baseline/Optimized) ARM TX2 : 300/290 230/290 70/290 8.5/290 Type: improvement Signed-off-by: Zachary Leaf <zachary.leaf@arm.com> Signed-off-by: mgovind <govindarajan.Mohandoss@arm.com> Tested-by: Jieqiang Wang <jieqiang.wang@arm.com> Change-Id: I8be2ad4715accbb335c38cd933904119db75827b
2021-06-25 08:11:15 -05:00
ipv4-inbound-spd-flow-cache-entries: 0
ipv4-outbound-spd-flow-cache-entries: 0
ipsec: Performance improvement of ipsec4_output_node using flow cache Adding flow cache support to improve outbound IPv4/IPSec SPD lookup performance. Details about flow cache: Mechanism: 1. First packet of a flow will undergo linear search in SPD table. Once a policy match is found, a new entry will be added into the flow cache. From 2nd packet onwards, the policy lookup will happen in flow cache. 2. The flow cache is implemented using bihash without collision handling. This will avoid the logic to age out or recycle the old flows in flow cache. Whenever a collision occurs, old entry will be overwritten by the new entry. Worst case is when all the 256 packets in a batch result in collision and fall back to linear search. Average and best case will be O(1). 3. The size of flow cache is fixed and decided based on the number of flows to be supported. The default is set to 1 million flows. This can be made as a configurable option as a next step. 4. Whenever a SPD rule is added/deleted by the control plane, the flow cache entries will be completely deleted (reset) in the control plane. The assumption here is that SPD rule add/del is not a frequent operation from control plane. Flow cache reset is done, by putting the data plane in fall back mode, to bypass flow cache and do linear search till the SPD rule add/delete operation is complete. Once the rule is successfully added/deleted, the data plane will be allowed to make use of the flow cache. The flow cache will be reset only after flushing out the inflight packets from all the worker cores using vlib_worker_wait_one_loop(). Details about bihash usage: 1. A new bihash template (16_8) is added to support IPv4 5 tuple. BIHASH_KVP_PER_PAGE and BIHASH_KVP_AT_BUCKET_LEVEL are set to 1 in the new template. It means only one KVP is supported per bucket. 2. Collision handling is avoided by calling BV (clib_bihash_add_or_overwrite_stale) function. Through the stale callback function pointer, the KVP entry will be overwritten during collision. 3. Flow cache reset is done using BV (clib_bihash_foreach_key_value_pair) function. Through the callback function pointer, the KVP value is reset to ~0ULL. MRR performance numbers with 1 core, 1 ESP Tunnel, null-encrypt, 64B for different SPD policy matching indices: SPD Policy index : 1 10 100 1000 Throughput : MPPS/MPPS MPPS/MPPS MPPS/MPPS KPPS/MPPS (Baseline/Optimized) ARM Neoverse N1 : 5.2/4.84 4.55/4.84 2.11/4.84 329.5/4.84 ARM TX2 : 2.81/2.6 2.51/2.6 1.27/2.6 176.62/2.6 INTEL SKX : 4.93/4.48 4.29/4.46 2.05/4.48 336.79/4.47 Next Steps: Following can be made as a configurable option through startup conf at IPSec level: 1. Enable/Disable Flow cache. 2. Bihash configuration like number of buckets and memory size. 3. Dual/Quad loop unroll can be applied around bihash to further improve the performance. 4. The same flow cache logic can be applied for IPv6 as well as in IPSec inbound direction. A deeper and wider flow cache using bihash_40_8 can replace existing bihash_16_8, to make it common for both IPv4 and IPv6 in both outbound and inbound directions. Following changes are made based on the review comments: 1. ON/OFF flow cache through startup conf. Default: OFF 2. Flow cache stale entry detection using epoch counter. 3. Avoid host order endianness conversion during flow cache lookup. 4. Move IPSec startup conf to a common file. 5. Added SPD flow cache unit test case 6. Replaced bihash with vectors to implement flow cache. 7. ipsec_add_del_policy API is not mpsafe. Cleaned up inflight packets check in control plane. Type: improvement Signed-off-by: mgovind <govindarajan.Mohandoss@arm.com> Signed-off-by: Zachary Leaf <zachary.leaf@arm.com> Tested-by: Jieqiang Wang <jieqiang.wang@arm.com> Change-Id: I62b4d6625fbc6caf292427a5d2046aa5672b2006
2021-03-19 19:20:49 +00:00
"""
show_ipsec_reply = self.vapi.cli("show ipsec spd")
# match the relevant section of 'show ipsec spd' output
if outbound:
ipsec: perf improvement of ipsec4_input_node using flow cache Adding flow cache support to improve inbound IPv4/IPSec Security Policy Database (SPD) lookup performance. By enabling the flow cache in startup conf, this replaces a linear O(N) SPD search, with an O(1) hash table search. This patch is the ipsec4_input_node counterpart to https://gerrit.fd.io/r/c/vpp/+/31694, and shares much of the same code, theory and mechanism of action. Details about the flow cache: Mechanism: 1. First packet of a flow will undergo linear search in SPD table. Once a policy match is found, a new entry will be added into the flow cache. From 2nd packet onwards, the policy lookup will happen in flow cache. 2. The flow cache is implemented using a hash table without collision handling. This will avoid the logic to age out or recycle the old flows in flow cache. Whenever a collision occurs, the old entry will be overwritten by the new entry. Worst case is when all the 256 packets in a batch result in collision, falling back to linear search. Average and best case will be O(1). 3. The size of flow cache is fixed and decided based on the number of flows to be supported. The default is set to 1 million flows, but is configurable by a startup.conf option. 4. Whenever a SPD rule is added/deleted by the control plane, all current flow cache entries will be invalidated. As the SPD API is not mp-safe, the data plane will wait for the control plane operation to complete. Cache invalidation is via an epoch counter that is incremented on policy add/del and stored with each entry in the flow cache. If the epoch counter in the flow cache does not match the current count, the entry is considered stale, and we fall back to linear search. The following configurable options are available through startup conf under the ipsec{} entry: 1. ipv4-inbound-spd-flow-cache on/off - enable SPD flow cache (default off) 2. ipv4-inbound-spd-hash-buckets %d - set number of hash buckets (default 4,194,304: ~1 million flows with 25% load factor) Performance with 1 core, 1 ESP Tunnel, null-decrypt then bypass, 94B (null encrypted packet) for different SPD policy matching indices: SPD Policy index : 2 10 100 1000 Throughput : Mbps/Mbps Mbps/Mbps Mbps/Mbps Mbps/Mbps (Baseline/Optimized) ARM TX2 : 300/290 230/290 70/290 8.5/290 Type: improvement Signed-off-by: Zachary Leaf <zachary.leaf@arm.com> Signed-off-by: mgovind <govindarajan.Mohandoss@arm.com> Tested-by: Jieqiang Wang <jieqiang.wang@arm.com> Change-Id: I8be2ad4715accbb335c38cd933904119db75827b
2021-06-25 08:11:15 -05:00
regex_match = re.search(
"ipv4-outbound-spd-flow-cache-entries: (.*)",
show_ipsec_reply,
re.DOTALL,
)
ipsec: perf improvement of ipsec4_input_node using flow cache Adding flow cache support to improve inbound IPv4/IPSec Security Policy Database (SPD) lookup performance. By enabling the flow cache in startup conf, this replaces a linear O(N) SPD search, with an O(1) hash table search. This patch is the ipsec4_input_node counterpart to https://gerrit.fd.io/r/c/vpp/+/31694, and shares much of the same code, theory and mechanism of action. Details about the flow cache: Mechanism: 1. First packet of a flow will undergo linear search in SPD table. Once a policy match is found, a new entry will be added into the flow cache. From 2nd packet onwards, the policy lookup will happen in flow cache. 2. The flow cache is implemented using a hash table without collision handling. This will avoid the logic to age out or recycle the old flows in flow cache. Whenever a collision occurs, the old entry will be overwritten by the new entry. Worst case is when all the 256 packets in a batch result in collision, falling back to linear search. Average and best case will be O(1). 3. The size of flow cache is fixed and decided based on the number of flows to be supported. The default is set to 1 million flows, but is configurable by a startup.conf option. 4. Whenever a SPD rule is added/deleted by the control plane, all current flow cache entries will be invalidated. As the SPD API is not mp-safe, the data plane will wait for the control plane operation to complete. Cache invalidation is via an epoch counter that is incremented on policy add/del and stored with each entry in the flow cache. If the epoch counter in the flow cache does not match the current count, the entry is considered stale, and we fall back to linear search. The following configurable options are available through startup conf under the ipsec{} entry: 1. ipv4-inbound-spd-flow-cache on/off - enable SPD flow cache (default off) 2. ipv4-inbound-spd-hash-buckets %d - set number of hash buckets (default 4,194,304: ~1 million flows with 25% load factor) Performance with 1 core, 1 ESP Tunnel, null-decrypt then bypass, 94B (null encrypted packet) for different SPD policy matching indices: SPD Policy index : 2 10 100 1000 Throughput : Mbps/Mbps Mbps/Mbps Mbps/Mbps Mbps/Mbps (Baseline/Optimized) ARM TX2 : 300/290 230/290 70/290 8.5/290 Type: improvement Signed-off-by: Zachary Leaf <zachary.leaf@arm.com> Signed-off-by: mgovind <govindarajan.Mohandoss@arm.com> Tested-by: Jieqiang Wang <jieqiang.wang@arm.com> Change-Id: I8be2ad4715accbb335c38cd933904119db75827b
2021-06-25 08:11:15 -05:00
else:
regex_match = re.search(
"ipv4-inbound-spd-flow-cache-entries: (.*)", show_ipsec_reply, re.DOTALL
)
ipsec: Performance improvement of ipsec4_output_node using flow cache Adding flow cache support to improve outbound IPv4/IPSec SPD lookup performance. Details about flow cache: Mechanism: 1. First packet of a flow will undergo linear search in SPD table. Once a policy match is found, a new entry will be added into the flow cache. From 2nd packet onwards, the policy lookup will happen in flow cache. 2. The flow cache is implemented using bihash without collision handling. This will avoid the logic to age out or recycle the old flows in flow cache. Whenever a collision occurs, old entry will be overwritten by the new entry. Worst case is when all the 256 packets in a batch result in collision and fall back to linear search. Average and best case will be O(1). 3. The size of flow cache is fixed and decided based on the number of flows to be supported. The default is set to 1 million flows. This can be made as a configurable option as a next step. 4. Whenever a SPD rule is added/deleted by the control plane, the flow cache entries will be completely deleted (reset) in the control plane. The assumption here is that SPD rule add/del is not a frequent operation from control plane. Flow cache reset is done, by putting the data plane in fall back mode, to bypass flow cache and do linear search till the SPD rule add/delete operation is complete. Once the rule is successfully added/deleted, the data plane will be allowed to make use of the flow cache. The flow cache will be reset only after flushing out the inflight packets from all the worker cores using vlib_worker_wait_one_loop(). Details about bihash usage: 1. A new bihash template (16_8) is added to support IPv4 5 tuple. BIHASH_KVP_PER_PAGE and BIHASH_KVP_AT_BUCKET_LEVEL are set to 1 in the new template. It means only one KVP is supported per bucket. 2. Collision handling is avoided by calling BV (clib_bihash_add_or_overwrite_stale) function. Through the stale callback function pointer, the KVP entry will be overwritten during collision. 3. Flow cache reset is done using BV (clib_bihash_foreach_key_value_pair) function. Through the callback function pointer, the KVP value is reset to ~0ULL. MRR performance numbers with 1 core, 1 ESP Tunnel, null-encrypt, 64B for different SPD policy matching indices: SPD Policy index : 1 10 100 1000 Throughput : MPPS/MPPS MPPS/MPPS MPPS/MPPS KPPS/MPPS (Baseline/Optimized) ARM Neoverse N1 : 5.2/4.84 4.55/4.84 2.11/4.84 329.5/4.84 ARM TX2 : 2.81/2.6 2.51/2.6 1.27/2.6 176.62/2.6 INTEL SKX : 4.93/4.48 4.29/4.46 2.05/4.48 336.79/4.47 Next Steps: Following can be made as a configurable option through startup conf at IPSec level: 1. Enable/Disable Flow cache. 2. Bihash configuration like number of buckets and memory size. 3. Dual/Quad loop unroll can be applied around bihash to further improve the performance. 4. The same flow cache logic can be applied for IPv6 as well as in IPSec inbound direction. A deeper and wider flow cache using bihash_40_8 can replace existing bihash_16_8, to make it common for both IPv4 and IPv6 in both outbound and inbound directions. Following changes are made based on the review comments: 1. ON/OFF flow cache through startup conf. Default: OFF 2. Flow cache stale entry detection using epoch counter. 3. Avoid host order endianness conversion during flow cache lookup. 4. Move IPSec startup conf to a common file. 5. Added SPD flow cache unit test case 6. Replaced bihash with vectors to implement flow cache. 7. ipsec_add_del_policy API is not mpsafe. Cleaned up inflight packets check in control plane. Type: improvement Signed-off-by: mgovind <govindarajan.Mohandoss@arm.com> Signed-off-by: Zachary Leaf <zachary.leaf@arm.com> Tested-by: Jieqiang Wang <jieqiang.wang@arm.com> Change-Id: I62b4d6625fbc6caf292427a5d2046aa5672b2006
2021-03-19 19:20:49 +00:00
if regex_match is None:
raise Exception(
"Unable to find spd flow cache entries \
in 'show ipsec spd' CLI output - regex failed to match"
)
ipsec: Performance improvement of ipsec4_output_node using flow cache Adding flow cache support to improve outbound IPv4/IPSec SPD lookup performance. Details about flow cache: Mechanism: 1. First packet of a flow will undergo linear search in SPD table. Once a policy match is found, a new entry will be added into the flow cache. From 2nd packet onwards, the policy lookup will happen in flow cache. 2. The flow cache is implemented using bihash without collision handling. This will avoid the logic to age out or recycle the old flows in flow cache. Whenever a collision occurs, old entry will be overwritten by the new entry. Worst case is when all the 256 packets in a batch result in collision and fall back to linear search. Average and best case will be O(1). 3. The size of flow cache is fixed and decided based on the number of flows to be supported. The default is set to 1 million flows. This can be made as a configurable option as a next step. 4. Whenever a SPD rule is added/deleted by the control plane, the flow cache entries will be completely deleted (reset) in the control plane. The assumption here is that SPD rule add/del is not a frequent operation from control plane. Flow cache reset is done, by putting the data plane in fall back mode, to bypass flow cache and do linear search till the SPD rule add/delete operation is complete. Once the rule is successfully added/deleted, the data plane will be allowed to make use of the flow cache. The flow cache will be reset only after flushing out the inflight packets from all the worker cores using vlib_worker_wait_one_loop(). Details about bihash usage: 1. A new bihash template (16_8) is added to support IPv4 5 tuple. BIHASH_KVP_PER_PAGE and BIHASH_KVP_AT_BUCKET_LEVEL are set to 1 in the new template. It means only one KVP is supported per bucket. 2. Collision handling is avoided by calling BV (clib_bihash_add_or_overwrite_stale) function. Through the stale callback function pointer, the KVP entry will be overwritten during collision. 3. Flow cache reset is done using BV (clib_bihash_foreach_key_value_pair) function. Through the callback function pointer, the KVP value is reset to ~0ULL. MRR performance numbers with 1 core, 1 ESP Tunnel, null-encrypt, 64B for different SPD policy matching indices: SPD Policy index : 1 10 100 1000 Throughput : MPPS/MPPS MPPS/MPPS MPPS/MPPS KPPS/MPPS (Baseline/Optimized) ARM Neoverse N1 : 5.2/4.84 4.55/4.84 2.11/4.84 329.5/4.84 ARM TX2 : 2.81/2.6 2.51/2.6 1.27/2.6 176.62/2.6 INTEL SKX : 4.93/4.48 4.29/4.46 2.05/4.48 336.79/4.47 Next Steps: Following can be made as a configurable option through startup conf at IPSec level: 1. Enable/Disable Flow cache. 2. Bihash configuration like number of buckets and memory size. 3. Dual/Quad loop unroll can be applied around bihash to further improve the performance. 4. The same flow cache logic can be applied for IPv6 as well as in IPSec inbound direction. A deeper and wider flow cache using bihash_40_8 can replace existing bihash_16_8, to make it common for both IPv4 and IPv6 in both outbound and inbound directions. Following changes are made based on the review comments: 1. ON/OFF flow cache through startup conf. Default: OFF 2. Flow cache stale entry detection using epoch counter. 3. Avoid host order endianness conversion during flow cache lookup. 4. Move IPSec startup conf to a common file. 5. Added SPD flow cache unit test case 6. Replaced bihash with vectors to implement flow cache. 7. ipsec_add_del_policy API is not mpsafe. Cleaned up inflight packets check in control plane. Type: improvement Signed-off-by: mgovind <govindarajan.Mohandoss@arm.com> Signed-off-by: Zachary Leaf <zachary.leaf@arm.com> Tested-by: Jieqiang Wang <jieqiang.wang@arm.com> Change-Id: I62b4d6625fbc6caf292427a5d2046aa5672b2006
2021-03-19 19:20:49 +00:00
else:
try:
num_entries = int(regex_match.group(1))
except ValueError:
raise Exception(
"Unable to get spd flow cache entries \
from 'show ipsec spd' string: %s",
regex_match.group(0),
)
ipsec: Performance improvement of ipsec4_output_node using flow cache Adding flow cache support to improve outbound IPv4/IPSec SPD lookup performance. Details about flow cache: Mechanism: 1. First packet of a flow will undergo linear search in SPD table. Once a policy match is found, a new entry will be added into the flow cache. From 2nd packet onwards, the policy lookup will happen in flow cache. 2. The flow cache is implemented using bihash without collision handling. This will avoid the logic to age out or recycle the old flows in flow cache. Whenever a collision occurs, old entry will be overwritten by the new entry. Worst case is when all the 256 packets in a batch result in collision and fall back to linear search. Average and best case will be O(1). 3. The size of flow cache is fixed and decided based on the number of flows to be supported. The default is set to 1 million flows. This can be made as a configurable option as a next step. 4. Whenever a SPD rule is added/deleted by the control plane, the flow cache entries will be completely deleted (reset) in the control plane. The assumption here is that SPD rule add/del is not a frequent operation from control plane. Flow cache reset is done, by putting the data plane in fall back mode, to bypass flow cache and do linear search till the SPD rule add/delete operation is complete. Once the rule is successfully added/deleted, the data plane will be allowed to make use of the flow cache. The flow cache will be reset only after flushing out the inflight packets from all the worker cores using vlib_worker_wait_one_loop(). Details about bihash usage: 1. A new bihash template (16_8) is added to support IPv4 5 tuple. BIHASH_KVP_PER_PAGE and BIHASH_KVP_AT_BUCKET_LEVEL are set to 1 in the new template. It means only one KVP is supported per bucket. 2. Collision handling is avoided by calling BV (clib_bihash_add_or_overwrite_stale) function. Through the stale callback function pointer, the KVP entry will be overwritten during collision. 3. Flow cache reset is done using BV (clib_bihash_foreach_key_value_pair) function. Through the callback function pointer, the KVP value is reset to ~0ULL. MRR performance numbers with 1 core, 1 ESP Tunnel, null-encrypt, 64B for different SPD policy matching indices: SPD Policy index : 1 10 100 1000 Throughput : MPPS/MPPS MPPS/MPPS MPPS/MPPS KPPS/MPPS (Baseline/Optimized) ARM Neoverse N1 : 5.2/4.84 4.55/4.84 2.11/4.84 329.5/4.84 ARM TX2 : 2.81/2.6 2.51/2.6 1.27/2.6 176.62/2.6 INTEL SKX : 4.93/4.48 4.29/4.46 2.05/4.48 336.79/4.47 Next Steps: Following can be made as a configurable option through startup conf at IPSec level: 1. Enable/Disable Flow cache. 2. Bihash configuration like number of buckets and memory size. 3. Dual/Quad loop unroll can be applied around bihash to further improve the performance. 4. The same flow cache logic can be applied for IPv6 as well as in IPSec inbound direction. A deeper and wider flow cache using bihash_40_8 can replace existing bihash_16_8, to make it common for both IPv4 and IPv6 in both outbound and inbound directions. Following changes are made based on the review comments: 1. ON/OFF flow cache through startup conf. Default: OFF 2. Flow cache stale entry detection using epoch counter. 3. Avoid host order endianness conversion during flow cache lookup. 4. Move IPSec startup conf to a common file. 5. Added SPD flow cache unit test case 6. Replaced bihash with vectors to implement flow cache. 7. ipsec_add_del_policy API is not mpsafe. Cleaned up inflight packets check in control plane. Type: improvement Signed-off-by: mgovind <govindarajan.Mohandoss@arm.com> Signed-off-by: Zachary Leaf <zachary.leaf@arm.com> Tested-by: Jieqiang Wang <jieqiang.wang@arm.com> Change-Id: I62b4d6625fbc6caf292427a5d2046aa5672b2006
2021-03-19 19:20:49 +00:00
self.logger.info("%s", regex_match.group(0))
return num_entries
def verify_num_outbound_flow_cache_entries(self, expected_elements):
self.assertEqual(
self.get_spd_flow_cache_entries(outbound=True), expected_elements
)
ipsec: perf improvement of ipsec4_input_node using flow cache Adding flow cache support to improve inbound IPv4/IPSec Security Policy Database (SPD) lookup performance. By enabling the flow cache in startup conf, this replaces a linear O(N) SPD search, with an O(1) hash table search. This patch is the ipsec4_input_node counterpart to https://gerrit.fd.io/r/c/vpp/+/31694, and shares much of the same code, theory and mechanism of action. Details about the flow cache: Mechanism: 1. First packet of a flow will undergo linear search in SPD table. Once a policy match is found, a new entry will be added into the flow cache. From 2nd packet onwards, the policy lookup will happen in flow cache. 2. The flow cache is implemented using a hash table without collision handling. This will avoid the logic to age out or recycle the old flows in flow cache. Whenever a collision occurs, the old entry will be overwritten by the new entry. Worst case is when all the 256 packets in a batch result in collision, falling back to linear search. Average and best case will be O(1). 3. The size of flow cache is fixed and decided based on the number of flows to be supported. The default is set to 1 million flows, but is configurable by a startup.conf option. 4. Whenever a SPD rule is added/deleted by the control plane, all current flow cache entries will be invalidated. As the SPD API is not mp-safe, the data plane will wait for the control plane operation to complete. Cache invalidation is via an epoch counter that is incremented on policy add/del and stored with each entry in the flow cache. If the epoch counter in the flow cache does not match the current count, the entry is considered stale, and we fall back to linear search. The following configurable options are available through startup conf under the ipsec{} entry: 1. ipv4-inbound-spd-flow-cache on/off - enable SPD flow cache (default off) 2. ipv4-inbound-spd-hash-buckets %d - set number of hash buckets (default 4,194,304: ~1 million flows with 25% load factor) Performance with 1 core, 1 ESP Tunnel, null-decrypt then bypass, 94B (null encrypted packet) for different SPD policy matching indices: SPD Policy index : 2 10 100 1000 Throughput : Mbps/Mbps Mbps/Mbps Mbps/Mbps Mbps/Mbps (Baseline/Optimized) ARM TX2 : 300/290 230/290 70/290 8.5/290 Type: improvement Signed-off-by: Zachary Leaf <zachary.leaf@arm.com> Signed-off-by: mgovind <govindarajan.Mohandoss@arm.com> Tested-by: Jieqiang Wang <jieqiang.wang@arm.com> Change-Id: I8be2ad4715accbb335c38cd933904119db75827b
2021-06-25 08:11:15 -05:00
def verify_num_inbound_flow_cache_entries(self, expected_elements):
self.assertEqual(
self.get_spd_flow_cache_entries(outbound=False), expected_elements
)
ipsec: Performance improvement of ipsec4_output_node using flow cache Adding flow cache support to improve outbound IPv4/IPSec SPD lookup performance. Details about flow cache: Mechanism: 1. First packet of a flow will undergo linear search in SPD table. Once a policy match is found, a new entry will be added into the flow cache. From 2nd packet onwards, the policy lookup will happen in flow cache. 2. The flow cache is implemented using bihash without collision handling. This will avoid the logic to age out or recycle the old flows in flow cache. Whenever a collision occurs, old entry will be overwritten by the new entry. Worst case is when all the 256 packets in a batch result in collision and fall back to linear search. Average and best case will be O(1). 3. The size of flow cache is fixed and decided based on the number of flows to be supported. The default is set to 1 million flows. This can be made as a configurable option as a next step. 4. Whenever a SPD rule is added/deleted by the control plane, the flow cache entries will be completely deleted (reset) in the control plane. The assumption here is that SPD rule add/del is not a frequent operation from control plane. Flow cache reset is done, by putting the data plane in fall back mode, to bypass flow cache and do linear search till the SPD rule add/delete operation is complete. Once the rule is successfully added/deleted, the data plane will be allowed to make use of the flow cache. The flow cache will be reset only after flushing out the inflight packets from all the worker cores using vlib_worker_wait_one_loop(). Details about bihash usage: 1. A new bihash template (16_8) is added to support IPv4 5 tuple. BIHASH_KVP_PER_PAGE and BIHASH_KVP_AT_BUCKET_LEVEL are set to 1 in the new template. It means only one KVP is supported per bucket. 2. Collision handling is avoided by calling BV (clib_bihash_add_or_overwrite_stale) function. Through the stale callback function pointer, the KVP entry will be overwritten during collision. 3. Flow cache reset is done using BV (clib_bihash_foreach_key_value_pair) function. Through the callback function pointer, the KVP value is reset to ~0ULL. MRR performance numbers with 1 core, 1 ESP Tunnel, null-encrypt, 64B for different SPD policy matching indices: SPD Policy index : 1 10 100 1000 Throughput : MPPS/MPPS MPPS/MPPS MPPS/MPPS KPPS/MPPS (Baseline/Optimized) ARM Neoverse N1 : 5.2/4.84 4.55/4.84 2.11/4.84 329.5/4.84 ARM TX2 : 2.81/2.6 2.51/2.6 1.27/2.6 176.62/2.6 INTEL SKX : 4.93/4.48 4.29/4.46 2.05/4.48 336.79/4.47 Next Steps: Following can be made as a configurable option through startup conf at IPSec level: 1. Enable/Disable Flow cache. 2. Bihash configuration like number of buckets and memory size. 3. Dual/Quad loop unroll can be applied around bihash to further improve the performance. 4. The same flow cache logic can be applied for IPv6 as well as in IPSec inbound direction. A deeper and wider flow cache using bihash_40_8 can replace existing bihash_16_8, to make it common for both IPv4 and IPv6 in both outbound and inbound directions. Following changes are made based on the review comments: 1. ON/OFF flow cache through startup conf. Default: OFF 2. Flow cache stale entry detection using epoch counter. 3. Avoid host order endianness conversion during flow cache lookup. 4. Move IPSec startup conf to a common file. 5. Added SPD flow cache unit test case 6. Replaced bihash with vectors to implement flow cache. 7. ipsec_add_del_policy API is not mpsafe. Cleaned up inflight packets check in control plane. Type: improvement Signed-off-by: mgovind <govindarajan.Mohandoss@arm.com> Signed-off-by: Zachary Leaf <zachary.leaf@arm.com> Tested-by: Jieqiang Wang <jieqiang.wang@arm.com> Change-Id: I62b4d6625fbc6caf292427a5d2046aa5672b2006
2021-03-19 19:20:49 +00:00
def crc32_supported(self):
# lscpu is part of util-linux package, available on all Linux Distros
stream = os.popen("lscpu")
ipsec: Performance improvement of ipsec4_output_node using flow cache Adding flow cache support to improve outbound IPv4/IPSec SPD lookup performance. Details about flow cache: Mechanism: 1. First packet of a flow will undergo linear search in SPD table. Once a policy match is found, a new entry will be added into the flow cache. From 2nd packet onwards, the policy lookup will happen in flow cache. 2. The flow cache is implemented using bihash without collision handling. This will avoid the logic to age out or recycle the old flows in flow cache. Whenever a collision occurs, old entry will be overwritten by the new entry. Worst case is when all the 256 packets in a batch result in collision and fall back to linear search. Average and best case will be O(1). 3. The size of flow cache is fixed and decided based on the number of flows to be supported. The default is set to 1 million flows. This can be made as a configurable option as a next step. 4. Whenever a SPD rule is added/deleted by the control plane, the flow cache entries will be completely deleted (reset) in the control plane. The assumption here is that SPD rule add/del is not a frequent operation from control plane. Flow cache reset is done, by putting the data plane in fall back mode, to bypass flow cache and do linear search till the SPD rule add/delete operation is complete. Once the rule is successfully added/deleted, the data plane will be allowed to make use of the flow cache. The flow cache will be reset only after flushing out the inflight packets from all the worker cores using vlib_worker_wait_one_loop(). Details about bihash usage: 1. A new bihash template (16_8) is added to support IPv4 5 tuple. BIHASH_KVP_PER_PAGE and BIHASH_KVP_AT_BUCKET_LEVEL are set to 1 in the new template. It means only one KVP is supported per bucket. 2. Collision handling is avoided by calling BV (clib_bihash_add_or_overwrite_stale) function. Through the stale callback function pointer, the KVP entry will be overwritten during collision. 3. Flow cache reset is done using BV (clib_bihash_foreach_key_value_pair) function. Through the callback function pointer, the KVP value is reset to ~0ULL. MRR performance numbers with 1 core, 1 ESP Tunnel, null-encrypt, 64B for different SPD policy matching indices: SPD Policy index : 1 10 100 1000 Throughput : MPPS/MPPS MPPS/MPPS MPPS/MPPS KPPS/MPPS (Baseline/Optimized) ARM Neoverse N1 : 5.2/4.84 4.55/4.84 2.11/4.84 329.5/4.84 ARM TX2 : 2.81/2.6 2.51/2.6 1.27/2.6 176.62/2.6 INTEL SKX : 4.93/4.48 4.29/4.46 2.05/4.48 336.79/4.47 Next Steps: Following can be made as a configurable option through startup conf at IPSec level: 1. Enable/Disable Flow cache. 2. Bihash configuration like number of buckets and memory size. 3. Dual/Quad loop unroll can be applied around bihash to further improve the performance. 4. The same flow cache logic can be applied for IPv6 as well as in IPSec inbound direction. A deeper and wider flow cache using bihash_40_8 can replace existing bihash_16_8, to make it common for both IPv4 and IPv6 in both outbound and inbound directions. Following changes are made based on the review comments: 1. ON/OFF flow cache through startup conf. Default: OFF 2. Flow cache stale entry detection using epoch counter. 3. Avoid host order endianness conversion during flow cache lookup. 4. Move IPSec startup conf to a common file. 5. Added SPD flow cache unit test case 6. Replaced bihash with vectors to implement flow cache. 7. ipsec_add_del_policy API is not mpsafe. Cleaned up inflight packets check in control plane. Type: improvement Signed-off-by: mgovind <govindarajan.Mohandoss@arm.com> Signed-off-by: Zachary Leaf <zachary.leaf@arm.com> Tested-by: Jieqiang Wang <jieqiang.wang@arm.com> Change-Id: I62b4d6625fbc6caf292427a5d2046aa5672b2006
2021-03-19 19:20:49 +00:00
cpu_info = stream.read()
# feature/flag "crc32" on Aarch64 and "sse4_2" on x86
# see vppinfra/crc32.h
if "crc32" or "sse4_2" in cpu_info:
self.logger.info("\ncrc32 supported:\n" + cpu_info)
return True
else:
self.logger.info("\ncrc32 NOT supported:\n" + cpu_info)
return False
def create_stream(
cls, src_if, dst_if, pkt_count, src_prt=1234, dst_prt=4500, proto="UDP-ESP"
):
packets = []
packets = super(SpdFlowCacheTemplate, cls).create_stream(
src_if, dst_if, pkt_count, src_prt, dst_prt, proto
)
return packets
def verify_capture(
self, src_if, dst_if, capture, tcp_port_in=1234, udp_port_in=4500
):
super(SpdFlowCacheTemplate, self).verify_l3_l4_capture(
src_if, dst_if, capture, tcp_port_in, udp_port_in
)
class SpdFastPathTemplate(IPSecIPv4Fwd):
@classmethod
def setUpConstants(cls):
super(SpdFastPathTemplate, cls).setUpConstants()
# Override this method with required cmdline parameters e.g.
# cls.vpp_cmdline.extend(["ipsec", "{",
# "ipv4-outbound-spd-flow-cache on",
# "}"])
# cls.logger.info("VPP modified cmdline is %s" % " "
# .join(cls.vpp_cmdline))
def setUp(self):
super(SpdFastPathTemplate, self).setUp()
def tearDown(self):
super(SpdFastPathTemplate, self).tearDown()
def create_stream(
cls, src_if, dst_if, pkt_count, src_prt=1234, dst_prt=4500, proto="UDP-ESP"
):
packets = []
packets = super(SpdFastPathTemplate, cls).create_stream(
src_if, dst_if, pkt_count, src_prt, dst_prt, proto
)
return packets
def verify_capture(
self, src_if, dst_if, capture, tcp_port_in=1234, udp_port_in=4500
):
super(SpdFastPathTemplate, self).verify_l3_l4_capture(
src_if, dst_if, capture, tcp_port_in, udp_port_in
)
class IpsecDefaultTemplate(IPSecIPv4Fwd):
@classmethod
def setUpConstants(cls):
super(IpsecDefaultTemplate, cls).setUpConstants()
def setUp(self):
super(IpsecDefaultTemplate, self).setUp()
def tearDown(self):
super(IpsecDefaultTemplate, self).tearDown()
def create_stream(
cls, src_if, dst_if, pkt_count, src_prt=1234, dst_prt=4500, proto="UDP-ESP"
):
packets = []
packets = super(IpsecDefaultTemplate, cls).create_stream(
src_if, dst_if, pkt_count, src_prt, dst_prt, proto
)
return packets
def verify_capture(
self, src_if, dst_if, capture, tcp_port_in=1234, udp_port_in=4500
):
super(IpsecDefaultTemplate, self).verify_l3_l4_capture(
src_if, dst_if, capture, tcp_port_in, udp_port_in
)
class IPSecIPv6Fwd(VppTestCase):
"""Test IPSec by capturing and verifying IPv6 forwarded pkts"""
@classmethod
def setUpConstants(cls):
super(IPSecIPv6Fwd, cls).setUpConstants()
def setUp(self):
super(IPSecIPv6Fwd, self).setUp()
# store SPD objects so we can remove configs on tear down
self.spd_objs = []
self.spd_policies = []
def tearDown(self):
# remove SPD policies
for obj in self.spd_policies:
obj.remove_vpp_config()
self.spd_policies = []
# remove SPD items (interface bindings first, then SPD)
for obj in reversed(self.spd_objs):
obj.remove_vpp_config()
self.spd_objs = []
# close down pg intfs
for pg in self.pg_interfaces:
pg.unconfig_ip6()
pg.admin_down()
super(IPSecIPv6Fwd, self).tearDown()
def create_interfaces(self, num_ifs=2):
# create interfaces pg0 ... pg<num_ifs>
self.create_pg_interfaces(range(num_ifs))
for pg in self.pg_interfaces:
# put the interface up
pg.admin_up()
# configure IPv6 address on the interface
pg.config_ip6()
pg.resolve_ndp()
self.logger.info(self.vapi.ppcli("show int addr"))
def spd_create_and_intf_add(self, spd_id, pg_list):
spd = VppIpsecSpd(self, spd_id)
spd.add_vpp_config()
self.spd_objs.append(spd)
for pg in pg_list:
spdItf = VppIpsecSpdItfBinding(self, spd, pg)
spdItf.add_vpp_config()
self.spd_objs.append(spdItf)
def get_policy(self, policy_type):
e = VppEnum.vl_api_ipsec_spd_action_t
if policy_type == "protect":
return e.IPSEC_API_SPD_ACTION_PROTECT
elif policy_type == "bypass":
return e.IPSEC_API_SPD_ACTION_BYPASS
elif policy_type == "discard":
return e.IPSEC_API_SPD_ACTION_DISCARD
else:
raise Exception("Invalid policy type: %s", policy_type)
def spd_add_rem_policy(
self,
spd_id,
src_if,
dst_if,
proto,
is_out,
priority,
policy_type,
remove=False,
all_ips=False,
ip_range=False,
local_ip_start=ip_address("0::0"),
local_ip_stop=ip_address("ffff:ffff:ffff:ffff:ffff:ffff:ffff:ffff"),
remote_ip_start=ip_address("0::0"),
remote_ip_stop=ip_address("ffff:ffff:ffff:ffff:ffff:ffff:ffff:ffff"),
remote_port_start=0,
remote_port_stop=65535,
local_port_start=0,
local_port_stop=65535,
):
spd = VppIpsecSpd(self, spd_id)
if all_ips:
src_range_low = ip_address("0::0")
src_range_high = ip_address("ffff:ffff:ffff:ffff:ffff:ffff:ffff:ffff")
dst_range_low = ip_address("0::0")
dst_range_high = ip_address("ffff:ffff:ffff:ffff:ffff:ffff:ffff:ffff")
elif ip_range:
src_range_low = local_ip_start
src_range_high = local_ip_stop
dst_range_low = remote_ip_start
dst_range_high = remote_ip_stop
else:
src_range_low = src_if.remote_ip6
src_range_high = src_if.remote_ip6
dst_range_low = dst_if.remote_ip6
dst_range_high = dst_if.remote_ip6
spdEntry = VppIpsecSpdEntry(
self,
spd,
0,
src_range_low,
src_range_high,
dst_range_low,
dst_range_high,
proto,
priority=priority,
policy=self.get_policy(policy_type),
is_outbound=is_out,
remote_port_start=remote_port_start,
remote_port_stop=remote_port_stop,
local_port_start=local_port_start,
local_port_stop=local_port_stop,
)
if remove is False:
spdEntry.add_vpp_config()
self.spd_policies.append(spdEntry)
else:
spdEntry.remove_vpp_config()
self.spd_policies.remove(spdEntry)
self.logger.info(self.vapi.ppcli("show ipsec all"))
return spdEntry
def create_stream(self, src_if, dst_if, pkt_count, src_prt=1234, dst_prt=5678):
packets = []
for i in range(pkt_count):
# create packet info stored in the test case instance
info = self.create_packet_info(src_if, dst_if)
# convert the info into packet payload
payload = self.info_to_payload(info)
# create the packet itself
p = (
Ether(dst=src_if.local_mac, src=src_if.remote_mac)
/ IPv6(src=src_if.remote_ip6, dst=dst_if.remote_ip6)
/ UDP(sport=src_prt, dport=dst_prt)
/ Raw(payload)
)
# store a copy of the packet in the packet info
info.data = p.copy()
# append the packet to the list
packets.append(p)
# return the created packet list
return packets
def verify_capture(self, src_if, dst_if, capture):
packet_info = None
for packet in capture:
try:
ip = packet[IPv6]
udp = packet[UDP]
# convert the payload to packet info object
payload_info = self.payload_to_info(packet)
# make sure the indexes match
self.assert_equal(
payload_info.src, src_if.sw_if_index, "source sw_if_index"
)
self.assert_equal(
payload_info.dst, dst_if.sw_if_index, "destination sw_if_index"
)
packet_info = self.get_next_packet_info_for_interface2(
src_if.sw_if_index, dst_if.sw_if_index, packet_info
)
# make sure we didn't run out of saved packets
self.assertIsNotNone(packet_info)
self.assert_equal(
payload_info.index, packet_info.index, "packet info index"
)
saved_packet = packet_info.data # fetch the saved packet
# assert the values match
self.assert_equal(ip.src, saved_packet[IPv6].src, "IP source address")
# ... more assertions here
self.assert_equal(udp.sport, saved_packet[UDP].sport, "UDP source port")
except Exception as e:
self.logger.error(ppp("Unexpected or invalid packet:", packet))
raise
remaining_packet = self.get_next_packet_info_for_interface2(
src_if.sw_if_index, dst_if.sw_if_index, packet_info
)
self.assertIsNone(
remaining_packet,
"Interface %s: Packet expected from interface "
"%s didn't arrive" % (dst_if.name, src_if.name),
)
def verify_policy_match(self, pkt_count, spdEntry):
self.logger.info("XXXX %s %s", str(spdEntry), str(spdEntry.get_stats()))
matched_pkts = spdEntry.get_stats().get("packets")
self.logger.info("Policy %s matched: %d pkts", str(spdEntry), matched_pkts)
self.assert_equal(pkt_count, matched_pkts)
if __name__ == "__main__":
unittest.main(testRunner=VppTestRunner)