vpp/test/test_ipsec_esp.py

2571 lines
81 KiB
Python
Raw Permalink Normal View History

import socket
import unittest
from scapy.layers.ipsec import ESP
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
from scapy.layers.inet import IP, ICMP, UDP
from scapy.layers.inet6 import IPv6
from scapy.layers.l2 import Ether
from scapy.packet import Raw
from template_ipsec import (
IpsecTra46Tests,
IpsecTun46Tests,
TemplateIpsec,
IpsecTcpTests,
IpsecTra4Tests,
config_tra_params,
config_tun_params,
IPsecIPv4Params,
IPsecIPv6Params,
IpsecTra4,
IpsecTun4,
IpsecTra6,
IpsecTun6,
IpsecTun6HandoffTests,
IpsecTun4HandoffTests,
IpsecTra6ExtTests,
)
from vpp_ipsec import VppIpsecSpd, VppIpsecSpdEntry, VppIpsecSA, VppIpsecSpdItfBinding
from vpp_ip_route import VppIpRoute, VppRoutePath
from vpp_ip import DpoProto
from vpp_papi import VppEnum
from config import config
NUM_PKTS = 67
engines_supporting_chain_bufs = ["openssl", "async"]
engines = ["ia32", "ipsecmb", "openssl"]
class ConfigIpsecESP(TemplateIpsec):
encryption_type = ESP
tra4_encrypt_node_name = "esp4-encrypt"
tra4_decrypt_node_name = ["esp4-decrypt", "esp4-decrypt-post"]
tra6_encrypt_node_name = "esp6-encrypt"
tra6_decrypt_node_name = ["esp6-decrypt", "esp6-decrypt-post"]
tun4_encrypt_node_name = "esp4-encrypt"
tun4_decrypt_node_name = ["esp4-decrypt", "esp4-decrypt-post"]
tun6_encrypt_node_name = "esp6-encrypt"
tun6_decrypt_node_name = ["esp6-decrypt", "esp6-decrypt-post"]
@classmethod
def setUpClass(cls):
super(ConfigIpsecESP, cls).setUpClass()
@classmethod
def tearDownClass(cls):
super(ConfigIpsecESP, cls).tearDownClass()
def setUp(self):
super(ConfigIpsecESP, self).setUp()
def tearDown(self):
super(ConfigIpsecESP, self).tearDown()
def config_anti_replay(self, params, anti_replay_window_size=64):
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
saf = VppEnum.vl_api_ipsec_sad_flags_t
for p in params:
p.flags |= saf.IPSEC_API_SAD_FLAG_USE_ANTI_REPLAY
p.anti_replay_window_size = anti_replay_window_size
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 config_network(self, params):
self.net_objs = []
self.tun_if = self.pg0
self.tra_if = self.pg2
self.logger.info(self.vapi.ppcli("show int addr"))
self.tra_spd = VppIpsecSpd(self, self.tra_spd_id)
self.tra_spd.add_vpp_config()
self.net_objs.append(self.tra_spd)
self.tun_spd = VppIpsecSpd(self, self.tun_spd_id)
self.tun_spd.add_vpp_config()
self.net_objs.append(self.tun_spd)
b = VppIpsecSpdItfBinding(self, self.tun_spd, self.tun_if)
b.add_vpp_config()
self.net_objs.append(b)
b = VppIpsecSpdItfBinding(self, self.tra_spd, self.tra_if)
b.add_vpp_config()
self.net_objs.append(b)
for p in params:
self.config_esp_tra(p)
config_tra_params(p, self.encryption_type)
for p in params:
self.config_esp_tun(p)
config_tun_params(p, self.encryption_type, self.tun_if)
for p in params:
d = DpoProto.DPO_PROTO_IP6 if p.is_ipv6 else DpoProto.DPO_PROTO_IP4
r = VppIpRoute(
self,
p.remote_tun_if_host,
p.addr_len,
[
VppRoutePath(
self.tun_if.remote_addr[p.addr_type], 0xFFFFFFFF, proto=d
)
],
)
r.add_vpp_config()
self.net_objs.append(r)
self.logger.info(self.vapi.ppcli("show ipsec all"))
def unconfig_network(self):
for o in reversed(self.net_objs):
o.remove_vpp_config()
self.net_objs = []
def config_esp_tun(self, params):
addr_type = params.addr_type
scapy_tun_sa_id = params.scapy_tun_sa_id
scapy_tun_spi = params.scapy_tun_spi
vpp_tun_sa_id = params.vpp_tun_sa_id
vpp_tun_spi = params.vpp_tun_spi
auth_algo_vpp_id = params.auth_algo_vpp_id
auth_key = params.auth_key
crypt_algo_vpp_id = params.crypt_algo_vpp_id
crypt_key = params.crypt_key
remote_tun_if_host = params.remote_tun_if_host
addr_any = params.addr_any
addr_bcast = params.addr_bcast
e = VppEnum.vl_api_ipsec_spd_action_t
flags = params.flags
tun_flags = params.tun_flags
salt = params.salt
anti_replay_window_size = params.anti_replay_window_size
objs = []
params.tun_sa_in = VppIpsecSA(
self,
scapy_tun_sa_id,
scapy_tun_spi,
auth_algo_vpp_id,
auth_key,
crypt_algo_vpp_id,
crypt_key,
self.vpp_esp_protocol,
self.tun_if.remote_addr[addr_type],
self.tun_if.local_addr[addr_type],
tun_flags=tun_flags,
dscp=params.dscp,
flags=flags,
salt=salt,
hop_limit=params.outer_hop_limit,
anti_replay_window_size=anti_replay_window_size,
)
params.tun_sa_out = VppIpsecSA(
self,
vpp_tun_sa_id,
vpp_tun_spi,
auth_algo_vpp_id,
auth_key,
crypt_algo_vpp_id,
crypt_key,
self.vpp_esp_protocol,
self.tun_if.local_addr[addr_type],
self.tun_if.remote_addr[addr_type],
tun_flags=tun_flags,
dscp=params.dscp,
flags=flags,
salt=salt,
hop_limit=params.outer_hop_limit,
anti_replay_window_size=anti_replay_window_size,
)
objs.append(params.tun_sa_in)
objs.append(params.tun_sa_out)
params.spd_policy_in_any = VppIpsecSpdEntry(
self,
self.tun_spd,
scapy_tun_sa_id,
addr_any,
addr_bcast,
addr_any,
addr_bcast,
socket.IPPROTO_ESP,
)
params.spd_policy_out_any = VppIpsecSpdEntry(
self,
self.tun_spd,
scapy_tun_sa_id,
addr_any,
addr_bcast,
addr_any,
addr_bcast,
socket.IPPROTO_ESP,
is_outbound=0,
)
objs.append(params.spd_policy_out_any)
objs.append(params.spd_policy_in_any)
objs.append(
VppIpsecSpdEntry(
self,
self.tun_spd,
scapy_tun_sa_id,
remote_tun_if_host,
remote_tun_if_host,
self.pg1.remote_addr[addr_type],
self.pg1.remote_addr[addr_type],
socket.IPPROTO_RAW,
priority=10,
policy=e.IPSEC_API_SPD_ACTION_PROTECT,
is_outbound=0,
)
)
objs.append(
VppIpsecSpdEntry(
self,
self.tun_spd,
vpp_tun_sa_id,
self.pg1.remote_addr[addr_type],
self.pg1.remote_addr[addr_type],
remote_tun_if_host,
remote_tun_if_host,
socket.IPPROTO_RAW,
policy=e.IPSEC_API_SPD_ACTION_PROTECT,
priority=10,
)
)
objs.append(
VppIpsecSpdEntry(
self,
self.tun_spd,
scapy_tun_sa_id,
remote_tun_if_host,
remote_tun_if_host,
self.pg0.local_addr[addr_type],
self.pg0.local_addr[addr_type],
socket.IPPROTO_RAW,
priority=20,
policy=e.IPSEC_API_SPD_ACTION_PROTECT,
is_outbound=0,
)
)
objs.append(
VppIpsecSpdEntry(
self,
self.tun_spd,
vpp_tun_sa_id,
self.pg0.local_addr[addr_type],
self.pg0.local_addr[addr_type],
remote_tun_if_host,
remote_tun_if_host,
socket.IPPROTO_RAW,
policy=e.IPSEC_API_SPD_ACTION_PROTECT,
priority=20,
)
)
for o in objs:
o.add_vpp_config()
self.net_objs = self.net_objs + objs
def config_esp_tra(self, params):
addr_type = params.addr_type
scapy_tra_sa_id = params.scapy_tra_sa_id
scapy_tra_spi = params.scapy_tra_spi
vpp_tra_sa_id = params.vpp_tra_sa_id
vpp_tra_spi = params.vpp_tra_spi
auth_algo_vpp_id = params.auth_algo_vpp_id
auth_key = params.auth_key
crypt_algo_vpp_id = params.crypt_algo_vpp_id
crypt_key = params.crypt_key
addr_any = params.addr_any
addr_bcast = params.addr_bcast
e = VppEnum.vl_api_ipsec_spd_action_t
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
flags = params.flags
salt = params.salt
anti_replay_window_size = params.anti_replay_window_size
objs = []
params.tra_sa_in = VppIpsecSA(
self,
scapy_tra_sa_id,
scapy_tra_spi,
auth_algo_vpp_id,
auth_key,
crypt_algo_vpp_id,
crypt_key,
self.vpp_esp_protocol,
flags=flags,
salt=salt,
anti_replay_window_size=anti_replay_window_size,
)
params.tra_sa_out = VppIpsecSA(
self,
vpp_tra_sa_id,
vpp_tra_spi,
auth_algo_vpp_id,
auth_key,
crypt_algo_vpp_id,
crypt_key,
self.vpp_esp_protocol,
flags=flags,
salt=salt,
anti_replay_window_size=anti_replay_window_size,
)
objs.append(params.tra_sa_in)
objs.append(params.tra_sa_out)
objs.append(
VppIpsecSpdEntry(
self,
self.tra_spd,
vpp_tra_sa_id,
addr_any,
addr_bcast,
addr_any,
addr_bcast,
socket.IPPROTO_ESP,
)
)
objs.append(
VppIpsecSpdEntry(
self,
self.tra_spd,
vpp_tra_sa_id,
addr_any,
addr_bcast,
addr_any,
addr_bcast,
socket.IPPROTO_ESP,
is_outbound=0,
)
)
objs.append(
VppIpsecSpdEntry(
self,
self.tra_spd,
scapy_tra_sa_id,
self.tra_if.local_addr[addr_type],
self.tra_if.local_addr[addr_type],
self.tra_if.remote_addr[addr_type],
self.tra_if.remote_addr[addr_type],
socket.IPPROTO_RAW,
priority=10,
policy=e.IPSEC_API_SPD_ACTION_PROTECT,
is_outbound=0,
)
)
objs.append(
VppIpsecSpdEntry(
self,
self.tra_spd,
vpp_tra_sa_id,
self.tra_if.local_addr[addr_type],
self.tra_if.local_addr[addr_type],
self.tra_if.remote_addr[addr_type],
self.tra_if.remote_addr[addr_type],
socket.IPPROTO_RAW,
policy=e.IPSEC_API_SPD_ACTION_PROTECT,
priority=10,
)
)
for o in objs:
o.add_vpp_config()
self.net_objs = self.net_objs + objs
class TemplateIpsecEsp(ConfigIpsecESP):
"""
Basic test for ipsec esp sanity - tunnel and transport modes.
Below 4 cases are covered as part of this test
#. ipsec esp v4 transport basic test - IPv4 Transport mode
scenario using HMAC-SHA1-96 integrity algo
#. ipsec esp v4 transport burst test
Above test for 257 pkts
#. ipsec esp 4o4 tunnel basic test - IPv4 Tunnel mode
scenario using HMAC-SHA1-96 integrity algo
#. ipsec esp 4o4 tunnel burst test
Above test for 257 pkts
TRANSPORT MODE::
--- encrypt ---
|pg2| <-------> |VPP|
--- decrypt ---
TUNNEL MODE::
--- encrypt --- plain ---
|pg0| <------- |VPP| <------ |pg1|
--- --- ---
--- decrypt --- plain ---
|pg0| -------> |VPP| ------> |pg1|
--- --- ---
"""
@classmethod
def setUpClass(cls):
super(TemplateIpsecEsp, cls).setUpClass()
@classmethod
def tearDownClass(cls):
super(TemplateIpsecEsp, cls).tearDownClass()
def setUp(self):
super(TemplateIpsecEsp, self).setUp()
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.config_anti_replay(self.params.values())
self.config_network(self.params.values())
def tearDown(self):
self.unconfig_network()
super(TemplateIpsecEsp, self).tearDown()
class TestIpsecEsp1(
TemplateIpsecEsp, IpsecTra46Tests, IpsecTun46Tests, IpsecTra6ExtTests
):
"""Ipsec ESP - TUN & TRA tests"""
@classmethod
def setUpClass(cls):
super(TestIpsecEsp1, cls).setUpClass()
@classmethod
def tearDownClass(cls):
super(TestIpsecEsp1, cls).tearDownClass()
def setUp(self):
super(TestIpsecEsp1, self).setUp()
def tearDown(self):
super(TestIpsecEsp1, self).tearDown()
def test_tun_46(self):
"""ipsec 4o6 tunnel"""
# add an SPD entry to direct 2.2.2.2 to the v6 tunnel SA
p6 = self.ipv6_params
p4 = self.ipv4_params
p6.remote_tun_if_host4 = "2.2.2.2"
e = VppEnum.vl_api_ipsec_spd_action_t
VppIpsecSpdEntry(
self,
self.tun_spd,
p6.vpp_tun_sa_id,
self.pg1.remote_addr[p4.addr_type],
self.pg1.remote_addr[p4.addr_type],
p6.remote_tun_if_host4,
p6.remote_tun_if_host4,
socket.IPPROTO_RAW,
priority=10,
policy=e.IPSEC_API_SPD_ACTION_PROTECT,
is_outbound=1,
).add_vpp_config()
VppIpRoute(
self,
p6.remote_tun_if_host4,
p4.addr_len,
[VppRoutePath(self.tun_if.remote_addr[p4.addr_type], 0xFFFFFFFF)],
).add_vpp_config()
old_name = self.tun6_encrypt_node_name
self.tun6_encrypt_node_name = "esp4-encrypt"
self.verify_tun_46(p6, count=63)
self.tun6_encrypt_node_name = old_name
def test_tun_64(self):
"""ipsec 6o4 tunnel"""
# add an SPD entry to direct 4444::4 to the v4 tunnel SA
p6 = self.ipv6_params
p4 = self.ipv4_params
p4.remote_tun_if_host6 = "4444::4"
e = VppEnum.vl_api_ipsec_spd_action_t
VppIpsecSpdEntry(
self,
self.tun_spd,
p4.vpp_tun_sa_id,
self.pg1.remote_addr[p6.addr_type],
self.pg1.remote_addr[p6.addr_type],
p4.remote_tun_if_host6,
p4.remote_tun_if_host6,
socket.IPPROTO_RAW,
priority=10,
policy=e.IPSEC_API_SPD_ACTION_PROTECT,
is_outbound=1,
).add_vpp_config()
d = DpoProto.DPO_PROTO_IP6
VppIpRoute(
self,
p4.remote_tun_if_host6,
p6.addr_len,
[VppRoutePath(self.tun_if.remote_addr[p6.addr_type], 0xFFFFFFFF, proto=d)],
).add_vpp_config()
old_name = self.tun4_encrypt_node_name
self.tun4_encrypt_node_name = "esp6-encrypt"
self.verify_tun_64(p4, count=63)
self.tun4_encrypt_node_name = old_name
class TestIpsecEspTun(TemplateIpsecEsp, IpsecTun46Tests):
"""Ipsec ESP - TUN encap tests"""
def setUp(self):
self.ipv4_params = IPsecIPv4Params()
self.ipv6_params = IPsecIPv6Params()
c = (
VppEnum.vl_api_tunnel_encap_decap_flags_t.TUNNEL_API_ENCAP_DECAP_FLAG_ENCAP_COPY_DSCP
)
c1 = c | (
VppEnum.vl_api_tunnel_encap_decap_flags_t.TUNNEL_API_ENCAP_DECAP_FLAG_ENCAP_COPY_ECN
)
self.ipv4_params.tun_flags = c
self.ipv6_params.tun_flags = c1
super(TestIpsecEspTun, self).setUp()
def gen_pkts(self, sw_intf, src, dst, count=1, payload_size=54):
# set the DSCP + ECN - flags are set to copy only DSCP
return [
Ether(src=sw_intf.remote_mac, dst=sw_intf.local_mac)
/ IP(src=src, dst=dst, tos=5)
/ UDP(sport=4444, dport=4444)
/ Raw(b"X" * payload_size)
for i in range(count)
]
def gen_pkts6(self, p, sw_intf, src, dst, count=1, payload_size=54):
# set the DSCP + ECN - flags are set to copy both
return [
Ether(src=sw_intf.remote_mac, dst=sw_intf.local_mac)
/ IPv6(src=src, dst=dst, tc=5)
/ UDP(sport=4444, dport=4444)
/ Raw(b"X" * payload_size)
for i in range(count)
]
def verify_encrypted(self, p, sa, rxs):
# just check that only the DSCP is copied
for rx in rxs:
self.assertEqual(rx[IP].tos, 4)
def verify_encrypted6(self, p, sa, rxs):
# just check that the DSCP & ECN are copied
for rx in rxs:
self.assertEqual(rx[IPv6].tc, 5)
class TestIpsecEspTun2(TemplateIpsecEsp, IpsecTun46Tests):
"""Ipsec ESP - TUN DSCP tests"""
def setUp(self):
self.ipv4_params = IPsecIPv4Params()
self.ipv6_params = IPsecIPv6Params()
self.ipv4_params.dscp = VppEnum.vl_api_ip_dscp_t.IP_API_DSCP_EF
self.ipv6_params.dscp = VppEnum.vl_api_ip_dscp_t.IP_API_DSCP_AF11
super(TestIpsecEspTun2, self).setUp()
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)
/ UDP(sport=4444, dport=4444)
/ 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)
/ UDP(sport=4444, dport=4444)
/ Raw(b"X" * payload_size)
for i in range(count)
]
def verify_encrypted(self, p, sa, rxs):
# just check that only the DSCP is set
for rx in rxs:
self.assertEqual(rx[IP].tos, VppEnum.vl_api_ip_dscp_t.IP_API_DSCP_EF << 2)
def verify_encrypted6(self, p, sa, rxs):
# just check that the DSCP is set
for rx in rxs:
self.assertEqual(
rx[IPv6].tc, VppEnum.vl_api_ip_dscp_t.IP_API_DSCP_AF11 << 2
)
class TestIpsecEsp2(TemplateIpsecEsp, IpsecTcpTests):
"""Ipsec ESP - TCP tests"""
pass
class TestIpsecEspAsync(TemplateIpsecEsp):
"""Ipsec ESP - Aysnc tests"""
vpp_worker_count = 2
def setUp(self):
super(TestIpsecEspAsync, self).setUp()
self.p_sync = IPsecIPv4Params()
self.p_sync.crypt_algo_vpp_id = (
VppEnum.vl_api_ipsec_crypto_alg_t.IPSEC_API_CRYPTO_ALG_AES_CBC_256
)
self.p_sync.crypt_algo = "AES-CBC" # scapy name
self.p_sync.crypt_key = b"JPjyOWBeVEQiMe7hJPjyOWBeVEQiMe7h"
self.p_sync.scapy_tun_sa_id += 0xF0000
self.p_sync.scapy_tun_spi += 0xF0000
self.p_sync.vpp_tun_sa_id += 0xF0000
self.p_sync.vpp_tun_spi += 0xF0000
self.p_sync.remote_tun_if_host = "2.2.2.2"
e = VppEnum.vl_api_ipsec_spd_action_t
self.p_sync.sa = VppIpsecSA(
self,
self.p_sync.vpp_tun_sa_id,
self.p_sync.vpp_tun_spi,
self.p_sync.auth_algo_vpp_id,
self.p_sync.auth_key,
self.p_sync.crypt_algo_vpp_id,
self.p_sync.crypt_key,
self.vpp_esp_protocol,
self.tun_if.local_addr[self.p_sync.addr_type],
self.tun_if.remote_addr[self.p_sync.addr_type],
).add_vpp_config()
self.p_sync.spd = VppIpsecSpdEntry(
self,
self.tun_spd,
self.p_sync.vpp_tun_sa_id,
self.pg1.remote_addr[self.p_sync.addr_type],
self.pg1.remote_addr[self.p_sync.addr_type],
self.p_sync.remote_tun_if_host,
self.p_sync.remote_tun_if_host,
socket.IPPROTO_RAW,
priority=1,
policy=e.IPSEC_API_SPD_ACTION_PROTECT,
is_outbound=1,
).add_vpp_config()
VppIpRoute(
self,
self.p_sync.remote_tun_if_host,
self.p_sync.addr_len,
[VppRoutePath(self.tun_if.remote_addr[self.p_sync.addr_type], 0xFFFFFFFF)],
).add_vpp_config()
config_tun_params(self.p_sync, self.encryption_type, self.tun_if)
self.p_async = IPsecIPv4Params()
self.p_async.crypt_algo_vpp_id = (
VppEnum.vl_api_ipsec_crypto_alg_t.IPSEC_API_CRYPTO_ALG_AES_GCM_256
)
self.p_async.auth_algo_vpp_id = (
VppEnum.vl_api_ipsec_integ_alg_t.IPSEC_API_INTEG_ALG_NONE
)
self.p_async.crypt_algo = "AES-GCM" # scapy name
self.p_async.crypt_key = b"JPjyOWBeVEQiMe7hJPjyOWBeVEQiMe7h"
self.p_async.auth_algo = "NULL"
self.p_async.scapy_tun_sa_id += 0xE0000
self.p_async.scapy_tun_spi += 0xE0000
self.p_async.vpp_tun_sa_id += 0xE0000
self.p_async.vpp_tun_spi += 0xE0000
self.p_async.remote_tun_if_host = "2.2.2.3"
iflags = VppEnum.vl_api_ipsec_sad_flags_t
self.p_async.flags = (
iflags.IPSEC_API_SAD_FLAG_USE_ESN
| iflags.IPSEC_API_SAD_FLAG_USE_ANTI_REPLAY
| iflags.IPSEC_API_SAD_FLAG_ASYNC
)
self.p_async.sa = VppIpsecSA(
self,
self.p_async.vpp_tun_sa_id,
self.p_async.vpp_tun_spi,
self.p_async.auth_algo_vpp_id,
self.p_async.auth_key,
self.p_async.crypt_algo_vpp_id,
self.p_async.crypt_key,
self.vpp_esp_protocol,
self.tun_if.local_addr[self.p_async.addr_type],
self.tun_if.remote_addr[self.p_async.addr_type],
flags=self.p_async.flags,
).add_vpp_config()
self.p_async.spd = VppIpsecSpdEntry(
self,
self.tun_spd,
self.p_async.vpp_tun_sa_id,
self.pg1.remote_addr[self.p_async.addr_type],
self.pg1.remote_addr[self.p_async.addr_type],
self.p_async.remote_tun_if_host,
self.p_async.remote_tun_if_host,
socket.IPPROTO_RAW,
priority=2,
policy=e.IPSEC_API_SPD_ACTION_PROTECT,
is_outbound=1,
).add_vpp_config()
VppIpRoute(
self,
self.p_async.remote_tun_if_host,
self.p_async.addr_len,
[VppRoutePath(self.tun_if.remote_addr[self.p_async.addr_type], 0xFFFFFFFF)],
).add_vpp_config()
config_tun_params(self.p_async, self.encryption_type, self.tun_if)
def test_dual_stream(self):
"""Alternating SAs"""
p = self.params[self.p_sync.addr_type]
self.vapi.ipsec_set_async_mode(async_enable=True)
pkts = [
(
Ether(src=self.pg1.remote_mac, dst=self.pg1.local_mac)
/ IP(src=self.pg1.remote_ip4, dst=self.p_sync.remote_tun_if_host)
/ UDP(sport=4444, dport=4444)
/ Raw(b"0x0" * 200)
),
(
Ether(src=self.pg1.remote_mac, dst=self.pg1.local_mac)
/ IP(src=self.pg1.remote_ip4, dst=p.remote_tun_if_host)
/ UDP(sport=4444, dport=4444)
/ Raw(b"0x0" * 200)
),
]
pkts *= 1023
rxs = self.send_and_expect(self.pg1, pkts, self.pg0)
self.assertEqual(len(rxs), len(pkts))
for rx in rxs:
if rx[ESP].spi == p.vpp_tun_spi:
decrypted = p.vpp_tun_sa.decrypt(rx[IP])
elif rx[ESP].spi == self.p_sync.vpp_tun_spi:
decrypted = self.p_sync.vpp_tun_sa.decrypt(rx[IP])
else:
rx.show()
self.assertTrue(False)
self.p_sync.spd.remove_vpp_config()
self.p_sync.sa.remove_vpp_config()
self.p_async.spd.remove_vpp_config()
self.p_async.sa.remove_vpp_config()
self.vapi.ipsec_set_async_mode(async_enable=False)
def test_sync_async_noop_stream(self):
"""Alternating SAs sync/async/noop"""
p = self.params[self.p_sync.addr_type]
# first pin the default/noop SA to worker 0
pkts = [
(
Ether(src=self.pg1.remote_mac, dst=self.pg1.local_mac)
/ IP(src=self.pg1.remote_ip4, dst=p.remote_tun_if_host)
/ UDP(sport=4444, dport=4444)
/ Raw(b"0x0" * 200)
)
]
rxs = self.send_and_expect(self.pg1, pkts, self.pg0, worker=0)
self.logger.info(self.vapi.cli("sh ipsec sa"))
self.logger.info(self.vapi.cli("sh crypto async status"))
# then use all the other SAs on worker 1.
# some will handoff, other take the sync and async paths
pkts = [
(
Ether(src=self.pg1.remote_mac, dst=self.pg1.local_mac)
/ IP(src=self.pg1.remote_ip4, dst=self.p_sync.remote_tun_if_host)
/ UDP(sport=4444, dport=4444)
/ Raw(b"0x0" * 200)
),
(
Ether(src=self.pg1.remote_mac, dst=self.pg1.local_mac)
/ IP(src=self.pg1.remote_ip4, dst=p.remote_tun_if_host)
/ UDP(sport=4444, dport=4444)
/ Raw(b"0x0" * 200)
),
(
Ether(src=self.pg1.remote_mac, dst=self.pg1.local_mac)
/ IP(src=self.pg1.remote_ip4, dst=self.p_async.remote_tun_if_host)
/ UDP(sport=4444, dport=4444)
/ Raw(b"0x0" * 200)
),
]
pkts *= 1023
rxs = self.send_and_expect(self.pg1, pkts, self.pg0, worker=1)
self.assertEqual(len(rxs), len(pkts))
for rx in rxs:
if rx[ESP].spi == p.vpp_tun_spi:
decrypted = p.vpp_tun_sa.decrypt(rx[IP])
elif rx[ESP].spi == self.p_sync.vpp_tun_spi:
decrypted = self.p_sync.vpp_tun_sa.decrypt(rx[IP])
elif rx[ESP].spi == self.p_async.vpp_tun_spi:
decrypted = self.p_async.vpp_tun_sa.decrypt(rx[IP])
else:
rx.show()
self.assertTrue(False)
self.p_sync.spd.remove_vpp_config()
self.p_sync.sa.remove_vpp_config()
self.p_async.spd.remove_vpp_config()
self.p_async.sa.remove_vpp_config()
class TestIpsecEspHandoff(
TemplateIpsecEsp, IpsecTun6HandoffTests, IpsecTun4HandoffTests
):
"""Ipsec ESP - handoff tests"""
pass
class TemplateIpsecEspUdp(ConfigIpsecESP):
"""
UDP encapped ESP
"""
@classmethod
def setUpClass(cls):
super(TemplateIpsecEspUdp, cls).setUpClass()
@classmethod
def tearDownClass(cls):
super(TemplateIpsecEspUdp, cls).tearDownClass()
def setUp(self):
super(TemplateIpsecEspUdp, self).setUp()
self.net_objs = []
self.tun_if = self.pg0
self.tra_if = self.pg2
self.logger.info(self.vapi.ppcli("show int addr"))
p = self.ipv4_params
p.flags = (
VppEnum.vl_api_ipsec_sad_flags_t.IPSEC_API_SAD_FLAG_UDP_ENCAP
| VppEnum.vl_api_ipsec_sad_flags_t.IPSEC_API_SAD_FLAG_USE_ANTI_REPLAY
)
p.nat_header = UDP(sport=5454, dport=4500)
self.tra_spd = VppIpsecSpd(self, self.tra_spd_id)
self.tra_spd.add_vpp_config()
VppIpsecSpdItfBinding(self, self.tra_spd, self.tra_if).add_vpp_config()
self.config_esp_tra(p)
config_tra_params(p, self.encryption_type)
self.tun_spd = VppIpsecSpd(self, self.tun_spd_id)
self.tun_spd.add_vpp_config()
VppIpsecSpdItfBinding(self, self.tun_spd, self.tun_if).add_vpp_config()
self.config_esp_tun(p)
self.logger.info(self.vapi.ppcli("show ipsec all"))
d = DpoProto.DPO_PROTO_IP4
VppIpRoute(
self,
p.remote_tun_if_host,
p.addr_len,
[VppRoutePath(self.tun_if.remote_addr[p.addr_type], 0xFFFFFFFF, proto=d)],
).add_vpp_config()
def tearDown(self):
super(TemplateIpsecEspUdp, self).tearDown()
def show_commands_at_teardown(self):
self.logger.info(self.vapi.cli("show hardware"))
class TestIpsecEspUdp(TemplateIpsecEspUdp, IpsecTra4Tests):
"""Ipsec NAT-T ESP UDP tests"""
pass
class MyParameters:
def __init__(self):
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
saf = VppEnum.vl_api_ipsec_sad_flags_t
flag_esn = saf.IPSEC_API_SAD_FLAG_USE_ESN
flag_ar = saf.IPSEC_API_SAD_FLAG_USE_ANTI_REPLAY
self.flags = [0, flag_esn, flag_ar]
# foreach crypto algorithm
self.algos = {
"AES-GCM-128/NONE": {
"vpp-crypto": (
VppEnum.vl_api_ipsec_crypto_alg_t.IPSEC_API_CRYPTO_ALG_AES_GCM_128
),
"vpp-integ": (
VppEnum.vl_api_ipsec_integ_alg_t.IPSEC_API_INTEG_ALG_NONE
),
"scapy-crypto": "AES-GCM",
"scapy-integ": "NULL",
"key": b"JPjyOWBeVEQiMe7h",
"salt": 0,
},
"AES-GCM-192/NONE": {
"vpp-crypto": (
VppEnum.vl_api_ipsec_crypto_alg_t.IPSEC_API_CRYPTO_ALG_AES_GCM_192
),
"vpp-integ": (
VppEnum.vl_api_ipsec_integ_alg_t.IPSEC_API_INTEG_ALG_NONE
),
"scapy-crypto": "AES-GCM",
"scapy-integ": "NULL",
"key": b"JPjyOWBeVEQiMe7h01234567",
"salt": 1010,
},
"AES-GCM-256/NONE": {
"vpp-crypto": (
VppEnum.vl_api_ipsec_crypto_alg_t.IPSEC_API_CRYPTO_ALG_AES_GCM_256
),
"vpp-integ": (
VppEnum.vl_api_ipsec_integ_alg_t.IPSEC_API_INTEG_ALG_NONE
),
"scapy-crypto": "AES-GCM",
"scapy-integ": "NULL",
"key": b"JPjyOWBeVEQiMe7h0123456787654321",
"salt": 2020,
},
"AES-CBC-128/MD5-96": {
"vpp-crypto": (
VppEnum.vl_api_ipsec_crypto_alg_t.IPSEC_API_CRYPTO_ALG_AES_CBC_128
),
"vpp-integ": (
VppEnum.vl_api_ipsec_integ_alg_t.IPSEC_API_INTEG_ALG_MD5_96
),
"scapy-crypto": "AES-CBC",
"scapy-integ": "HMAC-MD5-96",
"salt": 0,
"key": b"JPjyOWBeVEQiMe7h",
},
"AES-CBC-192/SHA1-96": {
"vpp-crypto": (
VppEnum.vl_api_ipsec_crypto_alg_t.IPSEC_API_CRYPTO_ALG_AES_CBC_192
),
"vpp-integ": (
VppEnum.vl_api_ipsec_integ_alg_t.IPSEC_API_INTEG_ALG_SHA1_96
),
"scapy-crypto": "AES-CBC",
"scapy-integ": "HMAC-SHA1-96",
"salt": 0,
"key": b"JPjyOWBeVEQiMe7hJPjyOWBe",
},
"AES-CBC-256/SHA1-96": {
"vpp-crypto": (
VppEnum.vl_api_ipsec_crypto_alg_t.IPSEC_API_CRYPTO_ALG_AES_CBC_256
),
"vpp-integ": (
VppEnum.vl_api_ipsec_integ_alg_t.IPSEC_API_INTEG_ALG_SHA1_96
),
"scapy-crypto": "AES-CBC",
"scapy-integ": "HMAC-SHA1-96",
"salt": 0,
"key": b"JPjyOWBeVEQiMe7hJPjyOWBeVEQiMe7h",
},
"3DES-CBC/SHA1-96": {
"vpp-crypto": (
VppEnum.vl_api_ipsec_crypto_alg_t.IPSEC_API_CRYPTO_ALG_3DES_CBC
),
"vpp-integ": (
VppEnum.vl_api_ipsec_integ_alg_t.IPSEC_API_INTEG_ALG_SHA1_96
),
"scapy-crypto": "3DES",
"scapy-integ": "HMAC-SHA1-96",
"salt": 0,
"key": b"JPjyOWBeVEQiMe7h00112233",
},
"NONE/SHA1-96": {
"vpp-crypto": (
VppEnum.vl_api_ipsec_crypto_alg_t.IPSEC_API_CRYPTO_ALG_NONE
),
"vpp-integ": (
VppEnum.vl_api_ipsec_integ_alg_t.IPSEC_API_INTEG_ALG_SHA1_96
),
"scapy-crypto": "NULL",
"scapy-integ": "HMAC-SHA1-96",
"salt": 0,
"key": b"JPjyOWBeVEQiMe7h00112233",
},
"AES-CTR-128/SHA1-96": {
"vpp-crypto": (
VppEnum.vl_api_ipsec_crypto_alg_t.IPSEC_API_CRYPTO_ALG_AES_CTR_128
),
"vpp-integ": (
VppEnum.vl_api_ipsec_integ_alg_t.IPSEC_API_INTEG_ALG_SHA1_96
),
"scapy-crypto": "AES-CTR",
"scapy-integ": "HMAC-SHA1-96",
"salt": 0,
"key": b"JPjyOWBeVEQiMe7h",
},
"AES-CTR-192/SHA1-96": {
"vpp-crypto": (
VppEnum.vl_api_ipsec_crypto_alg_t.IPSEC_API_CRYPTO_ALG_AES_CTR_192
),
"vpp-integ": (
VppEnum.vl_api_ipsec_integ_alg_t.IPSEC_API_INTEG_ALG_SHA1_96
),
"scapy-crypto": "AES-CTR",
"scapy-integ": "HMAC-SHA1-96",
"salt": 1010,
"key": b"JPjyOWBeVEQiMe7hJPjyOWBe",
},
"AES-CTR-256/SHA1-96": {
"vpp-crypto": (
VppEnum.vl_api_ipsec_crypto_alg_t.IPSEC_API_CRYPTO_ALG_AES_CTR_256
),
"vpp-integ": (
VppEnum.vl_api_ipsec_integ_alg_t.IPSEC_API_INTEG_ALG_SHA1_96
),
"scapy-crypto": "AES-CTR",
"scapy-integ": "HMAC-SHA1-96",
"salt": 2020,
"key": b"JPjyOWBeVEQiMe7hJPjyOWBeVEQiMe7h",
},
"AES-NULL-GMAC-128/NONE": {
"vpp-crypto": (
VppEnum.vl_api_ipsec_crypto_alg_t.IPSEC_API_CRYPTO_ALG_AES_NULL_GMAC_128
),
"vpp-integ": (
VppEnum.vl_api_ipsec_integ_alg_t.IPSEC_API_INTEG_ALG_NONE
),
"scapy-crypto": "AES-NULL-GMAC",
"scapy-integ": "NULL",
"key": b"JPjyOWBeVEQiMe7h",
"salt": 0,
},
"AES-NULL-GMAC-192/NONE": {
"vpp-crypto": (
VppEnum.vl_api_ipsec_crypto_alg_t.IPSEC_API_CRYPTO_ALG_AES_NULL_GMAC_192
),
"vpp-integ": (
VppEnum.vl_api_ipsec_integ_alg_t.IPSEC_API_INTEG_ALG_NONE
),
"scapy-crypto": "AES-NULL-GMAC",
"scapy-integ": "NULL",
"key": b"JPjyOWBeVEQiMe7h01234567",
"salt": 1010,
},
"AES-NULL-GMAC-256/NONE": {
"vpp-crypto": (
VppEnum.vl_api_ipsec_crypto_alg_t.IPSEC_API_CRYPTO_ALG_AES_NULL_GMAC_256
),
"vpp-integ": (
VppEnum.vl_api_ipsec_integ_alg_t.IPSEC_API_INTEG_ALG_NONE
),
"scapy-crypto": "AES-NULL-GMAC",
"scapy-integ": "NULL",
"key": b"JPjyOWBeVEQiMe7h0123456787654321",
"salt": 2020,
},
}
@unittest.skipIf(
"ping" in config.excluded_plugins, "Exclude tests requiring Ping plugin"
)
class RunTestIpsecEspAll(ConfigIpsecESP, IpsecTra4, IpsecTra6, IpsecTun4, IpsecTun6):
"""Ipsec ESP all Algos"""
@classmethod
def setUpConstants(cls):
test_args = str.split(cls.__doc__, " ")
engine = test_args[0]
if engine == "async":
cls.vpp_worker_count = 2
super(RunTestIpsecEspAll, cls).setUpConstants()
def setUp(self):
super(RunTestIpsecEspAll, self).setUp()
test_args = str.split(self.__doc__, " ")
params = MyParameters()
self.engine = test_args[0]
self.flag = params.flags[0]
if test_args[1] == "ESNon":
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.flag |= params.flags[1]
if test_args[2] == "ARon":
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.flag |= params.flags[2]
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.algo = params.algos[test_args[3]]
self.async_mode = False
if self.engine == "async":
self.async_mode = True
def tearDown(self):
super(RunTestIpsecEspAll, self).tearDown()
def run_test(self):
self.run_a_test(self.engine, self.flag, self.algo)
def run_a_test(self, engine, flag, algo, payload_size=None):
if self.async_mode:
self.vapi.cli("set ipsec async mode on")
else:
self.vapi.cli("set crypto handler all %s" % engine)
self.logger.info(self.vapi.cli("show crypto async status"))
self.ipv4_params = IPsecIPv4Params()
self.ipv6_params = IPsecIPv6Params()
self.params = {
self.ipv4_params.addr_type: self.ipv4_params,
self.ipv6_params.addr_type: self.ipv6_params,
}
for _, p in self.params.items():
p.auth_algo_vpp_id = algo["vpp-integ"]
p.crypt_algo_vpp_id = algo["vpp-crypto"]
p.crypt_algo = algo["scapy-crypto"]
p.auth_algo = algo["scapy-integ"]
p.crypt_key = algo["key"]
p.salt = algo["salt"]
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
p.flags = flag
p.outer_flow_label = 243224
p.async_mode = self.async_mode
self.reporter.send_keep_alive(self)
#
# configure the SPDs. SAs, etc
#
self.config_network(self.params.values())
#
# run some traffic.
# An exhautsive 4o6, 6o4 is not necessary
# for each algo
#
self.verify_tra_basic6(count=NUM_PKTS)
self.verify_tra_basic4(count=NUM_PKTS)
self.verify_tun_66(self.params[socket.AF_INET6], count=NUM_PKTS)
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
#
# Use an odd-byte payload size to check for correct padding.
#
# 49 + 2 == 51 which should pad +1 to 52 for 4 byte alignment, +5
# to 56 for 8 byte alignment, and +13 to 64 for 64 byte alignment.
# This should catch bugs where the code is incorrectly over-padding
# for algorithms that don't require it
psz = 49 - len(IP() / ICMP()) if payload_size is None else payload_size
self.verify_tun_44(
self.params[socket.AF_INET], count=NUM_PKTS, payload_size=psz
)
LARGE_PKT_SZ = [
1970, # results in 2 chained buffers entering decrypt node
# but leaving as simple buffer due to ICV removal (tra4)
2004, # footer+ICV will be added to 2nd buffer (tun4)
4010, # ICV ends up splitted across 2 buffers in esp_decrypt
# for transport4; transport6 takes normal path
4020, # same as above but tra4 and tra6 are switched
]
if self.engine in engines_supporting_chain_bufs:
for sz in LARGE_PKT_SZ:
self.verify_tra_basic4(count=NUM_PKTS, payload_size=sz)
self.verify_tra_basic6(count=NUM_PKTS, payload_size=sz)
self.verify_tun_66(
self.params[socket.AF_INET6], count=NUM_PKTS, payload_size=sz
)
self.verify_tun_44(
self.params[socket.AF_INET], count=NUM_PKTS, payload_size=sz
)
#
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
# reconfigure the network and SA to run the
# anti replay tests
#
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
saf = VppEnum.vl_api_ipsec_sad_flags_t
if flag & saf.IPSEC_API_SAD_FLAG_USE_ANTI_REPLAY:
for anti_replay_window_size in (
64,
131072,
):
self.unconfig_network()
self.config_anti_replay(self.params.values(), anti_replay_window_size)
self.config_network(self.params.values())
self.verify_tra_anti_replay()
self.verify_tra_anti_replay_algorithm()
self.config_anti_replay(self.params.values())
self.unconfig_network()
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.config_network(self.params.values())
self.verify_hi_seq_num()
self.unconfig_network()
self.config_network(self.params.values())
self.verify_tra_lost()
#
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
# swap the handlers while SAs are up
#
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.unconfig_network()
self.config_network(self.params.values())
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
for e in engines:
if e != engine:
self.vapi.cli("set crypto handler all %s" % e)
self.verify_tra_basic4(count=NUM_PKTS)
self.unconfig_network()
#
# To generate test classes, do:
# grep '# GEN' test_ipsec_esp.py | sed -e 's/# GEN //g' | bash
#
# GEN for ENG in native ipsecmb openssl; do \
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
# GEN for ESN in ESNon ESNoff; do for AR in ARon ARoff; do \
# GEN for ALG in AES-GCM-128/NONE \
# GEN AES-GCM-192/NONE AES-GCM-256/NONE AES-CBC-128/MD5-96 \
# GEN AES-CBC-192/SHA1-96 AES-CBC-256/SHA1-96 \
# GEN 3DES-CBC/SHA1-96 NONE/SHA1-96 \
# GEN AES-CTR-128/SHA1-96 AES-CTR-192/SHA1-96 AES-CTR-256/SHA1-96 \
# GEN AES-NULL-GMAC-128/NONE AES-NULL-GMAC-192/NONE AES-NULL-GMAC-256/NONE; do \
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
# GEN echo -en "\n\nclass "
# GEN echo -e "Test_${ENG}_${ESN}_${AR}_${ALG}(RunTestIpsecEspAll):" |
# GEN sed -e 's/-/_/g' -e 's#/#_#g' ;
# GEN echo ' """'$ENG $ESN $AR $ALG IPSec test'"""' ;
# GEN echo " def test_ipsec(self):";
ipsec: Fix setting the hi-sequence number for decrypt Type: fix two problems; 1 - just because anti-reply is not enabled doesn't mean the high sequence number should not be used. - fix, there needs to be some means to detect a wrapped packet, so we use a window size of 2^30. 2 - The SA object was used as a scratch pad for the high-sequence number used during decryption. That means that once the batch has been processed the high-sequence number used is lost. This means it is not possible to distinguish this case: if (seq < IPSEC_SA_ANTI_REPLAY_WINDOW_LOWER_BOUND (tl)) { ... if (post_decrypt) { if (hi_seq_used == sa->seq_hi) /* the high sequence number used to succesfully decrypt this * packet is the same as the last-sequnence number of the SA. * that means this packet did not cause a wrap. * this packet is thus out of window and should be dropped */ return 1; else /* The packet decrypted with a different high sequence number * to the SA, that means it is the wrap packet and should be * accepted */ return 0; } - fix: don't use the SA as a scratch pad, use the 'packet_data' - the same place that is used as the scratch pad for the low sequence number. other consequences: - An SA doesn't have seq and last_seq, it has only seq; the sequence numnber of the last packet tx'd or rx'd. - there's 64bits of space available on the SA's first cache line. move the AES CTR mode IV there. - test the ESN/AR combinations to catch the bugs this fixes. This doubles the amount of tests, but without AR on they only run for 2 seconds. In the AR tests, the time taken to wait for packets that won't arrive is dropped from 1 to 0.2 seconds thus reducing the runtime of these tests from 10-15 to about 5 sceonds. Signed-off-by: Neale Ranns <neale@graphiant.com> Change-Id: Iaac78905289a272dc01930d70decd8109cf5e7a5
2021-06-28 13:31:28 +00:00
# GEN echo ' """'$ENG $ESN $AR $ALG IPSec test'"""' ;
# GEN echo " self.run_test()";
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
# GEN done; done; done; done
#
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
# GEN for ESN in ESNon ESNoff; do for AR in ARon ARoff; do \
# GEN for ALG in \
# GEN AES-GCM-128/NONE AES-GCM-192/NONE AES-GCM-256/NONE \
# GEN AES-CBC-192/SHA1-96 AES-CBC-256/SHA1-96; do \
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
# GEN echo -en "\n\nclass "
# GEN echo -e "Test_async_${ESN}_${AR}_${ALG}(RunTestIpsecEspAll):" |
# GEN sed -e 's/-/_/g' -e 's#/#_#g' ;
# GEN echo ' """'async $ESN $AR $ALG IPSec test'"""' ;
# GEN echo " def test_ipsec(self):";
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
# GEN echo ' """'async $ESN $AR $ALG IPSec test'"""' ;
# GEN echo " self.run_test()";
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
# GEN done; done; done;
class Test_native_ESNon_ARon_AES_GCM_128_NONE(RunTestIpsecEspAll):
"""native ESNon ARon AES-GCM-128/NONE IPSec test"""
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 test_ipsec(self):
"""native ESNon ARon AES-GCM-128/NONE IPSec test"""
self.run_test()
class Test_native_ESNon_ARon_AES_GCM_192_NONE(RunTestIpsecEspAll):
"""native ESNon ARon AES-GCM-192/NONE IPSec test"""
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 test_ipsec(self):
"""native ESNon ARon AES-GCM-192/NONE IPSec test"""
self.run_test()
class Test_native_ESNon_ARon_AES_GCM_256_NONE(RunTestIpsecEspAll):
"""native ESNon ARon AES-GCM-256/NONE IPSec test"""
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 test_ipsec(self):
"""native ESNon ARon AES-GCM-256/NONE IPSec test"""
self.run_test()
class Test_native_ESNon_ARon_AES_CBC_128_MD5_96(RunTestIpsecEspAll):
"""native ESNon ARon AES-CBC-128/MD5-96 IPSec test"""
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 test_ipsec(self):
"""native ESNon ARon AES-CBC-128/MD5-96 IPSec test"""
self.run_test()
class Test_native_ESNon_ARon_AES_CBC_192_SHA1_96(RunTestIpsecEspAll):
"""native ESNon ARon AES-CBC-192/SHA1-96 IPSec test"""
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 test_ipsec(self):
"""native ESNon ARon AES-CBC-192/SHA1-96 IPSec test"""
self.run_test()
class Test_native_ESNon_ARon_AES_CBC_256_SHA1_96(RunTestIpsecEspAll):
"""native ESNon ARon AES-CBC-256/SHA1-96 IPSec test"""
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 test_ipsec(self):
"""native ESNon ARon AES-CBC-256/SHA1-96 IPSec test"""
self.run_test()
class Test_native_ESNon_ARon_3DES_CBC_SHA1_96(RunTestIpsecEspAll):
"""native ESNon ARon 3DES-CBC/SHA1-96 IPSec test"""
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 test_ipsec(self):
"""native ESNon ARon 3DES-CBC/SHA1-96 IPSec test"""
self.run_test()
class Test_native_ESNon_ARon_NONE_SHA1_96(RunTestIpsecEspAll):
"""native ESNon ARon NONE/SHA1-96 IPSec test"""
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 test_ipsec(self):
"""native ESNon ARon NONE/SHA1-96 IPSec test"""
self.run_test()
class Test_native_ESNon_ARon_AES_CTR_128_SHA1_96(RunTestIpsecEspAll):
"""native ESNon ARon AES-CTR-128/SHA1-96 IPSec test"""
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 test_ipsec(self):
"""native ESNon ARon AES-CTR-128/SHA1-96 IPSec test"""
self.run_test()
class Test_native_ESNon_ARon_AES_CTR_192_SHA1_96(RunTestIpsecEspAll):
"""native ESNon ARon AES-CTR-192/SHA1-96 IPSec test"""
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 test_ipsec(self):
"""native ESNon ARon AES-CTR-192/SHA1-96 IPSec test"""
self.run_test()
class Test_native_ESNon_ARon_AES_CTR_256_SHA1_96(RunTestIpsecEspAll):
"""native ESNon ARon AES-CTR-256/SHA1-96 IPSec test"""
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 test_ipsec(self):
"""native ESNon ARon AES-CTR-256/SHA1-96 IPSec test"""
self.run_test()
class Test_native_ESNon_ARoff_AES_GCM_128_NONE(RunTestIpsecEspAll):
"""native ESNon ARoff AES-GCM-128/NONE IPSec test"""
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 test_ipsec(self):
"""native ESNon ARoff AES-GCM-128/NONE IPSec test"""
self.run_test()
class Test_native_ESNon_ARoff_AES_GCM_192_NONE(RunTestIpsecEspAll):
"""native ESNon ARoff AES-GCM-192/NONE IPSec test"""
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 test_ipsec(self):
"""native ESNon ARoff AES-GCM-192/NONE IPSec test"""
self.run_test()
class Test_native_ESNon_ARoff_AES_GCM_256_NONE(RunTestIpsecEspAll):
"""native ESNon ARoff AES-GCM-256/NONE IPSec test"""
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 test_ipsec(self):
"""native ESNon ARoff AES-GCM-256/NONE IPSec test"""
self.run_test()
class Test_native_ESNon_ARoff_AES_CBC_128_MD5_96(RunTestIpsecEspAll):
"""native ESNon ARoff AES-CBC-128/MD5-96 IPSec test"""
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 test_ipsec(self):
"""native ESNon ARoff AES-CBC-128/MD5-96 IPSec test"""
self.run_test()
class Test_native_ESNon_ARoff_AES_CBC_192_SHA1_96(RunTestIpsecEspAll):
"""native ESNon ARoff AES-CBC-192/SHA1-96 IPSec test"""
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 test_ipsec(self):
"""native ESNon ARoff AES-CBC-192/SHA1-96 IPSec test"""
self.run_test()
class Test_native_ESNon_ARoff_AES_CBC_256_SHA1_96(RunTestIpsecEspAll):
"""native ESNon ARoff AES-CBC-256/SHA1-96 IPSec test"""
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 test_ipsec(self):
"""native ESNon ARoff AES-CBC-256/SHA1-96 IPSec test"""
self.run_test()
class Test_native_ESNon_ARoff_3DES_CBC_SHA1_96(RunTestIpsecEspAll):
"""native ESNon ARoff 3DES-CBC/SHA1-96 IPSec test"""
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 test_ipsec(self):
"""native ESNon ARoff 3DES-CBC/SHA1-96 IPSec test"""
self.run_test()
class Test_native_ESNon_ARoff_NONE_SHA1_96(RunTestIpsecEspAll):
"""native ESNon ARoff NONE/SHA1-96 IPSec test"""
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 test_ipsec(self):
"""native ESNon ARoff NONE/SHA1-96 IPSec test"""
self.run_test()
class Test_native_ESNon_ARoff_AES_CTR_128_SHA1_96(RunTestIpsecEspAll):
"""native ESNon ARoff AES-CTR-128/SHA1-96 IPSec test"""
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 test_ipsec(self):
"""native ESNon ARoff AES-CTR-128/SHA1-96 IPSec test"""
self.run_test()
class Test_native_ESNon_ARoff_AES_CTR_192_SHA1_96(RunTestIpsecEspAll):
"""native ESNon ARoff AES-CTR-192/SHA1-96 IPSec test"""
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 test_ipsec(self):
"""native ESNon ARoff AES-CTR-192/SHA1-96 IPSec test"""
self.run_test()
class Test_native_ESNon_ARoff_AES_CTR_256_SHA1_96(RunTestIpsecEspAll):
"""native ESNon ARoff AES-CTR-256/SHA1-96 IPSec test"""
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 test_ipsec(self):
"""native ESNon ARoff AES-CTR-256/SHA1-96 IPSec test"""
self.run_test()
class Test_native_ESNoff_ARon_AES_GCM_128_NONE(RunTestIpsecEspAll):
"""native ESNoff ARon AES-GCM-128/NONE IPSec test"""
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 test_ipsec(self):
"""native ESNoff ARon AES-GCM-128/NONE IPSec test"""
self.run_test()
class Test_native_ESNoff_ARon_AES_GCM_192_NONE(RunTestIpsecEspAll):
"""native ESNoff ARon AES-GCM-192/NONE IPSec test"""
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 test_ipsec(self):
"""native ESNoff ARon AES-GCM-192/NONE IPSec test"""
self.run_test()
class Test_native_ESNoff_ARon_AES_GCM_256_NONE(RunTestIpsecEspAll):
"""native ESNoff ARon AES-GCM-256/NONE IPSec test"""
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 test_ipsec(self):
"""native ESNoff ARon AES-GCM-256/NONE IPSec test"""
self.run_test()
class Test_native_ESNoff_ARon_AES_CBC_128_MD5_96(RunTestIpsecEspAll):
"""native ESNoff ARon AES-CBC-128/MD5-96 IPSec test"""
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 test_ipsec(self):
"""native ESNoff ARon AES-CBC-128/MD5-96 IPSec test"""
self.run_test()
class Test_native_ESNoff_ARon_AES_CBC_192_SHA1_96(RunTestIpsecEspAll):
"""native ESNoff ARon AES-CBC-192/SHA1-96 IPSec test"""
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 test_ipsec(self):
"""native ESNoff ARon AES-CBC-192/SHA1-96 IPSec test"""
self.run_test()
class Test_native_ESNoff_ARon_AES_CBC_256_SHA1_96(RunTestIpsecEspAll):
"""native ESNoff ARon AES-CBC-256/SHA1-96 IPSec test"""
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 test_ipsec(self):
"""native ESNoff ARon AES-CBC-256/SHA1-96 IPSec test"""
self.run_test()
class Test_native_ESNoff_ARon_3DES_CBC_SHA1_96(RunTestIpsecEspAll):
"""native ESNoff ARon 3DES-CBC/SHA1-96 IPSec test"""
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 test_ipsec(self):
"""native ESNoff ARon 3DES-CBC/SHA1-96 IPSec test"""
self.run_test()
class Test_native_ESNoff_ARon_NONE_SHA1_96(RunTestIpsecEspAll):
"""native ESNoff ARon NONE/SHA1-96 IPSec test"""
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 test_ipsec(self):
"""native ESNoff ARon NONE/SHA1-96 IPSec test"""
self.run_test()
class Test_native_ESNoff_ARon_AES_CTR_128_SHA1_96(RunTestIpsecEspAll):
"""native ESNoff ARon AES-CTR-128/SHA1-96 IPSec test"""
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 test_ipsec(self):
"""native ESNoff ARon AES-CTR-128/SHA1-96 IPSec test"""
self.run_test()
class Test_native_ESNoff_ARon_AES_CTR_192_SHA1_96(RunTestIpsecEspAll):
"""native ESNoff ARon AES-CTR-192/SHA1-96 IPSec test"""
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 test_ipsec(self):
"""native ESNoff ARon AES-CTR-192/SHA1-96 IPSec test"""
self.run_test()
class Test_native_ESNoff_ARon_AES_CTR_256_SHA1_96(RunTestIpsecEspAll):
"""native ESNoff ARon AES-CTR-256/SHA1-96 IPSec test"""
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 test_ipsec(self):
"""native ESNoff ARon AES-CTR-256/SHA1-96 IPSec test"""
self.run_test()
class Test_native_ESNoff_ARoff_AES_GCM_128_NONE(RunTestIpsecEspAll):
"""native ESNoff ARoff AES-GCM-128/NONE IPSec test"""
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 test_ipsec(self):
"""native ESNoff ARoff AES-GCM-128/NONE IPSec test"""
self.run_test()
class Test_native_ESNoff_ARoff_AES_GCM_192_NONE(RunTestIpsecEspAll):
"""native ESNoff ARoff AES-GCM-192/NONE IPSec test"""
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 test_ipsec(self):
"""native ESNoff ARoff AES-GCM-192/NONE IPSec test"""
self.run_test()
class Test_native_ESNoff_ARoff_AES_GCM_256_NONE(RunTestIpsecEspAll):
"""native ESNoff ARoff AES-GCM-256/NONE IPSec test"""
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 test_ipsec(self):
"""native ESNoff ARoff AES-GCM-256/NONE IPSec test"""
self.run_test()
class Test_native_ESNoff_ARoff_AES_CBC_128_MD5_96(RunTestIpsecEspAll):
"""native ESNoff ARoff AES-CBC-128/MD5-96 IPSec test"""
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 test_ipsec(self):
"""native ESNoff ARoff AES-CBC-128/MD5-96 IPSec test"""
self.run_test()
class Test_native_ESNoff_ARoff_AES_CBC_192_SHA1_96(RunTestIpsecEspAll):
"""native ESNoff ARoff AES-CBC-192/SHA1-96 IPSec test"""
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 test_ipsec(self):
"""native ESNoff ARoff AES-CBC-192/SHA1-96 IPSec test"""
self.run_test()
class Test_native_ESNoff_ARoff_AES_CBC_256_SHA1_96(RunTestIpsecEspAll):
"""native ESNoff ARoff AES-CBC-256/SHA1-96 IPSec test"""
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 test_ipsec(self):
"""native ESNoff ARoff AES-CBC-256/SHA1-96 IPSec test"""
self.run_test()
class Test_native_ESNoff_ARoff_3DES_CBC_SHA1_96(RunTestIpsecEspAll):
"""native ESNoff ARoff 3DES-CBC/SHA1-96 IPSec test"""
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 test_ipsec(self):
"""native ESNoff ARoff 3DES-CBC/SHA1-96 IPSec test"""
self.run_test()
class Test_native_ESNoff_ARoff_NONE_SHA1_96(RunTestIpsecEspAll):
"""native ESNoff ARoff NONE/SHA1-96 IPSec test"""
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 test_ipsec(self):
"""native ESNoff ARoff NONE/SHA1-96 IPSec test"""
self.run_test()
class Test_native_ESNoff_ARoff_AES_CTR_128_SHA1_96(RunTestIpsecEspAll):
"""native ESNoff ARoff AES-CTR-128/SHA1-96 IPSec test"""
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 test_ipsec(self):
"""native ESNoff ARoff AES-CTR-128/SHA1-96 IPSec test"""
self.run_test()
class Test_native_ESNoff_ARoff_AES_CTR_192_SHA1_96(RunTestIpsecEspAll):
"""native ESNoff ARoff AES-CTR-192/SHA1-96 IPSec test"""
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 test_ipsec(self):
"""native ESNoff ARoff AES-CTR-192/SHA1-96 IPSec test"""
self.run_test()
class Test_native_ESNoff_ARoff_AES_CTR_256_SHA1_96(RunTestIpsecEspAll):
"""native ESNoff ARoff AES-CTR-256/SHA1-96 IPSec test"""
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 test_ipsec(self):
"""native ESNoff ARoff AES-CTR-256/SHA1-96 IPSec test"""
self.run_test()
class Test_ipsecmb_ESNon_ARon_AES_GCM_128_NONE(RunTestIpsecEspAll):
"""ipsecmb ESNon ARon AES-GCM-128/NONE IPSec test"""
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 test_ipsec(self):
"""ipsecmb ESNon ARon AES-GCM-128/NONE IPSec test"""
self.run_test()
class Test_ipsecmb_ESNon_ARon_AES_GCM_192_NONE(RunTestIpsecEspAll):
"""ipsecmb ESNon ARon AES-GCM-192/NONE IPSec test"""
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 test_ipsec(self):
"""ipsecmb ESNon ARon AES-GCM-192/NONE IPSec test"""
self.run_test()
class Test_ipsecmb_ESNon_ARon_AES_GCM_256_NONE(RunTestIpsecEspAll):
"""ipsecmb ESNon ARon AES-GCM-256/NONE IPSec test"""
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 test_ipsec(self):
"""ipsecmb ESNon ARon AES-GCM-256/NONE IPSec test"""
self.run_test()
class Test_ipsecmb_ESNon_ARon_AES_CBC_128_MD5_96(RunTestIpsecEspAll):
"""ipsecmb ESNon ARon AES-CBC-128/MD5-96 IPSec test"""
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 test_ipsec(self):
"""ipsecmb ESNon ARon AES-CBC-128/MD5-96 IPSec test"""
self.run_test()
class Test_ipsecmb_ESNon_ARon_AES_CBC_192_SHA1_96(RunTestIpsecEspAll):
"""ipsecmb ESNon ARon AES-CBC-192/SHA1-96 IPSec test"""
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 test_ipsec(self):
"""ipsecmb ESNon ARon AES-CBC-192/SHA1-96 IPSec test"""
self.run_test()
class Test_ipsecmb_ESNon_ARon_AES_CBC_256_SHA1_96(RunTestIpsecEspAll):
"""ipsecmb ESNon ARon AES-CBC-256/SHA1-96 IPSec test"""
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 test_ipsec(self):
"""ipsecmb ESNon ARon AES-CBC-256/SHA1-96 IPSec test"""
self.run_test()
class Test_ipsecmb_ESNon_ARon_3DES_CBC_SHA1_96(RunTestIpsecEspAll):
"""ipsecmb ESNon ARon 3DES-CBC/SHA1-96 IPSec test"""
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 test_ipsec(self):
"""ipsecmb ESNon ARon 3DES-CBC/SHA1-96 IPSec test"""
self.run_test()
class Test_ipsecmb_ESNon_ARon_NONE_SHA1_96(RunTestIpsecEspAll):
"""ipsecmb ESNon ARon NONE/SHA1-96 IPSec test"""
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 test_ipsec(self):
"""ipsecmb ESNon ARon NONE/SHA1-96 IPSec test"""
self.run_test()
class Test_ipsecmb_ESNon_ARon_AES_CTR_128_SHA1_96(RunTestIpsecEspAll):
"""ipsecmb ESNon ARon AES-CTR-128/SHA1-96 IPSec test"""
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 test_ipsec(self):
"""ipsecmb ESNon ARon AES-CTR-128/SHA1-96 IPSec test"""
self.run_test()
class Test_ipsecmb_ESNon_ARon_AES_CTR_192_SHA1_96(RunTestIpsecEspAll):
"""ipsecmb ESNon ARon AES-CTR-192/SHA1-96 IPSec test"""
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 test_ipsec(self):
"""ipsecmb ESNon ARon AES-CTR-192/SHA1-96 IPSec test"""
self.run_test()
class Test_ipsecmb_ESNon_ARon_AES_CTR_256_SHA1_96(RunTestIpsecEspAll):
"""ipsecmb ESNon ARon AES-CTR-256/SHA1-96 IPSec test"""
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 test_ipsec(self):
"""ipsecmb ESNon ARon AES-CTR-256/SHA1-96 IPSec test"""
self.run_test()
class Test_ipsecmb_ESNon_ARoff_AES_GCM_128_NONE(RunTestIpsecEspAll):
"""ipsecmb ESNon ARoff AES-GCM-128/NONE IPSec test"""
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 test_ipsec(self):
"""ipsecmb ESNon ARoff AES-GCM-128/NONE IPSec test"""
self.run_test()
class Test_ipsecmb_ESNon_ARoff_AES_GCM_192_NONE(RunTestIpsecEspAll):
"""ipsecmb ESNon ARoff AES-GCM-192/NONE IPSec test"""
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 test_ipsec(self):
"""ipsecmb ESNon ARoff AES-GCM-192/NONE IPSec test"""
self.run_test()
class Test_ipsecmb_ESNon_ARoff_AES_GCM_256_NONE(RunTestIpsecEspAll):
"""ipsecmb ESNon ARoff AES-GCM-256/NONE IPSec test"""
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 test_ipsec(self):
"""ipsecmb ESNon ARoff AES-GCM-256/NONE IPSec test"""
self.run_test()
class Test_ipsecmb_ESNon_ARoff_AES_CBC_128_MD5_96(RunTestIpsecEspAll):
"""ipsecmb ESNon ARoff AES-CBC-128/MD5-96 IPSec test"""
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 test_ipsec(self):
"""ipsecmb ESNon ARoff AES-CBC-128/MD5-96 IPSec test"""
self.run_test()
class Test_ipsecmb_ESNon_ARoff_AES_CBC_192_SHA1_96(RunTestIpsecEspAll):
"""ipsecmb ESNon ARoff AES-CBC-192/SHA1-96 IPSec test"""
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 test_ipsec(self):
"""ipsecmb ESNon ARoff AES-CBC-192/SHA1-96 IPSec test"""
self.run_test()
class Test_ipsecmb_ESNon_ARoff_AES_CBC_256_SHA1_96(RunTestIpsecEspAll):
"""ipsecmb ESNon ARoff AES-CBC-256/SHA1-96 IPSec test"""
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 test_ipsec(self):
"""ipsecmb ESNon ARoff AES-CBC-256/SHA1-96 IPSec test"""
self.run_test()
class Test_ipsecmb_ESNon_ARoff_3DES_CBC_SHA1_96(RunTestIpsecEspAll):
"""ipsecmb ESNon ARoff 3DES-CBC/SHA1-96 IPSec test"""
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 test_ipsec(self):
"""ipsecmb ESNon ARoff 3DES-CBC/SHA1-96 IPSec test"""
self.run_test()
class Test_ipsecmb_ESNon_ARoff_NONE_SHA1_96(RunTestIpsecEspAll):
"""ipsecmb ESNon ARoff NONE/SHA1-96 IPSec test"""
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 test_ipsec(self):
"""ipsecmb ESNon ARoff NONE/SHA1-96 IPSec test"""
self.run_test()
class Test_ipsecmb_ESNon_ARoff_AES_CTR_128_SHA1_96(RunTestIpsecEspAll):
"""ipsecmb ESNon ARoff AES-CTR-128/SHA1-96 IPSec test"""
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 test_ipsec(self):
"""ipsecmb ESNon ARoff AES-CTR-128/SHA1-96 IPSec test"""
self.run_test()
class Test_ipsecmb_ESNon_ARoff_AES_CTR_192_SHA1_96(RunTestIpsecEspAll):
"""ipsecmb ESNon ARoff AES-CTR-192/SHA1-96 IPSec test"""
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 test_ipsec(self):
"""ipsecmb ESNon ARoff AES-CTR-192/SHA1-96 IPSec test"""
self.run_test()
class Test_ipsecmb_ESNon_ARoff_AES_CTR_256_SHA1_96(RunTestIpsecEspAll):
"""ipsecmb ESNon ARoff AES-CTR-256/SHA1-96 IPSec test"""
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 test_ipsec(self):
"""ipsecmb ESNon ARoff AES-CTR-256/SHA1-96 IPSec test"""
self.run_test()
class Test_ipsecmb_ESNoff_ARon_AES_GCM_128_NONE(RunTestIpsecEspAll):
"""ipsecmb ESNoff ARon AES-GCM-128/NONE IPSec test"""
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 test_ipsec(self):
"""ipsecmb ESNoff ARon AES-GCM-128/NONE IPSec test"""
self.run_test()
class Test_ipsecmb_ESNoff_ARon_AES_GCM_192_NONE(RunTestIpsecEspAll):
"""ipsecmb ESNoff ARon AES-GCM-192/NONE IPSec test"""
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 test_ipsec(self):
"""ipsecmb ESNoff ARon AES-GCM-192/NONE IPSec test"""
self.run_test()
class Test_ipsecmb_ESNoff_ARon_AES_GCM_256_NONE(RunTestIpsecEspAll):
"""ipsecmb ESNoff ARon AES-GCM-256/NONE IPSec test"""
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 test_ipsec(self):
"""ipsecmb ESNoff ARon AES-GCM-256/NONE IPSec test"""
self.run_test()
class Test_ipsecmb_ESNoff_ARon_AES_CBC_128_MD5_96(RunTestIpsecEspAll):
"""ipsecmb ESNoff ARon AES-CBC-128/MD5-96 IPSec test"""
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 test_ipsec(self):
"""ipsecmb ESNoff ARon AES-CBC-128/MD5-96 IPSec test"""
self.run_test()
class Test_ipsecmb_ESNoff_ARon_AES_CBC_192_SHA1_96(RunTestIpsecEspAll):
"""ipsecmb ESNoff ARon AES-CBC-192/SHA1-96 IPSec test"""
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 test_ipsec(self):
"""ipsecmb ESNoff ARon AES-CBC-192/SHA1-96 IPSec test"""
self.run_test()
class Test_ipsecmb_ESNoff_ARon_AES_CBC_256_SHA1_96(RunTestIpsecEspAll):
"""ipsecmb ESNoff ARon AES-CBC-256/SHA1-96 IPSec test"""
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 test_ipsec(self):
"""ipsecmb ESNoff ARon AES-CBC-256/SHA1-96 IPSec test"""
self.run_test()
class Test_ipsecmb_ESNoff_ARon_3DES_CBC_SHA1_96(RunTestIpsecEspAll):
"""ipsecmb ESNoff ARon 3DES-CBC/SHA1-96 IPSec test"""
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 test_ipsec(self):
"""ipsecmb ESNoff ARon 3DES-CBC/SHA1-96 IPSec test"""
self.run_test()
class Test_ipsecmb_ESNoff_ARon_NONE_SHA1_96(RunTestIpsecEspAll):
"""ipsecmb ESNoff ARon NONE/SHA1-96 IPSec test"""
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 test_ipsec(self):
"""ipsecmb ESNoff ARon NONE/SHA1-96 IPSec test"""
self.run_test()
class Test_ipsecmb_ESNoff_ARon_AES_CTR_128_SHA1_96(RunTestIpsecEspAll):
"""ipsecmb ESNoff ARon AES-CTR-128/SHA1-96 IPSec test"""
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 test_ipsec(self):
"""ipsecmb ESNoff ARon AES-CTR-128/SHA1-96 IPSec test"""
self.run_test()
class Test_ipsecmb_ESNoff_ARon_AES_CTR_192_SHA1_96(RunTestIpsecEspAll):
"""ipsecmb ESNoff ARon AES-CTR-192/SHA1-96 IPSec test"""
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 test_ipsec(self):
"""ipsecmb ESNoff ARon AES-CTR-192/SHA1-96 IPSec test"""
self.run_test()
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
class Test_ipsecmb_ESNoff_ARon_AES_CTR_256_SHA1_96(RunTestIpsecEspAll):
"""ipsecmb ESNoff ARon AES-CTR-256/SHA1-96 IPSec test"""
def test_ipsec(self):
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
"""ipsecmb ESNoff ARon AES-CTR-256/SHA1-96 IPSec test"""
self.run_test()
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
class Test_ipsecmb_ESNoff_ARoff_AES_GCM_128_NONE(RunTestIpsecEspAll):
"""ipsecmb ESNoff ARoff AES-GCM-128/NONE IPSec test"""
def test_ipsec(self):
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
"""ipsecmb ESNoff ARoff AES-GCM-128/NONE IPSec test"""
self.run_test()
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
class Test_ipsecmb_ESNoff_ARoff_AES_GCM_192_NONE(RunTestIpsecEspAll):
"""ipsecmb ESNoff ARoff AES-GCM-192/NONE IPSec test"""
def test_ipsec(self):
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
"""ipsecmb ESNoff ARoff AES-GCM-192/NONE IPSec test"""
self.run_test()
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
class Test_ipsecmb_ESNoff_ARoff_AES_GCM_256_NONE(RunTestIpsecEspAll):
"""ipsecmb ESNoff ARoff AES-GCM-256/NONE IPSec test"""
def test_ipsec(self):
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
"""ipsecmb ESNoff ARoff AES-GCM-256/NONE IPSec test"""
self.run_test()
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
class Test_ipsecmb_ESNoff_ARoff_AES_CBC_128_MD5_96(RunTestIpsecEspAll):
"""ipsecmb ESNoff ARoff AES-CBC-128/MD5-96 IPSec test"""
def test_ipsec(self):
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
"""ipsecmb ESNoff ARoff AES-CBC-128/MD5-96 IPSec test"""
self.run_test()
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
class Test_ipsecmb_ESNoff_ARoff_AES_CBC_192_SHA1_96(RunTestIpsecEspAll):
"""ipsecmb ESNoff ARoff AES-CBC-192/SHA1-96 IPSec test"""
def test_ipsec(self):
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
"""ipsecmb ESNoff ARoff AES-CBC-192/SHA1-96 IPSec test"""
self.run_test()
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
class Test_ipsecmb_ESNoff_ARoff_AES_CBC_256_SHA1_96(RunTestIpsecEspAll):
"""ipsecmb ESNoff ARoff AES-CBC-256/SHA1-96 IPSec test"""
def test_ipsec(self):
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
"""ipsecmb ESNoff ARoff AES-CBC-256/SHA1-96 IPSec test"""
self.run_test()
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
class Test_ipsecmb_ESNoff_ARoff_3DES_CBC_SHA1_96(RunTestIpsecEspAll):
"""ipsecmb ESNoff ARoff 3DES-CBC/SHA1-96 IPSec test"""
def test_ipsec(self):
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
"""ipsecmb ESNoff ARoff 3DES-CBC/SHA1-96 IPSec test"""
self.run_test()
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
class Test_ipsecmb_ESNoff_ARoff_NONE_SHA1_96(RunTestIpsecEspAll):
"""ipsecmb ESNoff ARoff NONE/SHA1-96 IPSec test"""
def test_ipsec(self):
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
"""ipsecmb ESNoff ARoff NONE/SHA1-96 IPSec test"""
self.run_test()
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
class Test_ipsecmb_ESNoff_ARoff_AES_CTR_128_SHA1_96(RunTestIpsecEspAll):
"""ipsecmb ESNoff ARoff AES-CTR-128/SHA1-96 IPSec test"""
def test_ipsec(self):
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
"""ipsecmb ESNoff ARoff AES-CTR-128/SHA1-96 IPSec test"""
self.run_test()
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
class Test_ipsecmb_ESNoff_ARoff_AES_CTR_192_SHA1_96(RunTestIpsecEspAll):
"""ipsecmb ESNoff ARoff AES-CTR-192/SHA1-96 IPSec test"""
def test_ipsec(self):
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
"""ipsecmb ESNoff ARoff AES-CTR-192/SHA1-96 IPSec test"""
self.run_test()
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
class Test_ipsecmb_ESNoff_ARoff_AES_CTR_256_SHA1_96(RunTestIpsecEspAll):
"""ipsecmb ESNoff ARoff AES-CTR-256/SHA1-96 IPSec test"""
def test_ipsec(self):
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
"""ipsecmb ESNoff ARoff AES-CTR-256/SHA1-96 IPSec test"""
self.run_test()
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
class Test_openssl_ESNon_ARon_AES_GCM_128_NONE(RunTestIpsecEspAll):
"""openssl ESNon ARon AES-GCM-128/NONE IPSec test"""
def test_ipsec(self):
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
"""openssl ESNon ARon AES-GCM-128/NONE IPSec test"""
self.run_test()
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
class Test_openssl_ESNon_ARon_AES_GCM_192_NONE(RunTestIpsecEspAll):
"""openssl ESNon ARon AES-GCM-192/NONE IPSec test"""
def test_ipsec(self):
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
"""openssl ESNon ARon AES-GCM-192/NONE IPSec test"""
self.run_test()
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
class Test_openssl_ESNon_ARon_AES_GCM_256_NONE(RunTestIpsecEspAll):
"""openssl ESNon ARon AES-GCM-256/NONE IPSec test"""
def test_ipsec(self):
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
"""openssl ESNon ARon AES-GCM-256/NONE IPSec test"""
self.run_test()
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
class Test_openssl_ESNon_ARon_AES_CBC_128_MD5_96(RunTestIpsecEspAll):
"""openssl ESNon ARon AES-CBC-128/MD5-96 IPSec test"""
def test_ipsec(self):
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
"""openssl ESNon ARon AES-CBC-128/MD5-96 IPSec test"""
self.run_test()
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
class Test_openssl_ESNon_ARon_AES_CBC_192_SHA1_96(RunTestIpsecEspAll):
"""openssl ESNon ARon AES-CBC-192/SHA1-96 IPSec test"""
def test_ipsec(self):
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
"""openssl ESNon ARon AES-CBC-192/SHA1-96 IPSec test"""
self.run_test()
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
class Test_openssl_ESNon_ARon_AES_CBC_256_SHA1_96(RunTestIpsecEspAll):
"""openssl ESNon ARon AES-CBC-256/SHA1-96 IPSec test"""
def test_ipsec(self):
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
"""openssl ESNon ARon AES-CBC-256/SHA1-96 IPSec test"""
self.run_test()
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
class Test_openssl_ESNon_ARon_3DES_CBC_SHA1_96(RunTestIpsecEspAll):
"""openssl ESNon ARon 3DES-CBC/SHA1-96 IPSec test"""
def test_ipsec(self):
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
"""openssl ESNon ARon 3DES-CBC/SHA1-96 IPSec test"""
self.run_test()
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
class Test_openssl_ESNon_ARon_NONE_SHA1_96(RunTestIpsecEspAll):
"""openssl ESNon ARon NONE/SHA1-96 IPSec test"""
def test_ipsec(self):
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
"""openssl ESNon ARon NONE/SHA1-96 IPSec test"""
self.run_test()
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
class Test_openssl_ESNon_ARon_AES_CTR_128_SHA1_96(RunTestIpsecEspAll):
"""openssl ESNon ARon AES-CTR-128/SHA1-96 IPSec test"""
def test_ipsec(self):
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
"""openssl ESNon ARon AES-CTR-128/SHA1-96 IPSec test"""
self.run_test()
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
class Test_openssl_ESNon_ARon_AES_CTR_192_SHA1_96(RunTestIpsecEspAll):
"""openssl ESNon ARon AES-CTR-192/SHA1-96 IPSec test"""
def test_ipsec(self):
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
"""openssl ESNon ARon AES-CTR-192/SHA1-96 IPSec test"""
self.run_test()
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
class Test_openssl_ESNon_ARon_AES_CTR_256_SHA1_96(RunTestIpsecEspAll):
"""openssl ESNon ARon AES-CTR-256/SHA1-96 IPSec test"""
def test_ipsec(self):
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
"""openssl ESNon ARon AES-CTR-256/SHA1-96 IPSec test"""
self.run_test()
class Test_openssl_ESNon_ARon_AES_NULL_GMAC_128_NONE(RunTestIpsecEspAll):
"""openssl ESNon ARon AES-NULL-GMAC-128/NONE IPSec test"""
def test_ipsec(self):
"""openssl ESNon ARon AES-NULL-GMAC-128/NONE IPSec test"""
self.run_test()
class Test_openssl_ESNon_ARon_AES_NULL_GMAC_192_NONE(RunTestIpsecEspAll):
"""openssl ESNon ARon AES-NULL-GMAC-192/NONE IPSec test"""
def test_ipsec(self):
"""openssl ESNon ARon AES-NULL-GMAC-192/NONE IPSec test"""
self.run_test()
class Test_openssl_ESNon_ARon_AES_NULL_GMAC_256_NONE(RunTestIpsecEspAll):
"""openssl ESNon ARon AES-NULL-GMAC-256/NONE IPSec test"""
def test_ipsec(self):
"""openssl ESNon ARon AES-NULL-GMAC-256/NONE IPSec test"""
self.run_test()
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
class Test_openssl_ESNon_ARoff_AES_GCM_128_NONE(RunTestIpsecEspAll):
"""openssl ESNon ARoff AES-GCM-128/NONE IPSec test"""
def test_ipsec(self):
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
"""openssl ESNon ARoff AES-GCM-128/NONE IPSec test"""
self.run_test()
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
class Test_openssl_ESNon_ARoff_AES_GCM_192_NONE(RunTestIpsecEspAll):
"""openssl ESNon ARoff AES-GCM-192/NONE IPSec test"""
def test_ipsec(self):
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
"""openssl ESNon ARoff AES-GCM-192/NONE IPSec test"""
self.run_test()
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
class Test_openssl_ESNon_ARoff_AES_GCM_256_NONE(RunTestIpsecEspAll):
"""openssl ESNon ARoff AES-GCM-256/NONE IPSec test"""
def test_ipsec(self):
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
"""openssl ESNon ARoff AES-GCM-256/NONE IPSec test"""
self.run_test()
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
class Test_openssl_ESNon_ARoff_AES_CBC_128_MD5_96(RunTestIpsecEspAll):
"""openssl ESNon ARoff AES-CBC-128/MD5-96 IPSec test"""
def test_ipsec(self):
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
"""openssl ESNon ARoff AES-CBC-128/MD5-96 IPSec test"""
self.run_test()
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
class Test_openssl_ESNon_ARoff_AES_CBC_192_SHA1_96(RunTestIpsecEspAll):
"""openssl ESNon ARoff AES-CBC-192/SHA1-96 IPSec test"""
def test_ipsec(self):
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
"""openssl ESNon ARoff AES-CBC-192/SHA1-96 IPSec test"""
self.run_test()
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
class Test_openssl_ESNon_ARoff_AES_CBC_256_SHA1_96(RunTestIpsecEspAll):
"""openssl ESNon ARoff AES-CBC-256/SHA1-96 IPSec test"""
def test_ipsec(self):
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
"""openssl ESNon ARoff AES-CBC-256/SHA1-96 IPSec test"""
self.run_test()
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
class Test_openssl_ESNon_ARoff_3DES_CBC_SHA1_96(RunTestIpsecEspAll):
"""openssl ESNon ARoff 3DES-CBC/SHA1-96 IPSec test"""
def test_ipsec(self):
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
"""openssl ESNon ARoff 3DES-CBC/SHA1-96 IPSec test"""
self.run_test()
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
class Test_openssl_ESNon_ARoff_NONE_SHA1_96(RunTestIpsecEspAll):
"""openssl ESNon ARoff NONE/SHA1-96 IPSec test"""
def test_ipsec(self):
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
"""openssl ESNon ARoff NONE/SHA1-96 IPSec test"""
self.run_test()
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
class Test_openssl_ESNon_ARoff_AES_CTR_128_SHA1_96(RunTestIpsecEspAll):
"""openssl ESNon ARoff AES-CTR-128/SHA1-96 IPSec test"""
def test_ipsec(self):
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
"""openssl ESNon ARoff AES-CTR-128/SHA1-96 IPSec test"""
self.run_test()
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
class Test_openssl_ESNon_ARoff_AES_CTR_192_SHA1_96(RunTestIpsecEspAll):
"""openssl ESNon ARoff AES-CTR-192/SHA1-96 IPSec test"""
def test_ipsec(self):
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
"""openssl ESNon ARoff AES-CTR-192/SHA1-96 IPSec test"""
self.run_test()
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
class Test_openssl_ESNon_ARoff_AES_CTR_256_SHA1_96(RunTestIpsecEspAll):
"""openssl ESNon ARoff AES-CTR-256/SHA1-96 IPSec test"""
def test_ipsec(self):
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
"""openssl ESNon ARoff AES-CTR-256/SHA1-96 IPSec test"""
self.run_test()
class Test_openssl_ESNon_ARoff_AES_NULL_GMAC_128_NONE(RunTestIpsecEspAll):
"""openssl ESNon ARoff AES-NULL-GMAC-128/NONE IPSec test"""
def test_ipsec(self):
"""openssl ESNon ARoff AES-NULL-GMAC-128/NONE IPSec test"""
self.run_test()
class Test_openssl_ESNon_ARoff_AES_NULL_GMAC_192_NONE(RunTestIpsecEspAll):
"""openssl ESNon ARoff AES-NULL-GMAC-192/NONE IPSec test"""
def test_ipsec(self):
"""openssl ESNon ARoff AES-NULL-GMAC-192/NONE IPSec test"""
self.run_test()
class Test_openssl_ESNon_ARoff_AES_NULL_GMAC_256_NONE(RunTestIpsecEspAll):
"""openssl ESNon ARoff AES-NULL-GMAC-256/NONE IPSec test"""
def test_ipsec(self):
"""openssl ESNon ARoff AES-NULL-GMAC-256/NONE IPSec test"""
self.run_test()
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
class Test_openssl_ESNoff_ARon_AES_GCM_128_NONE(RunTestIpsecEspAll):
"""openssl ESNoff ARon AES-GCM-128/NONE IPSec test"""
def test_ipsec(self):
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
"""openssl ESNoff ARon AES-GCM-128/NONE IPSec test"""
self.run_test()
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
class Test_openssl_ESNoff_ARon_AES_GCM_192_NONE(RunTestIpsecEspAll):
"""openssl ESNoff ARon AES-GCM-192/NONE IPSec test"""
def test_ipsec(self):
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
"""openssl ESNoff ARon AES-GCM-192/NONE IPSec test"""
self.run_test()
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
class Test_openssl_ESNoff_ARon_AES_GCM_256_NONE(RunTestIpsecEspAll):
"""openssl ESNoff ARon AES-GCM-256/NONE IPSec test"""
def test_ipsec(self):
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
"""openssl ESNoff ARon AES-GCM-256/NONE IPSec test"""
self.run_test()
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
class Test_openssl_ESNoff_ARon_AES_CBC_128_MD5_96(RunTestIpsecEspAll):
"""openssl ESNoff ARon AES-CBC-128/MD5-96 IPSec test"""
def test_ipsec(self):
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
"""openssl ESNoff ARon AES-CBC-128/MD5-96 IPSec test"""
self.run_test()
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
class Test_openssl_ESNoff_ARon_AES_CBC_192_SHA1_96(RunTestIpsecEspAll):
"""openssl ESNoff ARon AES-CBC-192/SHA1-96 IPSec test"""
def test_ipsec(self):
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
"""openssl ESNoff ARon AES-CBC-192/SHA1-96 IPSec test"""
self.run_test()
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
class Test_openssl_ESNoff_ARon_AES_CBC_256_SHA1_96(RunTestIpsecEspAll):
"""openssl ESNoff ARon AES-CBC-256/SHA1-96 IPSec test"""
def test_ipsec(self):
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
"""openssl ESNoff ARon AES-CBC-256/SHA1-96 IPSec test"""
self.run_test()
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
class Test_openssl_ESNoff_ARon_3DES_CBC_SHA1_96(RunTestIpsecEspAll):
"""openssl ESNoff ARon 3DES-CBC/SHA1-96 IPSec test"""
def test_ipsec(self):
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
"""openssl ESNoff ARon 3DES-CBC/SHA1-96 IPSec test"""
self.run_test()
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
class Test_openssl_ESNoff_ARon_NONE_SHA1_96(RunTestIpsecEspAll):
"""openssl ESNoff ARon NONE/SHA1-96 IPSec test"""
def test_ipsec(self):
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
"""openssl ESNoff ARon NONE/SHA1-96 IPSec test"""
self.run_test()
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
class Test_openssl_ESNoff_ARon_AES_CTR_128_SHA1_96(RunTestIpsecEspAll):
"""openssl ESNoff ARon AES-CTR-128/SHA1-96 IPSec test"""
def test_ipsec(self):
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
"""openssl ESNoff ARon AES-CTR-128/SHA1-96 IPSec test"""
self.run_test()
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
class Test_openssl_ESNoff_ARon_AES_CTR_192_SHA1_96(RunTestIpsecEspAll):
"""openssl ESNoff ARon AES-CTR-192/SHA1-96 IPSec test"""
def test_ipsec(self):
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
"""openssl ESNoff ARon AES-CTR-192/SHA1-96 IPSec test"""
self.run_test()
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
class Test_openssl_ESNoff_ARon_AES_CTR_256_SHA1_96(RunTestIpsecEspAll):
"""openssl ESNoff ARon AES-CTR-256/SHA1-96 IPSec test"""
def test_ipsec(self):
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
"""openssl ESNoff ARon AES-CTR-256/SHA1-96 IPSec test"""
self.run_test()
class Test_openssl_ESNoff_ARon_AES_NULL_GMAC_128_NONE(RunTestIpsecEspAll):
"""openssl ESNoff ARon AES-NULL-GMAC-128/NONE IPSec test"""
def test_ipsec(self):
"""openssl ESNoff ARon AES-NULL-GMAC-128/NONE IPSec test"""
self.run_test()
class Test_openssl_ESNoff_ARon_AES_NULL_GMAC_192_NONE(RunTestIpsecEspAll):
"""openssl ESNoff ARon AES-NULL-GMAC-192/NONE IPSec test"""
def test_ipsec(self):
"""openssl ESNoff ARon AES-NULL-GMAC-192/NONE IPSec test"""
self.run_test()
class Test_openssl_ESNoff_ARon_AES_NULL_GMAC_256_NONE(RunTestIpsecEspAll):
"""openssl ESNoff ARon AES-NULL-GMAC-256/NONE IPSec test"""
def test_ipsec(self):
"""openssl ESNoff ARon AES-NULL-GMAC-256/NONE IPSec test"""
self.run_test()
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
class Test_openssl_ESNoff_ARoff_AES_GCM_128_NONE(RunTestIpsecEspAll):
"""openssl ESNoff ARoff AES-GCM-128/NONE IPSec test"""
def test_ipsec(self):
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
"""openssl ESNoff ARoff AES-GCM-128/NONE IPSec test"""
self.run_test()
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
class Test_openssl_ESNoff_ARoff_AES_GCM_192_NONE(RunTestIpsecEspAll):
"""openssl ESNoff ARoff AES-GCM-192/NONE IPSec test"""
def test_ipsec(self):
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
"""openssl ESNoff ARoff AES-GCM-192/NONE IPSec test"""
self.run_test()
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
class Test_openssl_ESNoff_ARoff_AES_GCM_256_NONE(RunTestIpsecEspAll):
"""openssl ESNoff ARoff AES-GCM-256/NONE IPSec test"""
def test_ipsec(self):
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
"""openssl ESNoff ARoff AES-GCM-256/NONE IPSec test"""
self.run_test()
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
class Test_openssl_ESNoff_ARoff_AES_CBC_128_MD5_96(RunTestIpsecEspAll):
"""openssl ESNoff ARoff AES-CBC-128/MD5-96 IPSec test"""
def test_ipsec(self):
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
"""openssl ESNoff ARoff AES-CBC-128/MD5-96 IPSec test"""
self.run_test()
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
class Test_openssl_ESNoff_ARoff_AES_CBC_192_SHA1_96(RunTestIpsecEspAll):
"""openssl ESNoff ARoff AES-CBC-192/SHA1-96 IPSec test"""
def test_ipsec(self):
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
"""openssl ESNoff ARoff AES-CBC-192/SHA1-96 IPSec test"""
self.run_test()
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
class Test_openssl_ESNoff_ARoff_AES_CBC_256_SHA1_96(RunTestIpsecEspAll):
"""openssl ESNoff ARoff AES-CBC-256/SHA1-96 IPSec test"""
def test_ipsec(self):
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
"""openssl ESNoff ARoff AES-CBC-256/SHA1-96 IPSec test"""
self.run_test()
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
class Test_openssl_ESNoff_ARoff_3DES_CBC_SHA1_96(RunTestIpsecEspAll):
"""openssl ESNoff ARoff 3DES-CBC/SHA1-96 IPSec test"""
def test_ipsec(self):
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
"""openssl ESNoff ARoff 3DES-CBC/SHA1-96 IPSec test"""
self.run_test()
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
class Test_openssl_ESNoff_ARoff_NONE_SHA1_96(RunTestIpsecEspAll):
"""openssl ESNoff ARoff NONE/SHA1-96 IPSec test"""
def test_ipsec(self):
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
"""openssl ESNoff ARoff NONE/SHA1-96 IPSec test"""
self.run_test()
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
class Test_openssl_ESNoff_ARoff_AES_CTR_128_SHA1_96(RunTestIpsecEspAll):
"""openssl ESNoff ARoff AES-CTR-128/SHA1-96 IPSec test"""
def test_ipsec(self):
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
"""openssl ESNoff ARoff AES-CTR-128/SHA1-96 IPSec test"""
self.run_test()
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
class Test_openssl_ESNoff_ARoff_AES_CTR_192_SHA1_96(RunTestIpsecEspAll):
"""openssl ESNoff ARoff AES-CTR-192/SHA1-96 IPSec test"""
def test_ipsec(self):
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
"""openssl ESNoff ARoff AES-CTR-192/SHA1-96 IPSec test"""
self.run_test()
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
class Test_openssl_ESNoff_ARoff_AES_CTR_256_SHA1_96(RunTestIpsecEspAll):
"""openssl ESNoff ARoff AES-CTR-256/SHA1-96 IPSec test"""
def test_ipsec(self):
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
"""openssl ESNoff ARoff AES-CTR-256/SHA1-96 IPSec test"""
self.run_test()
class Test_openssl_ESNoff_ARoff_AES_NULL_GMAC_128_NONE(RunTestIpsecEspAll):
"""openssl ESNoff ARoff AES-NULL-GMAC-128/NONE IPSec test"""
def test_ipsec(self):
"""openssl ESNoff ARoff AES-NULL-GMAC-128/NONE IPSec test"""
self.run_test()
class Test_openssl_ESNoff_ARoff_AES_NULL_GMAC_192_NONE(RunTestIpsecEspAll):
"""openssl ESNoff ARoff AES-NULL-GMAC-192/NONE IPSec test"""
def test_ipsec(self):
"""openssl ESNoff ARoff AES-NULL-GMAC-192/NONE IPSec test"""
self.run_test()
class Test_openssl_ESNoff_ARoff_AES_NULL_GMAC_256_NONE(RunTestIpsecEspAll):
"""openssl ESNoff ARoff AES-NULL-GMAC-256/NONE IPSec test"""
def test_ipsec(self):
"""openssl ESNoff ARoff AES-NULL-GMAC-256/NONE IPSec test"""
self.run_test()
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
class Test_async_ESNon_ARon_AES_GCM_128_NONE(RunTestIpsecEspAll):
"""async ESNon ARon AES-GCM-128/NONE IPSec test"""
def test_ipsec(self):
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
"""async ESNon ARon AES-GCM-128/NONE IPSec test"""
self.run_test()
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
class Test_async_ESNon_ARon_AES_GCM_192_NONE(RunTestIpsecEspAll):
"""async ESNon ARon AES-GCM-192/NONE IPSec test"""
def test_ipsec(self):
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
"""async ESNon ARon AES-GCM-192/NONE IPSec test"""
self.run_test()
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
class Test_async_ESNon_ARon_AES_GCM_256_NONE(RunTestIpsecEspAll):
"""async ESNon ARon AES-GCM-256/NONE IPSec test"""
def test_ipsec(self):
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
"""async ESNon ARon AES-GCM-256/NONE IPSec test"""
self.run_test()
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
class Test_async_ESNon_ARon_AES_CBC_192_SHA1_96(RunTestIpsecEspAll):
"""async ESNon ARon AES-CBC-192/SHA1-96 IPSec test"""
def test_ipsec(self):
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
"""async ESNon ARon AES-CBC-192/SHA1-96 IPSec test"""
self.run_test()
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
class Test_async_ESNon_ARon_AES_CBC_256_SHA1_96(RunTestIpsecEspAll):
"""async ESNon ARon AES-CBC-256/SHA1-96 IPSec test"""
def test_ipsec(self):
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
"""async ESNon ARon AES-CBC-256/SHA1-96 IPSec test"""
self.run_test()
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
class Test_async_ESNon_ARoff_AES_GCM_128_NONE(RunTestIpsecEspAll):
"""async ESNon ARoff AES-GCM-128/NONE IPSec test"""
def test_ipsec(self):
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
"""async ESNon ARoff AES-GCM-128/NONE IPSec test"""
self.run_test()
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
class Test_async_ESNon_ARoff_AES_GCM_192_NONE(RunTestIpsecEspAll):
"""async ESNon ARoff AES-GCM-192/NONE IPSec test"""
def test_ipsec(self):
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
"""async ESNon ARoff AES-GCM-192/NONE IPSec test"""
self.run_test()
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
class Test_async_ESNon_ARoff_AES_GCM_256_NONE(RunTestIpsecEspAll):
"""async ESNon ARoff AES-GCM-256/NONE IPSec test"""
def test_ipsec(self):
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
"""async ESNon ARoff AES-GCM-256/NONE IPSec test"""
self.run_test()
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
class Test_async_ESNon_ARoff_AES_CBC_192_SHA1_96(RunTestIpsecEspAll):
"""async ESNon ARoff AES-CBC-192/SHA1-96 IPSec test"""
def test_ipsec(self):
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
"""async ESNon ARoff AES-CBC-192/SHA1-96 IPSec test"""
self.run_test()
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
class Test_async_ESNon_ARoff_AES_CBC_256_SHA1_96(RunTestIpsecEspAll):
"""async ESNon ARoff AES-CBC-256/SHA1-96 IPSec test"""
def test_ipsec(self):
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
"""async ESNon ARoff AES-CBC-256/SHA1-96 IPSec test"""
self.run_test()
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
class Test_async_ESNoff_ARon_AES_GCM_128_NONE(RunTestIpsecEspAll):
"""async ESNoff ARon AES-GCM-128/NONE IPSec test"""
def test_ipsec(self):
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
"""async ESNoff ARon AES-GCM-128/NONE IPSec test"""
self.run_test()
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
class Test_async_ESNoff_ARon_AES_GCM_192_NONE(RunTestIpsecEspAll):
"""async ESNoff ARon AES-GCM-192/NONE IPSec test"""
def test_ipsec(self):
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
"""async ESNoff ARon AES-GCM-192/NONE IPSec test"""
self.run_test()
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
class Test_async_ESNoff_ARon_AES_GCM_256_NONE(RunTestIpsecEspAll):
"""async ESNoff ARon AES-GCM-256/NONE IPSec test"""
def test_ipsec(self):
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
"""async ESNoff ARon AES-GCM-256/NONE IPSec test"""
self.run_test()
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
class Test_async_ESNoff_ARon_AES_CBC_192_SHA1_96(RunTestIpsecEspAll):
"""async ESNoff ARon AES-CBC-192/SHA1-96 IPSec test"""
def test_ipsec(self):
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
"""async ESNoff ARon AES-CBC-192/SHA1-96 IPSec test"""
self.run_test()
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
class Test_async_ESNoff_ARon_AES_CBC_256_SHA1_96(RunTestIpsecEspAll):
"""async ESNoff ARon AES-CBC-256/SHA1-96 IPSec test"""
def test_ipsec(self):
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
"""async ESNoff ARon AES-CBC-256/SHA1-96 IPSec test"""
self.run_test()
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
class Test_async_ESNoff_ARoff_AES_GCM_128_NONE(RunTestIpsecEspAll):
"""async ESNoff ARoff AES-GCM-128/NONE IPSec test"""
def test_ipsec(self):
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
"""async ESNoff ARoff AES-GCM-128/NONE IPSec test"""
self.run_test()
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
class Test_async_ESNoff_ARoff_AES_GCM_192_NONE(RunTestIpsecEspAll):
"""async ESNoff ARoff AES-GCM-192/NONE IPSec test"""
def test_ipsec(self):
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
"""async ESNoff ARoff AES-GCM-192/NONE IPSec test"""
self.run_test()
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
class Test_async_ESNoff_ARoff_AES_GCM_256_NONE(RunTestIpsecEspAll):
"""async ESNoff ARoff AES-GCM-256/NONE IPSec test"""
def test_ipsec(self):
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
"""async ESNoff ARoff AES-GCM-256/NONE IPSec test"""
self.run_test()
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
class Test_async_ESNoff_ARoff_AES_CBC_192_SHA1_96(RunTestIpsecEspAll):
"""async ESNoff ARoff AES-CBC-192/SHA1-96 IPSec test"""
def test_ipsec(self):
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
"""async ESNoff ARoff AES-CBC-192/SHA1-96 IPSec test"""
self.run_test()
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
class Test_async_ESNoff_ARoff_AES_CBC_256_SHA1_96(RunTestIpsecEspAll):
"""async ESNoff ARoff AES-CBC-256/SHA1-96 IPSec test"""
def test_ipsec(self):
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
"""async ESNoff ARoff AES-CBC-256/SHA1-96 IPSec test"""
self.run_test()