freebsd-nq/tests/sys/opencrypto/cryptotest.py

340 lines
10 KiB
Python
Raw Normal View History

2014-12-12 19:56:36 +00:00
#!/usr/bin/env python
#
# Copyright (c) 2014 The FreeBSD Foundation
# All rights reserved.
#
# This software was developed by John-Mark Gurney under
# the sponsorship from the FreeBSD Foundation.
# Redistribution and use in source and binary forms, with or without
# modification, are permitted provided that the following conditions
# are met:
# 1. Redistributions of source code must retain the above copyright
# notice, this list of conditions and the following disclaimer.
# 2. Redistributions in binary form must reproduce the above copyright
# notice, this list of conditions and the following disclaimer in the
# documentation and/or other materials provided with the distribution.
#
# THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
# ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
# ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
# FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
# DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
# OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
# HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
# LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
# OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
# SUCH DAMAGE.
#
# $FreeBSD$
#
from __future__ import print_function
import errno
2014-12-12 19:56:36 +00:00
import cryptodev
import itertools
import os
import struct
import unittest
from cryptodev import *
from glob import iglob
katdir = '/usr/local/share/nist-kat'
def katg(base, glob):
assert os.path.exists(os.path.join(katdir, base)), "Please 'pkg install nist-kat'"
2014-12-12 19:56:36 +00:00
return iglob(os.path.join(katdir, base, glob))
Add ccp(4): experimental driver for AMD Crypto Co-Processor * Registers TRNG source for random(4) * Finds available queues, LSBs; allocates static objects * Allocates a shared MSI-X for all queues. The hardware does not have separate interrupts per queue. Working interrupt mode driver. * Computes SHA hashes, HMAC. Passes cryptotest.py, cryptocheck tests. * Does AES-CBC, CTR mode, and XTS. cryptotest.py and cryptocheck pass. * Support for "authenc" (AES + HMAC). (SHA1 seems to result in "unaligned" cleartext inputs from cryptocheck -- which the engine cannot handle. SHA2 seems to work fine.) * GCM passes for block-multiple AAD, input lengths Largely based on ccr(4), part of cxgbe(4). Rough performance averages on AMD Ryzen 1950X (4kB buffer): aesni: SHA1: ~8300 Mb/s SHA256: ~8000 Mb/s ccp: ~630 Mb/s SHA256: ~660 Mb/s SHA512: ~700 Mb/s cryptosoft: ~1800 Mb/s SHA256: ~1800 Mb/s SHA512: ~2700 Mb/s As you can see, performance is poor in comparison to aesni(4) and even cryptosoft (due to high setup cost). At a larger buffer size (128kB), throughput is a little better (but still worse than aesni(4)): aesni: SHA1:~10400 Mb/s SHA256: ~9950 Mb/s ccp: ~2200 Mb/s SHA256: ~2600 Mb/s SHA512: ~3800 Mb/s cryptosoft: ~1750 Mb/s SHA256: ~1800 Mb/s SHA512: ~2700 Mb/s AES performance has a similar story: aesni: 4kB: ~11250 Mb/s 128kB: ~11250 Mb/s ccp: ~350 Mb/s 128kB: ~4600 Mb/s cryptosoft: ~1750 Mb/s 128kB: ~1700 Mb/s This driver is EXPERIMENTAL. You should verify cryptographic results on typical and corner case inputs from your application against a known- good implementation. Sponsored by: Dell EMC Isilon Differential Revision: https://reviews.freebsd.org/D12723
2018-01-18 22:01:30 +00:00
aesmodules = [ 'cryptosoft0', 'aesni0', 'ccr0', 'ccp0' ]
2014-12-12 19:56:36 +00:00
desmodules = [ 'cryptosoft0', ]
Add ccp(4): experimental driver for AMD Crypto Co-Processor * Registers TRNG source for random(4) * Finds available queues, LSBs; allocates static objects * Allocates a shared MSI-X for all queues. The hardware does not have separate interrupts per queue. Working interrupt mode driver. * Computes SHA hashes, HMAC. Passes cryptotest.py, cryptocheck tests. * Does AES-CBC, CTR mode, and XTS. cryptotest.py and cryptocheck pass. * Support for "authenc" (AES + HMAC). (SHA1 seems to result in "unaligned" cleartext inputs from cryptocheck -- which the engine cannot handle. SHA2 seems to work fine.) * GCM passes for block-multiple AAD, input lengths Largely based on ccr(4), part of cxgbe(4). Rough performance averages on AMD Ryzen 1950X (4kB buffer): aesni: SHA1: ~8300 Mb/s SHA256: ~8000 Mb/s ccp: ~630 Mb/s SHA256: ~660 Mb/s SHA512: ~700 Mb/s cryptosoft: ~1800 Mb/s SHA256: ~1800 Mb/s SHA512: ~2700 Mb/s As you can see, performance is poor in comparison to aesni(4) and even cryptosoft (due to high setup cost). At a larger buffer size (128kB), throughput is a little better (but still worse than aesni(4)): aesni: SHA1:~10400 Mb/s SHA256: ~9950 Mb/s ccp: ~2200 Mb/s SHA256: ~2600 Mb/s SHA512: ~3800 Mb/s cryptosoft: ~1750 Mb/s SHA256: ~1800 Mb/s SHA512: ~2700 Mb/s AES performance has a similar story: aesni: 4kB: ~11250 Mb/s 128kB: ~11250 Mb/s ccp: ~350 Mb/s 128kB: ~4600 Mb/s cryptosoft: ~1750 Mb/s 128kB: ~1700 Mb/s This driver is EXPERIMENTAL. You should verify cryptographic results on typical and corner case inputs from your application against a known- good implementation. Sponsored by: Dell EMC Isilon Differential Revision: https://reviews.freebsd.org/D12723
2018-01-18 22:01:30 +00:00
shamodules = [ 'cryptosoft0', 'aesni0', 'ccr0', 'ccp0' ]
2014-12-12 19:56:36 +00:00
def GenTestCase(cname):
try:
crid = cryptodev.Crypto.findcrid(cname)
except IOError:
return None
class GendCryptoTestCase(unittest.TestCase):
###############
##### AES #####
###############
@unittest.skipIf(cname not in aesmodules, 'skipping AES on %s' % (cname))
2014-12-12 19:56:36 +00:00
def test_xts(self):
for i in katg('XTSTestVectors/format tweak value input - data unit seq no', '*.rsp'):
self.runXTS(i, cryptodev.CRYPTO_AES_XTS)
@unittest.skipIf(cname not in aesmodules, 'skipping AES on %s' % (cname))
2014-12-12 19:56:36 +00:00
def test_cbc(self):
for i in katg('KAT_AES', 'CBC[GKV]*.rsp'):
self.runCBC(i)
@unittest.skipIf(cname not in aesmodules, 'skipping AES on %s' % (cname))
2014-12-12 19:56:36 +00:00
def test_gcm(self):
for i in katg('gcmtestvectors', 'gcmEncrypt*'):
self.runGCM(i, 'ENCRYPT')
for i in katg('gcmtestvectors', 'gcmDecrypt*'):
self.runGCM(i, 'DECRYPT')
_gmacsizes = { 32: cryptodev.CRYPTO_AES_256_NIST_GMAC,
24: cryptodev.CRYPTO_AES_192_NIST_GMAC,
16: cryptodev.CRYPTO_AES_128_NIST_GMAC,
}
def runGCM(self, fname, mode):
curfun = None
if mode == 'ENCRYPT':
swapptct = False
curfun = Crypto.encrypt
elif mode == 'DECRYPT':
swapptct = True
curfun = Crypto.decrypt
else:
raise RuntimeError('unknown mode: %r' % repr(mode))
2014-12-12 19:56:36 +00:00
for bogusmode, lines in cryptodev.KATParser(fname,
[ 'Count', 'Key', 'IV', 'CT', 'AAD', 'Tag', 'PT', ]):
for data in lines:
curcnt = int(data['Count'])
cipherkey = data['Key'].decode('hex')
iv = data['IV'].decode('hex')
aad = data['AAD'].decode('hex')
tag = data['Tag'].decode('hex')
if 'FAIL' not in data:
pt = data['PT'].decode('hex')
ct = data['CT'].decode('hex')
if len(iv) != 12:
# XXX - isn't supported
continue
Add ccp(4): experimental driver for AMD Crypto Co-Processor * Registers TRNG source for random(4) * Finds available queues, LSBs; allocates static objects * Allocates a shared MSI-X for all queues. The hardware does not have separate interrupts per queue. Working interrupt mode driver. * Computes SHA hashes, HMAC. Passes cryptotest.py, cryptocheck tests. * Does AES-CBC, CTR mode, and XTS. cryptotest.py and cryptocheck pass. * Support for "authenc" (AES + HMAC). (SHA1 seems to result in "unaligned" cleartext inputs from cryptocheck -- which the engine cannot handle. SHA2 seems to work fine.) * GCM passes for block-multiple AAD, input lengths Largely based on ccr(4), part of cxgbe(4). Rough performance averages on AMD Ryzen 1950X (4kB buffer): aesni: SHA1: ~8300 Mb/s SHA256: ~8000 Mb/s ccp: ~630 Mb/s SHA256: ~660 Mb/s SHA512: ~700 Mb/s cryptosoft: ~1800 Mb/s SHA256: ~1800 Mb/s SHA512: ~2700 Mb/s As you can see, performance is poor in comparison to aesni(4) and even cryptosoft (due to high setup cost). At a larger buffer size (128kB), throughput is a little better (but still worse than aesni(4)): aesni: SHA1:~10400 Mb/s SHA256: ~9950 Mb/s ccp: ~2200 Mb/s SHA256: ~2600 Mb/s SHA512: ~3800 Mb/s cryptosoft: ~1750 Mb/s SHA256: ~1800 Mb/s SHA512: ~2700 Mb/s AES performance has a similar story: aesni: 4kB: ~11250 Mb/s 128kB: ~11250 Mb/s ccp: ~350 Mb/s 128kB: ~4600 Mb/s cryptosoft: ~1750 Mb/s 128kB: ~1700 Mb/s This driver is EXPERIMENTAL. You should verify cryptographic results on typical and corner case inputs from your application against a known- good implementation. Sponsored by: Dell EMC Isilon Differential Revision: https://reviews.freebsd.org/D12723
2018-01-18 22:01:30 +00:00
try:
c = Crypto(cryptodev.CRYPTO_AES_NIST_GCM_16,
cipherkey,
mac=self._gmacsizes[len(cipherkey)],
mackey=cipherkey, crid=crid)
except EnvironmentError, e:
# Can't test algorithms the driver does not support.
if e.errno != errno.EOPNOTSUPP:
raise
continue
2014-12-12 19:56:36 +00:00
if mode == 'ENCRYPT':
Add ccp(4): experimental driver for AMD Crypto Co-Processor * Registers TRNG source for random(4) * Finds available queues, LSBs; allocates static objects * Allocates a shared MSI-X for all queues. The hardware does not have separate interrupts per queue. Working interrupt mode driver. * Computes SHA hashes, HMAC. Passes cryptotest.py, cryptocheck tests. * Does AES-CBC, CTR mode, and XTS. cryptotest.py and cryptocheck pass. * Support for "authenc" (AES + HMAC). (SHA1 seems to result in "unaligned" cleartext inputs from cryptocheck -- which the engine cannot handle. SHA2 seems to work fine.) * GCM passes for block-multiple AAD, input lengths Largely based on ccr(4), part of cxgbe(4). Rough performance averages on AMD Ryzen 1950X (4kB buffer): aesni: SHA1: ~8300 Mb/s SHA256: ~8000 Mb/s ccp: ~630 Mb/s SHA256: ~660 Mb/s SHA512: ~700 Mb/s cryptosoft: ~1800 Mb/s SHA256: ~1800 Mb/s SHA512: ~2700 Mb/s As you can see, performance is poor in comparison to aesni(4) and even cryptosoft (due to high setup cost). At a larger buffer size (128kB), throughput is a little better (but still worse than aesni(4)): aesni: SHA1:~10400 Mb/s SHA256: ~9950 Mb/s ccp: ~2200 Mb/s SHA256: ~2600 Mb/s SHA512: ~3800 Mb/s cryptosoft: ~1750 Mb/s SHA256: ~1800 Mb/s SHA512: ~2700 Mb/s AES performance has a similar story: aesni: 4kB: ~11250 Mb/s 128kB: ~11250 Mb/s ccp: ~350 Mb/s 128kB: ~4600 Mb/s cryptosoft: ~1750 Mb/s 128kB: ~1700 Mb/s This driver is EXPERIMENTAL. You should verify cryptographic results on typical and corner case inputs from your application against a known- good implementation. Sponsored by: Dell EMC Isilon Differential Revision: https://reviews.freebsd.org/D12723
2018-01-18 22:01:30 +00:00
try:
rct, rtag = c.encrypt(pt, iv, aad)
except EnvironmentError, e:
# Can't test inputs the driver does not support.
if e.errno != errno.EINVAL:
raise
continue
2014-12-12 19:56:36 +00:00
rtag = rtag[:len(tag)]
data['rct'] = rct.encode('hex')
data['rtag'] = rtag.encode('hex')
self.assertEqual(rct, ct, repr(data))
self.assertEqual(rtag, tag, repr(data))
2014-12-12 19:56:36 +00:00
else:
if len(tag) != 16:
continue
args = (ct, iv, aad, tag)
if 'FAIL' in data:
self.assertRaises(IOError,
c.decrypt, *args)
else:
Add ccp(4): experimental driver for AMD Crypto Co-Processor * Registers TRNG source for random(4) * Finds available queues, LSBs; allocates static objects * Allocates a shared MSI-X for all queues. The hardware does not have separate interrupts per queue. Working interrupt mode driver. * Computes SHA hashes, HMAC. Passes cryptotest.py, cryptocheck tests. * Does AES-CBC, CTR mode, and XTS. cryptotest.py and cryptocheck pass. * Support for "authenc" (AES + HMAC). (SHA1 seems to result in "unaligned" cleartext inputs from cryptocheck -- which the engine cannot handle. SHA2 seems to work fine.) * GCM passes for block-multiple AAD, input lengths Largely based on ccr(4), part of cxgbe(4). Rough performance averages on AMD Ryzen 1950X (4kB buffer): aesni: SHA1: ~8300 Mb/s SHA256: ~8000 Mb/s ccp: ~630 Mb/s SHA256: ~660 Mb/s SHA512: ~700 Mb/s cryptosoft: ~1800 Mb/s SHA256: ~1800 Mb/s SHA512: ~2700 Mb/s As you can see, performance is poor in comparison to aesni(4) and even cryptosoft (due to high setup cost). At a larger buffer size (128kB), throughput is a little better (but still worse than aesni(4)): aesni: SHA1:~10400 Mb/s SHA256: ~9950 Mb/s ccp: ~2200 Mb/s SHA256: ~2600 Mb/s SHA512: ~3800 Mb/s cryptosoft: ~1750 Mb/s SHA256: ~1800 Mb/s SHA512: ~2700 Mb/s AES performance has a similar story: aesni: 4kB: ~11250 Mb/s 128kB: ~11250 Mb/s ccp: ~350 Mb/s 128kB: ~4600 Mb/s cryptosoft: ~1750 Mb/s 128kB: ~1700 Mb/s This driver is EXPERIMENTAL. You should verify cryptographic results on typical and corner case inputs from your application against a known- good implementation. Sponsored by: Dell EMC Isilon Differential Revision: https://reviews.freebsd.org/D12723
2018-01-18 22:01:30 +00:00
try:
rpt, rtag = c.decrypt(*args)
except EnvironmentError, e:
# Can't test inputs the driver does not support.
if e.errno != errno.EINVAL:
raise
continue
2014-12-12 19:56:36 +00:00
data['rpt'] = rpt.encode('hex')
data['rtag'] = rtag.encode('hex')
self.assertEqual(rpt, pt,
repr(data))
2014-12-12 19:56:36 +00:00
def runCBC(self, fname):
curfun = None
for mode, lines in cryptodev.KATParser(fname,
[ 'COUNT', 'KEY', 'IV', 'PLAINTEXT', 'CIPHERTEXT', ]):
if mode == 'ENCRYPT':
swapptct = False
curfun = Crypto.encrypt
elif mode == 'DECRYPT':
swapptct = True
curfun = Crypto.decrypt
else:
raise RuntimeError('unknown mode: %r' % repr(mode))
2014-12-12 19:56:36 +00:00
for data in lines:
curcnt = int(data['COUNT'])
cipherkey = data['KEY'].decode('hex')
iv = data['IV'].decode('hex')
pt = data['PLAINTEXT'].decode('hex')
ct = data['CIPHERTEXT'].decode('hex')
if swapptct:
pt, ct = ct, pt
# run the fun
c = Crypto(cryptodev.CRYPTO_AES_CBC, cipherkey, crid=crid)
r = curfun(c, pt, iv)
self.assertEqual(r, ct)
def runXTS(self, fname, meth):
curfun = None
for mode, lines in cryptodev.KATParser(fname,
[ 'COUNT', 'DataUnitLen', 'Key', 'DataUnitSeqNumber', 'PT',
'CT' ]):
if mode == 'ENCRYPT':
swapptct = False
curfun = Crypto.encrypt
elif mode == 'DECRYPT':
swapptct = True
curfun = Crypto.decrypt
else:
raise RuntimeError('unknown mode: %r' % repr(mode))
2014-12-12 19:56:36 +00:00
for data in lines:
curcnt = int(data['COUNT'])
nbits = int(data['DataUnitLen'])
cipherkey = data['Key'].decode('hex')
iv = struct.pack('QQ', int(data['DataUnitSeqNumber']), 0)
pt = data['PT'].decode('hex')
ct = data['CT'].decode('hex')
if nbits % 128 != 0:
# XXX - mark as skipped
continue
if swapptct:
pt, ct = ct, pt
# run the fun
Add ccp(4): experimental driver for AMD Crypto Co-Processor * Registers TRNG source for random(4) * Finds available queues, LSBs; allocates static objects * Allocates a shared MSI-X for all queues. The hardware does not have separate interrupts per queue. Working interrupt mode driver. * Computes SHA hashes, HMAC. Passes cryptotest.py, cryptocheck tests. * Does AES-CBC, CTR mode, and XTS. cryptotest.py and cryptocheck pass. * Support for "authenc" (AES + HMAC). (SHA1 seems to result in "unaligned" cleartext inputs from cryptocheck -- which the engine cannot handle. SHA2 seems to work fine.) * GCM passes for block-multiple AAD, input lengths Largely based on ccr(4), part of cxgbe(4). Rough performance averages on AMD Ryzen 1950X (4kB buffer): aesni: SHA1: ~8300 Mb/s SHA256: ~8000 Mb/s ccp: ~630 Mb/s SHA256: ~660 Mb/s SHA512: ~700 Mb/s cryptosoft: ~1800 Mb/s SHA256: ~1800 Mb/s SHA512: ~2700 Mb/s As you can see, performance is poor in comparison to aesni(4) and even cryptosoft (due to high setup cost). At a larger buffer size (128kB), throughput is a little better (but still worse than aesni(4)): aesni: SHA1:~10400 Mb/s SHA256: ~9950 Mb/s ccp: ~2200 Mb/s SHA256: ~2600 Mb/s SHA512: ~3800 Mb/s cryptosoft: ~1750 Mb/s SHA256: ~1800 Mb/s SHA512: ~2700 Mb/s AES performance has a similar story: aesni: 4kB: ~11250 Mb/s 128kB: ~11250 Mb/s ccp: ~350 Mb/s 128kB: ~4600 Mb/s cryptosoft: ~1750 Mb/s 128kB: ~1700 Mb/s This driver is EXPERIMENTAL. You should verify cryptographic results on typical and corner case inputs from your application against a known- good implementation. Sponsored by: Dell EMC Isilon Differential Revision: https://reviews.freebsd.org/D12723
2018-01-18 22:01:30 +00:00
try:
c = Crypto(meth, cipherkey, crid=crid)
r = curfun(c, pt, iv)
except EnvironmentError, e:
# Can't test hashes the driver does not support.
if e.errno != errno.EOPNOTSUPP:
raise
continue
2014-12-12 19:56:36 +00:00
self.assertEqual(r, ct)
###############
##### DES #####
###############
@unittest.skipIf(cname not in desmodules, 'skipping DES on %s' % (cname))
2014-12-12 19:56:36 +00:00
def test_tdes(self):
for i in katg('KAT_TDES', 'TCBC[a-z]*.rsp'):
self.runTDES(i)
def runTDES(self, fname):
curfun = None
for mode, lines in cryptodev.KATParser(fname,
[ 'COUNT', 'KEYs', 'IV', 'PLAINTEXT', 'CIPHERTEXT', ]):
if mode == 'ENCRYPT':
swapptct = False
curfun = Crypto.encrypt
elif mode == 'DECRYPT':
swapptct = True
curfun = Crypto.decrypt
else:
raise RuntimeError('unknown mode: %r' % repr(mode))
2014-12-12 19:56:36 +00:00
for data in lines:
curcnt = int(data['COUNT'])
key = data['KEYs'] * 3
cipherkey = key.decode('hex')
iv = data['IV'].decode('hex')
pt = data['PLAINTEXT'].decode('hex')
ct = data['CIPHERTEXT'].decode('hex')
if swapptct:
pt, ct = ct, pt
# run the fun
c = Crypto(cryptodev.CRYPTO_3DES_CBC, cipherkey, crid=crid)
r = curfun(c, pt, iv)
self.assertEqual(r, ct)
###############
##### SHA #####
###############
@unittest.skipIf(cname not in shamodules, 'skipping SHA on %s' % str(cname))
2014-12-12 19:56:36 +00:00
def test_sha(self):
# SHA not available in software
pass
#for i in iglob('SHA1*'):
# self.runSHA(i)
@unittest.skipIf(cname not in shamodules, 'skipping SHA on %s' % str(cname))
2014-12-12 19:56:36 +00:00
def test_sha1hmac(self):
for i in katg('hmactestvectors', 'HMAC.rsp'):
self.runSHA1HMAC(i)
def runSHA1HMAC(self, fname):
for hashlength, lines in cryptodev.KATParser(fname,
2014-12-12 19:56:36 +00:00
[ 'Count', 'Klen', 'Tlen', 'Key', 'Msg', 'Mac' ]):
# E.g., hashlength will be "L=20" (bytes)
hashlen = int(hashlength.split("=")[1])
blocksize = None
if hashlen == 20:
alg = cryptodev.CRYPTO_SHA1_HMAC
blocksize = 64
elif hashlen == 28:
# Cryptodev doesn't support SHA-224
# Slurp remaining input in section
for data in lines:
continue
continue
elif hashlen == 32:
alg = cryptodev.CRYPTO_SHA2_256_HMAC
blocksize = 64
elif hashlen == 48:
alg = cryptodev.CRYPTO_SHA2_384_HMAC
blocksize = 128
elif hashlen == 64:
alg = cryptodev.CRYPTO_SHA2_512_HMAC
blocksize = 128
else:
# Skip unsupported hashes
# Slurp remaining input in section
for data in lines:
continue
continue
2014-12-12 19:56:36 +00:00
for data in lines:
key = data['Key'].decode('hex')
msg = data['Msg'].decode('hex')
mac = data['Mac'].decode('hex')
tlen = int(data['Tlen'])
2014-12-12 19:56:36 +00:00
if len(key) > blocksize:
2014-12-12 19:56:36 +00:00
continue
try:
c = Crypto(mac=alg, mackey=key,
crid=crid)
except EnvironmentError, e:
# Can't test hashes the driver does not support.
if e.errno != errno.EOPNOTSUPP:
raise
continue
_, r = c.encrypt(msg, iv="")
2014-12-12 19:56:36 +00:00
# A limitation in cryptodev.py means we
# can only store MACs up to 16 bytes.
# That's good enough to validate the
# correct behavior, more or less.
maclen = min(tlen, 16)
self.assertEqual(r[:maclen], mac[:maclen], "Actual: " + \
repr(r[:maclen].encode("hex")) + " Expected: " + repr(data))
2014-12-12 19:56:36 +00:00
return GendCryptoTestCase
cryptosoft = GenTestCase('cryptosoft0')
aesni = GenTestCase('aesni0')
ccr = GenTestCase('ccr0')
Add ccp(4): experimental driver for AMD Crypto Co-Processor * Registers TRNG source for random(4) * Finds available queues, LSBs; allocates static objects * Allocates a shared MSI-X for all queues. The hardware does not have separate interrupts per queue. Working interrupt mode driver. * Computes SHA hashes, HMAC. Passes cryptotest.py, cryptocheck tests. * Does AES-CBC, CTR mode, and XTS. cryptotest.py and cryptocheck pass. * Support for "authenc" (AES + HMAC). (SHA1 seems to result in "unaligned" cleartext inputs from cryptocheck -- which the engine cannot handle. SHA2 seems to work fine.) * GCM passes for block-multiple AAD, input lengths Largely based on ccr(4), part of cxgbe(4). Rough performance averages on AMD Ryzen 1950X (4kB buffer): aesni: SHA1: ~8300 Mb/s SHA256: ~8000 Mb/s ccp: ~630 Mb/s SHA256: ~660 Mb/s SHA512: ~700 Mb/s cryptosoft: ~1800 Mb/s SHA256: ~1800 Mb/s SHA512: ~2700 Mb/s As you can see, performance is poor in comparison to aesni(4) and even cryptosoft (due to high setup cost). At a larger buffer size (128kB), throughput is a little better (but still worse than aesni(4)): aesni: SHA1:~10400 Mb/s SHA256: ~9950 Mb/s ccp: ~2200 Mb/s SHA256: ~2600 Mb/s SHA512: ~3800 Mb/s cryptosoft: ~1750 Mb/s SHA256: ~1800 Mb/s SHA512: ~2700 Mb/s AES performance has a similar story: aesni: 4kB: ~11250 Mb/s 128kB: ~11250 Mb/s ccp: ~350 Mb/s 128kB: ~4600 Mb/s cryptosoft: ~1750 Mb/s 128kB: ~1700 Mb/s This driver is EXPERIMENTAL. You should verify cryptographic results on typical and corner case inputs from your application against a known- good implementation. Sponsored by: Dell EMC Isilon Differential Revision: https://reviews.freebsd.org/D12723
2018-01-18 22:01:30 +00:00
ccp = GenTestCase('ccp0')
2014-12-12 19:56:36 +00:00
if __name__ == '__main__':
unittest.main()