| Internet-Draft | Encryption algorithm Rocca-S | March 2023 |
| Nakano, et al. | Expires 3 September 2023 | [Page] |
This document defines Rocca-S encryption scheme, which is an Authenticated Encryption with Associated Data (AEAD), using a 256-bit key and can be efficiently implemented utilizing the AES New Instruction set (AES-NI).¶
This Internet-Draft is submitted in full conformance with the provisions of BCP 78 and BCP 79.¶
Internet-Drafts are working documents of the Internet Engineering Task Force (IETF). Note that other groups may also distribute working documents as Internet-Drafts. The list of current Internet-Drafts is at https://datatracker.ietf.org/drafts/current/.¶
Internet-Drafts are draft documents valid for a maximum of six months and may be updated, replaced, or obsoleted by other documents at any time. It is inappropriate to use Internet-Drafts as reference material or to cite them other than as "work in progress."¶
This Internet-Draft will expire on 3 September 2023.¶
Copyright (c) 2023 IETF Trust and the persons identified as the document authors. All rights reserved.¶
This document is subject to BCP 78 and the IETF Trust's Legal Provisions Relating to IETF Documents (https://trustee.ietf.org/license-info) in effect on the date of publication of this document. Please review these documents carefully, as they describe your rights and restrictions with respect to this document. Code Components extracted from this document must include Revised BSD License text as described in Section 4.e of the Trust Legal Provisions and are provided without warranty as described in the Revised BSD License.¶
Countries such as the USA, China, and South Korea are adapting to the fifth-generation mobile communication systems (5G) technology at an increasingly rapid pace. There are more than 1500 cities worldwide with access to 5G technology. Other countries are also taking significant steps to make 5G networks commercially available to their citizens. As the research in 5G technology is moving toward global standardization, it is important for the research community to focus on developing solutions beyond 5G and for the 6G era. The first white paper on 6G [WP-6G] was published by 6G Flagship, University of Oulu, Finland under the 6Genesis project in 2019. This white paper identified the key drivers, research requirements, challenges and essential research questions related to 6G. One of the main requirements as listed in this paper was to look at the problem of transmitting data at a speed of over 100 Gbps per user.¶
Additionally, 3GPP requires that the cryptographic algorithms proposed for 5G systems should support 256-bit keys [SPEC-5G]. Apart from the need of speeds of more than 100 Gbps and supporting 256-bit keys, 3GPP also discusses the possible impacts of quantum computing in the coming years, especially due to Grover's algorithm. While describing the impact of quantum computers on symmetric algorithms required for 5G and beyond, 3GPP states the following in Section 5.3 of [SPEC-5G]:¶
"The threat to symmetric cryptography from quantum computing is lower than that for asymmetric cryptography. As such there is little benefit in transitioning symmetric algorithms without corresponding changes to the asymmetric algorithms that accompany them."¶
However, it has been shown in numerous articles that quantum computers can be used to either efficiently break or drastically reduce the time necessary to attack some symmetric-key cryptography methods. These results require a serious reevaluation of the premise that has informed beyond 5G quantum security concerns up to this point. Additionally, since NIST will finally standardize quantum-resistant public key algorithms in the coming few years, we believe it is important for the research community to also focus on symmetric algorithms for future telecommunications that would provide security against quantum adversaries. The effectiveness of post-quantum asymmetric cryptography would only be improved if the symmetric cryptography used with it is also quantum resistant. Thus, a symmetric cryptographic algorithm that¶
is needed.¶
Rocca has been designed as an encryption algorithm for a high speed communication such as future internet and beyond 5G mobile communications. Rocca achieves an encryption/decryption speed of more than 100 Gbps in both the raw encryption scheme and the AEAD scheme. It supports a 256-bit key and provides 256-bit and 128-bit security against the key recovery and distinguishing attacks, respectively. The high throughput of Rocca can be achieved by utilizing the AES New Instruction set (AES-NI) [AES-NI]. Similar approach has been taken by AEGIS family [AEGIS] and Tiaoxin-346 [TIAOXIN], both are two submissions to the CAESAR competition [CAESAR]. SNOW-V [SNOW-V] also uses AES round function as a component so that AES-NI can be used.¶
As Rocca has been designed for future telecommunication services, Rocca satisfies two out of three above mentioned requirements. However, there is still room for the improvement with regard to security against quantum computers. This motivates us to propose a symmetric-key algorithm that satisfies all three of the above-mentioned requirements. In this document, we propose Rocca-S, which is an AES-based encryption scheme with a 256-bit key. Rocca-S provides both a raw encryption scheme and an AEAD scheme with a 256-bit tag. Rocca-S is designed to meet the requirements of high throughput of more than 100 Gbps as well as 256-bit security. Rocca-S achieves an encryption/decryption speed of more than 200 Gbps in both raw encryption scheme and AEAD scheme on Intel(R) Core(TM) i9-12900K, and can provide 256-bit and 128-bit security against classical and quantum adversaries respectively.¶
In this document, we present an AES-based AEAD encryption scheme with a 256-bit key and 256-bit tag called Rocca-S, which is a variant of Rocca described in [ROCCA]. The goal of Rocca-S is to further improve the security of Rocca while maintaining its performance advantage.¶
To achieve such a dramatically fast encryption/decryption speed, Rocca-S follows the same design principle as Rocca, such as the SIMD-friendly round function and an efficient permutation-based structure. We explore the class of AES-based structures to further increase its speed and reduce the state size. Specifically, we take the following different approaches.¶
aesenc) or XOR
while Jean and Nikolic consider the case of applying both aesenc and XOR
in a cascade way for one round,
and the most efficient structures in [DESIGN] are included in this class.¶
The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "NOT RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in BCP 14 [RFC2119] [RFC8174] when, and only when, they appear in all capitals, as shown here.¶
In this section, the notations and the specification of our designs will be described.¶
The following notations will be used in the document. Throughout this document, a block means a 2-octet value. For the constants Z0 and Z1, we utilize the same ones as Tiaoxin-346 [TIAOXIN].¶
X ^ Y: The bitwise Exclusive OR (XOR) of X and Y.¶
X#Y: For a number X and a positive integer Y, the Y-th power of X.¶
f#(N): For a function f and a non-negative integer N,
the N-th iteration of function f.¶
|X|: The length of X in bits.¶
X||Y : The concatenation of X and Y.¶
ZERO(l): A zero string of length l bits.¶
PAD(X): X||ZERO(l), where l is the minimal non-negative integer such that
|PAD(X)| is a multiple of 256.¶
PADN(X): X||ZERO(l), where l is the minimal non-negative integer such that
|PADN(X)| is a multiple of 128.¶
LE128(X): the little-endian encoding of 128-bit integer X.¶
X as X = X[0]||X[1]|| ... ||X[n] with |X[i]| = 256,
where n is |X|/256 - 1.
In addition, X[i] is written as X[i] = X[i]_0||X[i]_1 with |X[i]_0| = |X[i]_1| = 128.¶
S: The state of Rocca-S, which is composed of 7 blocks, i.e.,
S = (S[0], S[1], ..., S[6]),
where S[i] (0 <= i <= 6) are blocks and S[0] is the first block.¶
Z0: A 128-bit constant block defined as Z0 = 428a2f98d728ae227137449123ef65cd.¶
Z1: A 128-bit constant block defined as Z1 = b5c0fbcfec4d3b2fe9b5dba58189dbbc.¶
A(X): The AES round function without the constant addition operation, as defined
below:A(X) = MixColumns( ShiftRows( SubBytes(X) ) ),
where MixColumns, ShiftRows and SubBytes are the same operations as defined
in AES [AES].¶
AES(X,Y): One AES round is applied to the block X, where the round constant is Y,
as defined below:AES(X,Y) = A(X) ^ Y.aesenc, which is one of the instructions of AES-NI,
performs one regular (not the last) round of AES on an input state X with a subkey Y.¶
R(S,X0,X1): The round function is used to update the state S, as defined in Section 2.2.¶
The input of the round function R(S,X0,X1) of Rocca-S consists of the state S and two
blocks (X0,X1).
If denoting the output by Snew, Snew:=R(S,X0,X1) can be defined as follows:¶
Snew[0] = S[6] ^ S[1], Snew[1] = AES(S[0],X_0), Snew[2] = AES(S[1],S[0]), Snew[3] = AES(S[2],S[6]), Snew[4] = AES(S[3],X_1), Snew[5] = AES(S[4],S[3]), Snew[6] = AES(S[5],S[4]).¶
The corresponding illustration can be found in Figure 1.¶
+----+ +----+ +----+ +----+ +----+ +----+ +----+ |S[0]| |S[1]| |S[6]| |S[2]| |S[3]| |S[4]| |S[5]| +-+-++ ++-+-+ ++-+-+ ++---+ +-+-++ ++-+-+ ++---+ | | | | | | | | | | | | | +-----+ | +--+ | +--+ | | +-----+ | +--+ | v | v | v | v v | v | v +---+ | +---+ | +---+ | +---+ +---+ | +---+ | +---+ |AES|<-X0 +>|AES| +>|XOR| +>|AES| |AES|<-X1 +>|AES| +>|AES| +-+-+ +-+-+ +-+-+ +-+-+ +-+-+ +-+-+ +-+-+ | | | | | | | v v v v v v v +----+ +----+ +----+ +----+ +----+ +----+ +----+ |Snew| |Snew| |Snew| |Snew| |Snew| |Snew| |Snew| | [1]| | [2]| | [0]| | [3]| | [4]| | [5]| | [6]| +----+ +----+ +----+ +----+ +----+ +----+ +----+
Rocca-S is an AEAD scheme composed of four phases:
initialization, processing the associated data, encryption and finalization.
The input consists of a 256-bit key K = K0||K1,
a nonce N of between 12 and 16 octets (both inclusive) in length,
the associated data AD and the message M,
where K0 and K1 are elements of the binary finite field of 2#128.
The output is the corresponding ciphertext C and a 256-bit tag T.¶
The settings described below are required for the parameters:¶
K MUST be unpredictable for each invocation.¶
PADN(N), where N is the nonce, MUST be unique per invocation with the same key,
so N MUST NOT be randomly generated.¶
First, (N,K0,K1) is loaded into the state S in the following way:¶
S[0] = K1, S[1] = PADN(N), S[2] = Z0, S[3] = K0, S[4] = Z1, S[5] = PADN(N) ^ K1, S[6] = ZERO(128)¶
Then, 16 iterations of the round function R(S,Z0,Z1),
which is written as R(S,Z0,Z1)#(16),
are applied to state S.¶
After 16 iterations of the round function, two 128-bit keys are XORed with the state S in the following way:¶
S[0] = S[0] ^ K0, S[1] = S[1] ^ K0, S[2] = S[2] ^ K1, S[3] = S[3] ^ K0, S[4] = S[4] ^ K0, S[5] = S[5] ^ K1, S[6] = S[6] ^ K1.¶
If AD is empty, this phase will be skipped. Otherwise,
AD is padded to PAD(AD), and the state is updated as follows:¶
for i = 0 to d - 1
R(S, PAD(AD)[i]_0, PAD(AD)[i]_1),
end for
¶
where d = |PAD(AD)| / 256.¶
The encryption phase is similar to the phase to process the associated data.
If M is empty, the encryption phase will be skipped.
Otherwise, M is first padded to PAD(M), and then PAD(M) will be absorbed
with the round function.
During this procedure, the ciphertext C is generated.
If the last block of M is incomplete and its length is b bits, i.e.,
0 < b < 256, the last block of C will be truncated to the first b bits.
A detailed description is shown below:¶
for i = 0 to m - 1
C[i]_0 = AES(S[3] ^ S[5], S[0]) ^ PAD(M)[i]_0,
C[i]_1 = AES(S[4] ^ S[6], S[2]) ^ PAD(M)[i]_1,
R(S, PAD(M)[i]_0, PAD(M)[i]_1),
end for
¶
where m = |PAD(M)| / 256.¶
The state S will again pass through 16 iterations
of the round function R(S,LE128(|AD|),LE128(|M|)) and then the 256-bit tag T is computed in the
following way:¶
T = (S[0] ^ S[1] ^ S[2] ^ S[3]) || (S[4] ^ S[5] ^ S[6]).¶
A formal description of Rocca-S can be seen in Figure 2, and the corresponding illustration is shown in Figure 3.¶
// Rocca-S Algorithm. The specification of Rocca-S
procedure RoccaEncrypt(K0, K1, N, AD, M)
S = Initialization(N,K0,K1)
if |AD| > 0 then
S = ProcessAD(S,PAD(AD))
if |M| > 0 then
S = Encryption(S,PAD(M),C)
Truncate C
T = Finalization(S, |AD|, |M|)
return (C, T)
procedure RoccaDecrypt(K0, K1, N, AD, C, T)
S = Initialization(N,K0,K1)
if |AD| > 0 then
S = ProcessAD(S,PAD(AD))
if |C| > 0 then
S = Decryption(S,PAD(C),M)
Truncate M
if T == Finalization(S, |AD|, |C|) then
return M
else
return nil
procedure Initialization(N, K0, K1)
S[0] = K1,
S[1] = PADN(N),
S[2] = Z0,
S[3] = K0,
S[4] = Z1,
S[5] = PADN(N) ^ K1,
S[6] = ZERO(128)
for i = 0 to 15 do
S = R(S, Z0, Z1)
S[0] = S[0] ^ K0,
S[1] = S[1] ^ K0,
S[2] = S[2] ^ K1,
S[3] = S[3] ^ K0,
S[4] = S[4] ^ K0,
S[5] = S[5] ^ K1,
S[6] = S[6] ^ K1
return S
procedure ProcessAD(S, AD)
d = |PAD(AD)|/256
for i = 0 to d - 1 do
S = R(S, AD[i]_0, AD[i]_1)
return S
procedure Encryption(S, M, C)
m = |PAD(M)|/256
for i = 0 to m - 1 do
C[i]_0 = AES(S[3] ^ S[5], S[0]) ^ M[i]_0
C[i]_1 = AES(S[4] ^ S[6], S[2]) ^ M[i]_1
S = R(S,M[i]_0, M[i]_1)
return S
procedure Decryption(S, M, C)
c = |C|/256
for i = 0 to c - 1 do
M[i]_0 = AES(S[3] ^ S[5], S[0]) ^ C[i]_0
M[i]_1 = AES(S[4] ^ S[6], S[2]) ^ C[i]_1
S = R(S,M[i]_0, M[i]_1)
return S
procedure Finalization(S, |AD|, |M|)
for i = 0 to 15 do
S = R(S, |AD|, |M|)
T0 = 0
T1 = 0
for i = 0 to 3 do
T0 = T0 ^ S[i]
for i = 4 to 6 do
T1 = T1 ^ S[i]
return T0||T1
Z1 AD[0]_1 AD[1]_1
| | |
v v v
+------+ +---+ +---+
PADN(N)->| | | | | |
|R#(16)+------>| R +-->| R +->...---+
K0||K1->| | ^ | | | | |
+------+ | +---+ +---+ |
^ K0||K1 ^ ^ |
| | | |
Z0 AD[0]_0 AD[1]_0 |
|
+--------------------------------------------+
|
| C[0]_1 C[1]_1 C[m-1]_1
| ^ ^ ^
| | | | |AD|
| +-+-+ +-+-+ +-+-+ |
| AD[d-1]_1 |XOR|<-M[0]_1 |XOR|<-M[1]_1 |XOR|<-M[m-1]_1 |
| | +---+ | +---+ | +---+ | |
| v ^ v ^ v ^ v |
| +---+ | +---+ | +---+ | +---+ v
| | +---+ | +-----+ | | -+ | | +------+
| | | | | | | | | | |
+------>| R +------>| R +-------->| R +->...---->| R +------>|R#(16)+->T
| | | | | | | | | |
| +---+ | +-----+ | | -+ | | +------+
+---+ | +---+ | +---+ | +---+ ^
^ v ^ v ^ v ^ |
| +---+ | +---+ | +---+ | |
|XOR|<-M[0]_0 |XOR|<-M[1]_0 |XOR|<-M[m-1]_0 |M|
AD[d-1]_0 +-+-+ +-+-+ +-+-+
| | |
v v v
C[0]_0 C[1]_0 C[m-1]_0
If the phases of processing the associated data and finalization are removed, a raw encryption scheme is obtained.¶
If the phases of processing the associated data and finalization are removed, and there is no message injection into the round function such that R(S,0,0),
a keystream generation scheme is obtained.
This scheme can be used as a general stream cipher and random bit generation.¶
For Rocca-S to support 128-bit or 192-bit keys, the given key needs to be expanded to 256 bits.
When 128-bit key is given, it will be set to K0, and K1 is defined as K1 = ZERO(128).
When 192-bit key is given, the first 128-bit will be set to K0, and the remaining 64-bit will be set to K1_p.
Then K1 is defined as K1 = K1_p||ZERO(64).¶
Use of Key Derivation Functions (KDF) [KDF] to stretch the key length to 256-bit could be another option. The given 128-bit or 192-bit key will be used as a key derivation key, and the output of the KDF will be 256-bit.¶
To comply with the requirements defined in Section 4 of [RFC5116], the settings of the parameters for Rocca-S are defined as follows:¶
K_LEN (key length) is 32 octets (256 bits), and K (key) does not require any particular data format.¶
P_MAX (maximum size of the plaintext) is 2#125 octets.¶
A_MAX (maximum size of the associated data) is 2#61 octets.¶
N_MIN (minimum size of the nonce) = 12 octets,
and N_MAX (maximum size of the nonce) = 16 octets.¶
C_MAX (the largest possible AEAD ciphertext) = P_MAX + tag length = 2#125 + 32 octets.¶
In addition,¶
As described in Section 3, Rocca-S provides 256-bit security against key-recovery and forgery attacks in the nonce-respecting setting. Rocca-S also provides 128-bit key-committing security. We do not claim its security in the related-key and known-key settings.¶
The message length for a fixed key is limited to at most 2#128, and we also limit the number of different messages that are produced for a fixed key to be at most 2#128. The length of the associated data for a fixed key is up to 2#64.¶
There exist no quantum attacks for key-recovery and forgery (in nonce-respecting setting) on Rocca-S with time complexity lower than 2#128. Rocca-S does not provide security against related-key and known-key superposition attacks (as is the case of all known block ciphers).¶
Rocca-S is secure against the following attacks:¶
State-recovery Attack:¶
While there are many attack vectors for block ciphers, their application to Rocca-S is restrictive, as the attackers can only know partial information about the internal state from the ciphertext blocks. In other words, reversing the round function is impossible in Rocca-S without guessing many secret state blocks. Therefore, only the above potential attack vectors are taken into account. In addition, due to the usage of the constant (Z0,Z1) at the initialization phase, the attack based on the similarity in the four columns of the AES state is also excluded.¶
Inadvertent reuse of the same nonce by two invocations of the Rocca-S encryption operation, with the same key, undermines the security of the messages processed with those invocations. A loss of confidentiality ensues because an adversary will be able to reconstruct the bitwise exclusive-or of the two plaintext values.¶
IANA has assigned value TBD in the AEAD Algorithms registry to AEAD_ROCCA.¶
Figure 4 shows a sample implementation of Rocca-S.¶
#include <memory.h>
#include <immintrin.h>
#include <stdlib.h>
#include <stdint.h>
#define ROCCA_KEY_SIZE (32)
#define ROCCA_IV_SIZE (16)
#define ROCCA_MSG_BLOCK_SIZE (32)
#define ROCCA_TAG_SIZE (32)
#define ROCCA_STATE_NUM ( 7)
typedef struct ROCCA_CTX {
uint8_t key[ROCCA_KEY_SIZE/16][16];
uint8_t state[ROCCA_STATE_NUM][16];
size_t size_ad;
size_t size_m;
} rocca_context;
#define load(m) _mm_loadu_si128((const __m128i *)(m))
#define store(m,a) _mm_storeu_si128((__m128i *)(m),a)
#define xor(a,b) _mm_xor_si128(a,b)
#define and(a,b) _mm_and_si128(a,b)
#define enc(a,k) _mm_aesenc_si128(a,k)
#define setzero() _mm_setzero_si128()
#define ENCODE_IN_LITTLE_ENDIAN(bytes, v) \
bytes[ 0] = ((uint64_t)(v) << ( 3)); \
bytes[ 1] = ((uint64_t)(v) >> (1*8-3)); \
bytes[ 2] = ((uint64_t)(v) >> (2*8-3)); \
bytes[ 3] = ((uint64_t)(v) >> (3*8-3)); \
bytes[ 4] = ((uint64_t)(v) >> (4*8-3)); \
bytes[ 5] = ((uint64_t)(v) >> (5*8-3)); \
bytes[ 6] = ((uint64_t)(v) >> (6*8-3)); \
bytes[ 7] = ((uint64_t)(v) >> (7*8-3)); \
bytes[ 8] = ((uint64_t)(v) >> (8*8-3)); \
bytes[ 9] = 0; \
bytes[10] = 0; \
bytes[11] = 0; \
bytes[12] = 0; \
bytes[13] = 0; \
bytes[14] = 0; \
bytes[15] = 0;
#define FLOORTO(a,b) ((a) / (b) * (b))
#define S_NUM ROCCA_STATE_NUM
#define M_NUM ( 2)
#define INIT_LOOP (16)
#define TAG_LOOP (16)
#define VARS4UPDATE \
__m128i k[2], state[S_NUM], stateNew[S_NUM], M[M_NUM];
#define VARS4ENCRYPT \
VARS4UPDATE \
__m128i Z[M_NUM], C[M_NUM];
#define COPY_TO_LOCAL(ctx) \
for(size_t i = 0; i < S_NUM; ++i) \
{ state[i] = load(&((ctx)->state[i][0])); }
#define COPY_FROM_LOCAL(ctx) \
for(size_t i = 0; i < S_NUM; ++i) \
{ store(&((ctx)->state[i][0]), state[i]); }
#define COPY_TO_LOCAL_IN_TAG(ctx) \
COPY_TO_LOCAL(ctx) for(size_t i = 0; i < 2; ++i) \
{ k[i] = load(&((ctx)->key[i][0])); }
#define COPY_FROM_LOCAL_IN_INIT(ctx) \
COPY_FROM_LOCAL(ctx) for(size_t i = 0; i < 2; ++i) \
{ store(&((ctx)->key[i][0]), k[i]); }
#define UPDATE_STATE(X) \
stateNew[0] = xor(state[6], state[1]); \
stateNew[1] = enc(state[0], X[0]); \
stateNew[2] = enc(state[1], state[0]); \
stateNew[3] = enc(state[2], state[6]); \
stateNew[4] = enc(state[3], X[1]); \
stateNew[5] = enc(state[4], state[3]); \
stateNew[6] = enc(state[5], state[4]); \
for(size_t i = 0; i < S_NUM; ++i) \
{state[i] = stateNew[i];}
#define INIT_STATE(key, iv) \
k[0] = load((key) + 16*0); \
k[1] = load((key) + 16*1); \
state[0] = k[1]; \
state[1] = load(iv); \
state[2] = load(Z0); \
state[3] = k[0]; \
state[4] = load(Z1); \
state[5] = xor(state[1], state[0]); \
state[6] = setzero(); \
M[0] = state[2]; \
M[1] = state[4]; \
for(size_t i = 0; i < INIT_LOOP; ++i) { \
UPDATE_STATE(M) \
} \
state[0] = xor(state[0], k[0]); \
state[1] = xor(state[1], k[0]); \
state[2] = xor(state[2], k[1]); \
state[3] = xor(state[3], k[0]); \
state[4] = xor(state[4], k[0]); \
state[5] = xor(state[5], k[1]); \
state[6] = xor(state[6], k[1]);
#define MAKE_STRM \
Z[0] = enc(xor(state[3], state[5]), state[0]); \
Z[1] = enc(xor(state[4], state[6]), state[2]);
#define MSG_LOAD(mem, reg) \
reg[0] = load((mem) + 0); \
reg[1] = load((mem) + 16);
#define MSG_STORE(mem, reg) \
store((mem) + 0, reg[0]); \
store((mem) + 16, reg[1]);
#define XOR_BLOCK(dst, src1, src2) \
dst[0] = xor(src1[0], src2[0]); \
dst[1] = xor(src1[1], src2[1]);
#define MASKXOR_BLOCK(dst, src1, src2, mask) \
dst[0] = and(xor(src1[0], src2[0]), mask[0]); \
dst[1] = and(xor(src1[1], src2[1]), mask[1]);
#define ADD_AD(input) \
MSG_LOAD(input, M) \
UPDATE_STATE(M)
#define ADD_AD_LAST_BLOCK(input, size) \
uint8_t tmpblk[ROCCA_MSG_BLOCK_SIZE] = {0}; \
memcpy(tmpblk, input, size); \
MSG_LOAD(tmpblk, M) \
UPDATE_STATE(M)
#define ENCRYPT(output, input) \
MSG_LOAD(input, M) \
MAKE_STRM \
XOR_BLOCK(C, M, Z) \
MSG_STORE(output, C) \
UPDATE_STATE(M)
#define ENCRYPT_LAST_BLOCK(output, input, size) \
uint8_t tmpblk[ROCCA_MSG_BLOCK_SIZE] = {0}; \
memcpy(tmpblk, input, size); \
MSG_LOAD(tmpblk, M) \
MAKE_STRM \
XOR_BLOCK(C, M, Z) \
MSG_STORE(tmpblk, C) \
memcpy(output, tmpblk, size); \
UPDATE_STATE(M)
#define DECRYPT(output, input) \
MSG_LOAD(input, C) \
MAKE_STRM \
XOR_BLOCK(M, C, Z) \
MSG_STORE(output, M) \
UPDATE_STATE(M)
#define DECRYPT_LAST_BLOCK(output, input, size) \
uint8_t tmpblk[ROCCA_MSG_BLOCK_SIZE] = {0}; \
uint8_t tmpmsk[ROCCA_MSG_BLOCK_SIZE] = {0}; \
__m128i mask[M_NUM]; \
memcpy(tmpblk, input, size); \
memset(tmpmsk, 0xFF , size); \
MSG_LOAD(tmpblk, C ) \
MSG_LOAD(tmpmsk, mask) \
MAKE_STRM \
MASKXOR_BLOCK(M, C, Z, mask) \
MSG_STORE(tmpblk, M) \
memcpy(output, tmpblk, size); \
UPDATE_STATE(M)
#define SET_AD_BITLEN_MSG_BITLEN(sizeAD, sizeM) \
uint8_t bitlenAD[16]; \
uint8_t bitlenM [16]; \
ENCODE_IN_LITTLE_ENDIAN(bitlenAD, sizeAD); \
ENCODE_IN_LITTLE_ENDIAN(bitlenM , sizeM ); \
M[0] = load(bitlenAD); \
M[1] = load(bitlenM );
#define MAKE_TAG(sizeAD, sizeM, tag) \
SET_AD_BITLEN_MSG_BITLEN(sizeAD, sizeM) \
for(size_t i = 0; i < TAG_LOOP; ++i) { \
UPDATE_STATE(M) \
} \
__m128i tag128a = setzero(); \
for(size_t i = 0; i <= 3; ++i) { \
tag128a = xor(tag128a, state[i]); \
} \
__m128i tag128b = setzero(); \
for(size_t i = 4; i <= 6; ++i) { \
tag128b = xor(tag128b, state[i]); \
} \
store((tag) , tag128a); \
store((tag)+16, tag128b);
static const uint8_t Z0[] = {0xcd,0x65,0xef,0x23,0x91, \
0x44,0x37,0x71,0x22,0xae,0x28,0xd7,0x98,0x2f,0x8a,0x42};
static const uint8_t Z1[] = {0xbc,0xdb,0x89,0x81,0xa5, \
0xdb,0xb5,0xe9,0x2f,0x3b,0x4d,0xec,0xcf,0xfb,0xc0,0xb5};
void rocca_init(rocca_context * ctx, const uint8_t * key, \
const uint8_t * iv) {
VARS4UPDATE
INIT_STATE(key, iv);
COPY_FROM_LOCAL_IN_INIT(ctx);
ctx->size_ad = 0;
ctx->size_m = 0;
}
void rocca_add_ad(rocca_context * ctx, const uint8_t * in, size_t size)
{
VARS4UPDATE
COPY_TO_LOCAL(ctx);
size_t i = 0;
for(size_t size2 = FLOORTO(size, ROCCA_MSG_BLOCK_SIZE); \
i < size2; i += ROCCA_MSG_BLOCK_SIZE) {
ADD_AD(in + i);
}
if(i < size) {
ADD_AD_LAST_BLOCK(in + i, size - i);
}
COPY_FROM_LOCAL(ctx);
ctx->size_ad += size;
}
void rocca_encrypt(rocca_context * ctx, uint8_t * out, \
const uint8_t * in, size_t size) {
VARS4ENCRYPT
COPY_TO_LOCAL(ctx);
size_t i = 0;
for(size_t size2 = FLOORTO(size, ROCCA_MSG_BLOCK_SIZE); \
i < size2; i += ROCCA_MSG_BLOCK_SIZE) {
ENCRYPT(out + i, in + i);
}
if(i < size) {
ENCRYPT_LAST_BLOCK(out + i, in + i, size - i);
}
COPY_FROM_LOCAL(ctx);
ctx->size_m += size;
}
void rocca_decrypt(rocca_context * ctx, uint8_t * out, \
const uint8_t * in, size_t size) {
VARS4ENCRYPT
COPY_TO_LOCAL(ctx);
size_t i = 0;
for(size_t size2 = FLOORTO(size, ROCCA_MSG_BLOCK_SIZE); \
i < size2; i += ROCCA_MSG_BLOCK_SIZE) {
DECRYPT(out + i, in + i);
}
if(i < size) {
DECRYPT_LAST_BLOCK(out + i, in + i, size - i);
}
COPY_FROM_LOCAL(ctx);
ctx->size_m += size;
}
void rocca_tag(rocca_context * ctx, uint8_t *tag) {
VARS4UPDATE
COPY_TO_LOCAL_IN_TAG(ctx);
MAKE_TAG(ctx->size_ad, ctx->size_m, tag);
}
This section gives three test vectors of Rocca-S. The least significant octet of the vector is shown on the left and the first 128-bit value is shown on the first line.¶
=== test vector #1=== key = 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 nonce = 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 associated data = 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 plaintext = 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ciphertext = 9a c3 32 64 95 a8 d4 14 fe 40 7f 47 b5 44 10 50 24 81 cf 79 ca b8 c0 a6 69 32 3e 07 71 1e 46 17 0d e5 b2 fb ba 0f ae 8d e7 c1 fc ca ee fc 36 26 24 fc fd c1 5f 8b b3 e6 44 57 e8 b7 e3 75 57 bb tag = 8d f9 34 d1 48 37 10 c9 41 0f 6a 08 9c 4c ed 97 91 90 1b 7e 2e 66 12 06 20 2d b2 cc 7a 24 a3 86 === test vector #2=== key = 01 01 01 01 01 01 01 01 01 01 01 01 01 01 01 01 01 01 01 01 01 01 01 01 01 01 01 01 01 01 01 01 nonce = 01 01 01 01 01 01 01 01 01 01 01 01 01 01 01 01 associated data = 01 01 01 01 01 01 01 01 01 01 01 01 01 01 01 01 01 01 01 01 01 01 01 01 01 01 01 01 01 01 01 01 plaintext = 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ciphertext = 55 9e cb 25 3b cf e2 6b 48 3b f0 0e 9c 74 83 45 97 8f f9 21 03 6a 6c 1f dc b7 12 17 28 36 50 4f bc 64 d4 30 a7 3f c6 7a cd 3c 3b 9c 19 76 d8 07 90 f4 83 57 e7 fe 0c 06 82 62 45 69 d3 a6 58 fb tag = c1 fd f3 97 62 ec a7 7d a8 b0 f1 da e5 ff f7 5a 92 fb 0a df a7 94 0a 28 c8 ca db bb e8 e4 ca 8d === test vector #3=== key = 01 23 45 67 89 ab cd ef 01 23 45 67 89 ab cd ef 01 23 45 67 89 ab cd ef 01 23 45 67 89 ab cd ef nonce = 01 23 45 67 89 ab cd ef 01 23 45 67 89 ab cd ef associated data = 01 23 45 67 89 ab cd ef 01 23 45 67 89 ab cd ef 01 23 45 67 89 ab cd ef 01 23 45 67 89 ab cd ef plaintext = 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ciphertext = b5 fc 4e 2a 72 b8 6d 1a 13 3c 0f 02 02 bd f7 90 af 14 a2 4b 2c db 67 6e 42 78 65 e1 2f cc 9d 30 21 d1 84 18 fc 75 dc 19 12 dd 2c d7 9a 3b ee b2 a9 8b 23 5d e2 29 9b 9d da 93 fd 2b 5a c8 f4 36 tag = a0 78 e1 35 1e f2 42 0c 8e 3a 93 fd 31 f5 b1 13 5b 15 31 5a 5f 20 55 34 14 8e fb cd 63 f7 9f 00¶
This draft is partially supported by a contract of "Research and development on new generation cryptography for secure wireless communication services" among "Research and Development for Expansion of Radio Wave Resources (JPJ000254)", which was supported by the Ministry of Internal Affairs and Communications, Japan.¶