fzorb/REDACTED-rig
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Panthera algo implementation

Browse source
This commit is contained in:
MoneroOcean 2020-07-16 16:43:19 -07:00
parent 50b1bf8a1d
commit 5eafa9e455
20 changed files with 2897 additions and 426 deletions
Download patch file Download diff file Expand all files Collapse all files

15
cmake/randomx.cmake
View file

@ -51,11 +51,22 @@ if (WITH_RANDOMX)
src/crypto/randomx/defyx/KeccakSpongeWidth1600.c
src/crypto/randomx/defyx/KeccakSpongeWidth1600.h
src/crypto/randomx/defyx/Phases.h
src/crypto/randomx/defyx/sha256.c
src/crypto/randomx/defyx/sha256.h
src/crypto/randomx/defyx/sysendian.h
src/crypto/randomx/defyx/yescrypt.h
src/crypto/randomx/defyx/yescrypt-best.c
src/crypto/randomx/defyx/yescrypt-opt.c
src/crypto/randomx/defyx/yescrypt-platform.c
src/crypto/randomx/defyx/yescrypt-ref.c
src/crypto/randomx/defyx/yescrypt-simd.c
src/crypto/randomx/defyx/yescrypt.h
src/crypto/randomx/panthera/insecure_memzero.h
src/crypto/randomx/panthera/sha256.c
src/crypto/randomx/panthera/sha256.h
src/crypto/randomx/panthera/sysendian.h
src/crypto/randomx/panthera/yespower-opt.c
src/crypto/randomx/panthera/yespower-platform.c
src/crypto/randomx/panthera/yespower-ref.c
src/crypto/randomx/panthera/yespower.h
)
if (CMAKE_C_COMPILER_ID MATCHES MSVC)

2
src/backend/common/Threads.cpp
View file

@ -138,7 +138,7 @@ xmrig::String xmrig::Threads<T>::profileName(const Algorithm &algorithm, bool st
}
}
if (std::is_same<T, CpuThreads>::value && name == "defyx" && has("rx")) return "rx";
if (std::is_same<T, CpuThreads>::value && (name == "defyx" || name == "panthera") && has("rx")) return "rx";
if (has(kAsterisk)) {
return kAsterisk;

4
src/backend/cpu/CpuConfig_gen.h
View file

@ -142,6 +142,10 @@ size_t inline generate<Algorithm::RANDOM_X>(Threads<CpuThreads> &threads, uint32
count += generate("defyx", threads, Algorithm::RX_DEFYX, limit);
}
if (!threads.isExist(Algorithm::RX_XLA)) {
count += generate("panthera", threads, Algorithm::RX_XLA, limit);
}
count += generate("rx", threads, Algorithm::RX_0, limit);
return count;

1
src/backend/opencl/cl/cn/algorithm.cl
View file

@ -29,6 +29,7 @@
#define ALGO_ASTROBWT_DERO 28
#define ALGO_KAWPOW_RVN 29
#define ALGO_RX_DEFYX 30
#define ALGO_RX_XLA 31
#define FAMILY_UNKNOWN 0
#define FAMILY_CN 1

2
src/base/crypto/Algorithm.cpp
View file

@ -115,6 +115,8 @@ static AlgoName const algorithm_names[] = {
{ "RandomKEVA", nullptr, Algorithm::RX_KEVA },
{ "defyx", "defyx", Algorithm::RX_DEFYX },
{ "DefyX", nullptr, Algorithm::RX_DEFYX },
{ "panthera", "panthera", Algorithm::RX_XLA },
{ "Panthera", "panthera", Algorithm::RX_XLA },
# endif
# ifdef XMRIG_ALGO_ARGON2
{ "argon2/chukwa", nullptr, Algorithm::AR2_CHUKWA },

1
src/base/crypto/Algorithm.h
View file

@ -78,6 +78,7 @@ public:
ASTROBWT_DERO, // "astrobwt" AstroBWT (Dero)
KAWPOW_RVN, // "kawpow/rvn" KawPow (RVN)
RX_DEFYX, // "defyx" DefyX (Scala).
RX_XLA, // "panthera" Panthera (Scala2).
MAX,
MIN = 0,
INVALID = -1,

1
src/core/Benchmark.cpp
View file

@ -142,6 +142,7 @@ double Benchmark::get_algo_perf(Algorithm::Id algo) const {
case Algorithm::RX_ARQ: return m_bench_algo_perf[BenchAlgo::RX_ARQ];
case Algorithm::RX_KEVA: return m_bench_algo_perf[BenchAlgo::RX_KEVA];
case Algorithm::RX_DEFYX: return m_bench_algo_perf[BenchAlgo::RX_DEFYX];
case Algorithm::RX_XLA: return m_bench_algo_perf[BenchAlgo::RX_XLA];
default: return 0.0f;
}
}

2
src/core/Benchmark.h
View file

@ -46,6 +46,7 @@ class Benchmark : public IJobResultListener {
RX_ARQ, // "rx/arq" RandomARQ (Arqma).
RX_KEVA, // "rx/keva" RandomKEVA (Keva).
RX_DEFYX, // "defyx" DefyX (Scala).
RX_XLA, // "panthera" Panthera (Scala2).
MAX,
MIN = 0,
INVALID = -1,
@ -67,6 +68,7 @@ class Benchmark : public IJobResultListener {
Algorithm::RX_ARQ,
Algorithm::RX_KEVA,
Algorithm::RX_DEFYX,
Algorithm::RX_XLA,
};
Job* m_bench_job[BenchAlgo::MAX];

411
src/crypto/randomx/defyx/sha256.c
View file

@ -1,411 +0,0 @@
/*-
* Copyright 2005,2007,2009 Colin Percival
* All rights reserved.
*
* 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.
*/
#include <sys/types.h>
#include <stdint.h>
#include <string.h>
#include "sysendian.h"
#include "sha256.h"
/*
* Encode a length len/4 vector of (uint32_t) into a length len vector of
* (unsigned char) in big-endian form. Assumes len is a multiple of 4.
*/
static void
be32enc_vect(unsigned char *dst, const uint32_t *src, size_t len)
{
size_t i;
for (i = 0; i < len / 4; i++)
be32enc(dst + i * 4, src[i]);
}
/*
* Decode a big-endian length len vector of (unsigned char) into a length
* len/4 vector of (uint32_t). Assumes len is a multiple of 4.
*/
static void
be32dec_vect(uint32_t *dst, const unsigned char *src, size_t len)
{
size_t i;
for (i = 0; i < len / 4; i++)
dst[i] = be32dec(src + i * 4);
}
/* Elementary functions used by SHA256 */
#define Ch(x, y, z) ((x & (y ^ z)) ^ z)
#define Maj(x, y, z) ((x & (y | z)) | (y & z))
#define SHR(x, n) (x >> n)
#define ROTR(x, n) ((x >> n) | (x << (32 - n)))
#define S0(x) (ROTR(x, 2) ^ ROTR(x, 13) ^ ROTR(x, 22))
#define S1(x) (ROTR(x, 6) ^ ROTR(x, 11) ^ ROTR(x, 25))
#define s0(x) (ROTR(x, 7) ^ ROTR(x, 18) ^ SHR(x, 3))
#define s1(x) (ROTR(x, 17) ^ ROTR(x, 19) ^ SHR(x, 10))
/* SHA256 round function */
#define RND(a, b, c, d, e, f, g, h, k) \
t0 = h + S1(e) + Ch(e, f, g) + k; \
t1 = S0(a) + Maj(a, b, c); \
d += t0; \
h = t0 + t1;
/* Adjusted round function for rotating state */
#define RNDr(S, W, i, k) \
RND(S[(64 - i) % 8], S[(65 - i) % 8], \
S[(66 - i) % 8], S[(67 - i) % 8], \
S[(68 - i) % 8], S[(69 - i) % 8], \
S[(70 - i) % 8], S[(71 - i) % 8], \
W[i] + k)
/*
* SHA256 block compression function. The 256-bit state is transformed via
* the 512-bit input block to produce a new state.
*/
static void
SHA256_Transform(uint32_t * state, const unsigned char block[64])
{
uint32_t W[64];
uint32_t S[8];
uint32_t t0, t1;
int i;
/* 1. Prepare message schedule W. */
be32dec_vect(W, block, 64);
for (i = 16; i < 64; i++)
W[i] = s1(W[i - 2]) + W[i - 7] + s0(W[i - 15]) + W[i - 16];
/* 2. Initialize working variables. */
memcpy(S, state, 32);
/* 3. Mix. */
RNDr(S, W, 0, 0x428a2f98);
RNDr(S, W, 1, 0x71374491);
RNDr(S, W, 2, 0xb5c0fbcf);
RNDr(S, W, 3, 0xe9b5dba5);
RNDr(S, W, 4, 0x3956c25b);
RNDr(S, W, 5, 0x59f111f1);
RNDr(S, W, 6, 0x923f82a4);
RNDr(S, W, 7, 0xab1c5ed5);
RNDr(S, W, 8, 0xd807aa98);
RNDr(S, W, 9, 0x12835b01);
RNDr(S, W, 10, 0x243185be);
RNDr(S, W, 11, 0x550c7dc3);
RNDr(S, W, 12, 0x72be5d74);
RNDr(S, W, 13, 0x80deb1fe);
RNDr(S, W, 14, 0x9bdc06a7);
RNDr(S, W, 15, 0xc19bf174);
RNDr(S, W, 16, 0xe49b69c1);
RNDr(S, W, 17, 0xefbe4786);
RNDr(S, W, 18, 0x0fc19dc6);
RNDr(S, W, 19, 0x240ca1cc);
RNDr(S, W, 20, 0x2de92c6f);
RNDr(S, W, 21, 0x4a7484aa);
RNDr(S, W, 22, 0x5cb0a9dc);
RNDr(S, W, 23, 0x76f988da);
RNDr(S, W, 24, 0x983e5152);
RNDr(S, W, 25, 0xa831c66d);
RNDr(S, W, 26, 0xb00327c8);
RNDr(S, W, 27, 0xbf597fc7);
RNDr(S, W, 28, 0xc6e00bf3);
RNDr(S, W, 29, 0xd5a79147);
RNDr(S, W, 30, 0x06ca6351);
RNDr(S, W, 31, 0x14292967);
RNDr(S, W, 32, 0x27b70a85);
RNDr(S, W, 33, 0x2e1b2138);
RNDr(S, W, 34, 0x4d2c6dfc);
RNDr(S, W, 35, 0x53380d13);
RNDr(S, W, 36, 0x650a7354);
RNDr(S, W, 37, 0x766a0abb);
RNDr(S, W, 38, 0x81c2c92e);
RNDr(S, W, 39, 0x92722c85);
RNDr(S, W, 40, 0xa2bfe8a1);
RNDr(S, W, 41, 0xa81a664b);
RNDr(S, W, 42, 0xc24b8b70);
RNDr(S, W, 43, 0xc76c51a3);
RNDr(S, W, 44, 0xd192e819);
RNDr(S, W, 45, 0xd6990624);
RNDr(S, W, 46, 0xf40e3585);
RNDr(S, W, 47, 0x106aa070);
RNDr(S, W, 48, 0x19a4c116);
RNDr(S, W, 49, 0x1e376c08);
RNDr(S, W, 50, 0x2748774c);
RNDr(S, W, 51, 0x34b0bcb5);
RNDr(S, W, 52, 0x391c0cb3);
RNDr(S, W, 53, 0x4ed8aa4a);
RNDr(S, W, 54, 0x5b9cca4f);
RNDr(S, W, 55, 0x682e6ff3);
RNDr(S, W, 56, 0x748f82ee);
RNDr(S, W, 57, 0x78a5636f);
RNDr(S, W, 58, 0x84c87814);
RNDr(S, W, 59, 0x8cc70208);
RNDr(S, W, 60, 0x90befffa);
RNDr(S, W, 61, 0xa4506ceb);
RNDr(S, W, 62, 0xbef9a3f7);
RNDr(S, W, 63, 0xc67178f2);
/* 4. Mix local working variables into global state */
for (i = 0; i < 8; i++)
state[i] += S[i];
/* Clean the stack. */
memset(W, 0, 256);
memset(S, 0, 32);
t0 = t1 = 0;
}
static unsigned char PAD[64] = {
0x80, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0
};
/* Add padding and terminating bit-count. */
static void
SHA256_Pad(SHA256_CTX_Y * ctx)
{
unsigned char len[8];
uint32_t r, plen;
/*
* Convert length to a vector of bytes -- we do this now rather
* than later because the length will change after we pad.
*/
be32enc_vect(len, ctx->count, 8);
/* Add 1--64 bytes so that the resulting length is 56 mod 64 */
r = (ctx->count[1] >> 3) & 0x3f;
plen = (r < 56) ? (56 - r) : (120 - r);
SHA256_Update_Y(ctx, PAD, (size_t)plen);
/* Add the terminating bit-count */
SHA256_Update_Y(ctx, len, 8);
}
/* SHA-256 initialization. Begins a SHA-256 operation. */
void
SHA256_Init_Y(SHA256_CTX_Y * ctx)
{
/* Zero bits processed so far */
ctx->count[0] = ctx->count[1] = 0;
/* Magic initialization constants */
ctx->state[0] = 0x6A09E667;
ctx->state[1] = 0xBB67AE85;
ctx->state[2] = 0x3C6EF372;
ctx->state[3] = 0xA54FF53A;
ctx->state[4] = 0x510E527F;
ctx->state[5] = 0x9B05688C;
ctx->state[6] = 0x1F83D9AB;
ctx->state[7] = 0x5BE0CD19;
}
/* Add bytes into the hash */
void
SHA256_Update_Y(SHA256_CTX_Y * ctx, const void *in, size_t len)
{
uint32_t bitlen[2];
uint32_t r;
const unsigned char *src = in;
/* Number of bytes left in the buffer from previous updates */
r = (ctx->count[1] >> 3) & 0x3f;
/* Convert the length into a number of bits */
bitlen[1] = ((uint32_t)len) << 3;
bitlen[0] = (uint32_t)(len >> 29);
/* Update number of bits */
if ((ctx->count[1] += bitlen[1]) < bitlen[1])
ctx->count[0]++;
ctx->count[0] += bitlen[0];
/* Handle the case where we don't need to perform any transforms */
if (len < 64 - r) {
memcpy(&ctx->buf[r], src, len);
return;
}
/* Finish the current block */
memcpy(&ctx->buf[r], src, 64 - r);
SHA256_Transform(ctx->state, ctx->buf);
src += 64 - r;
len -= 64 - r;
/* Perform complete blocks */
while (len >= 64) {
SHA256_Transform(ctx->state, src);
src += 64;
len -= 64;
}
/* Copy left over data into buffer */
memcpy(ctx->buf, src, len);
}
/*
* SHA-256 finalization. Pads the input data, exports the hash value,
* and clears the context state.
*/
void
SHA256_Final_Y(unsigned char digest[32], SHA256_CTX_Y * ctx)
{
/* Add padding */
SHA256_Pad(ctx);
/* Write the hash */
be32enc_vect(digest, ctx->state, 32);
/* Clear the context state */
memset((void *)ctx, 0, sizeof(*ctx));
}
/* Initialize an HMAC-SHA256 operation with the given key. */
void
HMAC_SHA256_Init_Y(HMAC_SHA256_CTX_Y * ctx, const void * _K, size_t Klen)
{
unsigned char pad[64];
unsigned char khash[32];
const unsigned char * K = _K;
size_t i;
/* If Klen > 64, the key is really SHA256(K). */
if (Klen > 64) {
SHA256_Init_Y(&ctx->ictx);
SHA256_Update_Y(&ctx->ictx, K, Klen);
SHA256_Final_Y(khash, &ctx->ictx);
K = khash;
Klen = 32;
}
/* Inner SHA256 operation is SHA256(K xor [block of 0x36] || data). */
SHA256_Init_Y(&ctx->ictx);
memset(pad, 0x36, 64);
for (i = 0; i < Klen; i++)
pad[i] ^= K[i];
SHA256_Update_Y(&ctx->ictx, pad, 64);
/* Outer SHA256 operation is SHA256(K xor [block of 0x5c] || hash). */
SHA256_Init_Y(&ctx->octx);
memset(pad, 0x5c, 64);
for (i = 0; i < Klen; i++)
pad[i] ^= K[i];
SHA256_Update_Y(&ctx->octx, pad, 64);
/* Clean the stack. */
memset(khash, 0, 32);
}
/* Add bytes to the HMAC-SHA256 operation. */
void
HMAC_SHA256_Update_Y(HMAC_SHA256_CTX_Y * ctx, const void *in, size_t len)
{
/* Feed data to the inner SHA256 operation. */
SHA256_Update_Y(&ctx->ictx, in, len);
}
/* Finish an HMAC-SHA256 operation. */
void
HMAC_SHA256_Final_Y(unsigned char digest[32], HMAC_SHA256_CTX_Y * ctx)
{
unsigned char ihash[32];
/* Finish the inner SHA256 operation. */
SHA256_Final_Y(ihash, &ctx->ictx);
/* Feed the inner hash to the outer SHA256 operation. */
SHA256_Update_Y(&ctx->octx, ihash, 32);
/* Finish the outer SHA256 operation. */
SHA256_Final_Y(digest, &ctx->octx);
/* Clean the stack. */
memset(ihash, 0, 32);
}
/**
* PBKDF2_SHA256(passwd, passwdlen, salt, saltlen, c, buf, dkLen):
* Compute PBKDF2(passwd, salt, c, dkLen) using HMAC-SHA256 as the PRF, and
* write the output to buf. The value dkLen must be at most 32 * (2^32 - 1).
*/
void
PBKDF2_SHA256(const uint8_t * passwd, size_t passwdlen, const uint8_t * salt,
size_t saltlen, uint64_t c, uint8_t * buf, size_t dkLen)
{
HMAC_SHA256_CTX_Y PShctx, hctx;
size_t i;
uint8_t ivec[4];
uint8_t U[32];
uint8_t T[32];
uint64_t j;
int k;
size_t clen;
/* Compute HMAC state after processing P and S. */
HMAC_SHA256_Init_Y(&PShctx, passwd, passwdlen);
HMAC_SHA256_Update_Y(&PShctx, salt, saltlen);
/* Iterate through the blocks. */
for (i = 0; i * 32 < dkLen; i++) {
/* Generate INT(i + 1). */
be32enc(ivec, (uint32_t)(i + 1));
/* Compute U_1 = PRF(P, S || INT(i)). */
memcpy(&hctx, &PShctx, sizeof(HMAC_SHA256_CTX_Y));
HMAC_SHA256_Update_Y(&hctx, ivec, 4);
HMAC_SHA256_Final_Y(U, &hctx);
/* T_i = U_1 ... */
memcpy(T, U, 32);
for (j = 2; j <= c; j++) {
/* Compute U_j. */
HMAC_SHA256_Init_Y(&hctx, passwd, passwdlen);
HMAC_SHA256_Update_Y(&hctx, U, 32);
HMAC_SHA256_Final_Y(U, &hctx);
/* ... xor U_j ... */
for (k = 0; k < 32; k++)
T[k] ^= U[k];
}
/* Copy as many bytes as necessary into buf. */
clen = dkLen - i * 32;
if (clen > 32)
clen = 32;
memcpy(&buf[i * 32], T, clen);
}
/* Clean PShctx, since we never called _Final on it. */
memset(&PShctx, 0, sizeof(HMAC_SHA256_CTX_Y));
}

1
src/crypto/randomx/panthera/insecure_memzero.h Normal file
View file

@ -0,0 +1 @@
#define insecure_memzero(buf, len) /* empty */

646
src/crypto/randomx/panthera/sha256.c Normal file
View file

@ -0,0 +1,646 @@
/*-
* Copyright 2005-2016 Colin Percival
* Copyright 2016-2018 Alexander Peslyak
* All rights reserved.
*
* 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.
*/
#include <assert.h>
#include <stdint.h>
#include <string.h>
#include "insecure_memzero.h"
#include "sysendian.h"
#include "sha256.h"
#ifdef __ICC
/* Miscompile with icc 14.0.0 (at least), so don't use restrict there */
#define restrict
#elif __STDC_VERSION__ >= 199901L
/* Have restrict */
#elif defined(__GNUC__)
#define restrict __restrict
#else
#define restrict
#endif
/*
* Encode a length len*2 vector of (uint32_t) into a length len*8 vector of
* (uint8_t) in big-endian form.
*/
static void
be32enc_vect(uint8_t * dst, const uint32_t * src, size_t len)
{
/* Encode vector, two words at a time. */
do {
be32enc(&dst[0], src[0]);
be32enc(&dst[4], src[1]);
src += 2;
dst += 8;
} while (--len);
}
/*
* Decode a big-endian length len*8 vector of (uint8_t) into a length
* len*2 vector of (uint32_t).
*/
static void
be32dec_vect(uint32_t * dst, const uint8_t * src, size_t len)
{
/* Decode vector, two words at a time. */
do {
dst[0] = be32dec(&src[0]);
dst[1] = be32dec(&src[4]);
src += 8;
dst += 2;
} while (--len);
}
/* SHA256 round constants. */
static const uint32_t Krnd[64] = {
0x428a2f98, 0x71374491, 0xb5c0fbcf, 0xe9b5dba5,
0x3956c25b, 0x59f111f1, 0x923f82a4, 0xab1c5ed5,
0xd807aa98, 0x12835b01, 0x243185be, 0x550c7dc3,
0x72be5d74, 0x80deb1fe, 0x9bdc06a7, 0xc19bf174,
0xe49b69c1, 0xefbe4786, 0x0fc19dc6, 0x240ca1cc,
0x2de92c6f, 0x4a7484aa, 0x5cb0a9dc, 0x76f988da,
0x983e5152, 0xa831c66d, 0xb00327c8, 0xbf597fc7,
0xc6e00bf3, 0xd5a79147, 0x06ca6351, 0x14292967,
0x27b70a85, 0x2e1b2138, 0x4d2c6dfc, 0x53380d13,
0x650a7354, 0x766a0abb, 0x81c2c92e, 0x92722c85,
0xa2bfe8a1, 0xa81a664b, 0xc24b8b70, 0xc76c51a3,
0xd192e819, 0xd6990624, 0xf40e3585, 0x106aa070,
0x19a4c116, 0x1e376c08, 0x2748774c, 0x34b0bcb5,
0x391c0cb3, 0x4ed8aa4a, 0x5b9cca4f, 0x682e6ff3,
0x748f82ee, 0x78a5636f, 0x84c87814, 0x8cc70208,
0x90befffa, 0xa4506ceb, 0xbef9a3f7, 0xc67178f2
};
/* Elementary functions used by SHA256 */
#define Ch(x, y, z) ((x & (y ^ z)) ^ z)
#define Maj(x, y, z) ((x & (y | z)) | (y & z))
#define SHR(x, n) (x >> n)
#define ROTR(x, n) ((x >> n) | (x << (32 - n)))
#define S0(x) (ROTR(x, 2) ^ ROTR(x, 13) ^ ROTR(x, 22))
#define S1(x) (ROTR(x, 6) ^ ROTR(x, 11) ^ ROTR(x, 25))
#define s0(x) (ROTR(x, 7) ^ ROTR(x, 18) ^ SHR(x, 3))
#define s1(x) (ROTR(x, 17) ^ ROTR(x, 19) ^ SHR(x, 10))
/* SHA256 round function */
#define RND(a, b, c, d, e, f, g, h, k) \
h += S1(e) + Ch(e, f, g) + k; \
d += h; \
h += S0(a) + Maj(a, b, c);
/* Adjusted round function for rotating state */
#define RNDr(S, W, i, ii) \
RND(S[(64 - i) % 8], S[(65 - i) % 8], \
S[(66 - i) % 8], S[(67 - i) % 8], \
S[(68 - i) % 8], S[(69 - i) % 8], \
S[(70 - i) % 8], S[(71 - i) % 8], \
W[i + ii] + Krnd[i + ii])
/* Message schedule computation */
#define MSCH(W, ii, i) \
W[i + ii + 16] = s1(W[i + ii + 14]) + W[i + ii + 9] + s0(W[i + ii + 1]) + W[i + ii]
/*
* SHA256 block compression function. The 256-bit state is transformed via
* the 512-bit input block to produce a new state.
*/
static void
SHA256_Transform(uint32_t state[static restrict 8],
const uint8_t block[static restrict 64],
uint32_t W[static restrict 64], uint32_t S[static restrict 8])
{
int i;
/* 1. Prepare the first part of the message schedule W. */
be32dec_vect(W, block, 8);
/* 2. Initialize working variables. */
memcpy(S, state, 32);
/* 3. Mix. */
for (i = 0; i < 64; i += 16) {
RNDr(S, W, 0, i);
RNDr(S, W, 1, i);
RNDr(S, W, 2, i);
RNDr(S, W, 3, i);
RNDr(S, W, 4, i);
RNDr(S, W, 5, i);
RNDr(S, W, 6, i);
RNDr(S, W, 7, i);
RNDr(S, W, 8, i);
RNDr(S, W, 9, i);
RNDr(S, W, 10, i);
RNDr(S, W, 11, i);
RNDr(S, W, 12, i);
RNDr(S, W, 13, i);
RNDr(S, W, 14, i);
RNDr(S, W, 15, i);
if (i == 48)
break;
MSCH(W, 0, i);
MSCH(W, 1, i);
MSCH(W, 2, i);
MSCH(W, 3, i);
MSCH(W, 4, i);
MSCH(W, 5, i);
MSCH(W, 6, i);
MSCH(W, 7, i);
MSCH(W, 8, i);
MSCH(W, 9, i);
MSCH(W, 10, i);
MSCH(W, 11, i);
MSCH(W, 12, i);
MSCH(W, 13, i);
MSCH(W, 14, i);
MSCH(W, 15, i);
}
/* 4. Mix local working variables into global state. */
state[0] += S[0];
state[1] += S[1];
state[2] += S[2];
state[3] += S[3];
state[4] += S[4];
state[5] += S[5];
state[6] += S[6];
state[7] += S[7];
}
static const uint8_t PAD[64] = {
0x80, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0
};
/* Add padding and terminating bit-count. */
static void
SHA256_Pad(SHA256_CTX * ctx, uint32_t tmp32[static restrict 72])
{
size_t r;
/* Figure out how many bytes we have buffered. */
r = (ctx->count >> 3) & 0x3f;
/* Pad to 56 mod 64, transforming if we finish a block en route. */
if (r < 56) {
/* Pad to 56 mod 64. */
memcpy(&ctx->buf[r], PAD, 56 - r);
} else {
/* Finish the current block and mix. */
memcpy(&ctx->buf[r], PAD, 64 - r);
SHA256_Transform(ctx->state, ctx->buf, &tmp32[0], &tmp32[64]);
/* The start of the final block is all zeroes. */
memset(&ctx->buf[0], 0, 56);
}
/* Add the terminating bit-count. */
be64enc(&ctx->buf[56], ctx->count);
/* Mix in the final block. */
SHA256_Transform(ctx->state, ctx->buf, &tmp32[0], &tmp32[64]);
}
/* Magic initialization constants. */
static const uint32_t initial_state[8] = {
0x6A09E667, 0xBB67AE85, 0x3C6EF372, 0xA54FF53A,
0x510E527F, 0x9B05688C, 0x1F83D9AB, 0x5BE0CD19
};
/**
* SHA256_Init(ctx):
* Initialize the SHA256 context ${ctx}.
*/
void
SHA256_Init(SHA256_CTX * ctx)
{
/* Zero bits processed so far. */
ctx->count = 0;
/* Initialize state. */
memcpy(ctx->state, initial_state, sizeof(initial_state));
}
/**
* SHA256_Update(ctx, in, len):
* Input ${len} bytes from ${in} into the SHA256 context ${ctx}.
*/
static void
_SHA256_Update(SHA256_CTX * ctx, const void * in, size_t len,
uint32_t tmp32[static restrict 72])
{
uint32_t r;
const uint8_t * src = in;
/* Return immediately if we have nothing to do. */
if (len == 0)
return;
/* Number of bytes left in the buffer from previous updates. */
r = (ctx->count >> 3) & 0x3f;
/* Update number of bits. */
ctx->count += (uint64_t)(len) << 3;
/* Handle the case where we don't need to perform any transforms. */
if (len < 64 - r) {
memcpy(&ctx->buf[r], src, len);
return;
}
/* Finish the current block. */
memcpy(&ctx->buf[r], src, 64 - r);
SHA256_Transform(ctx->state, ctx->buf, &tmp32[0], &tmp32[64]);
src += 64 - r;
len -= 64 - r;
/* Perform complete blocks. */
while (len >= 64) {
SHA256_Transform(ctx->state, src, &tmp32[0], &tmp32[64]);
src += 64;
len -= 64;
}
/* Copy left over data into buffer. */
memcpy(ctx->buf, src, len);
}
/* Wrapper function for intermediate-values sanitization. */
void
SHA256_Update(SHA256_CTX * ctx, const void * in, size_t len)
{
uint32_t tmp32[72];
/* Call the real function. */
_SHA256_Update(ctx, in, len, tmp32);
/* Clean the stack. */
insecure_memzero(tmp32, 288);
}
/**
* SHA256_Final(digest, ctx):
* Output the SHA256 hash of the data input to the context ${ctx} into the
* buffer ${digest}.
*/
static void
_SHA256_Final(uint8_t digest[32], SHA256_CTX * ctx,
uint32_t tmp32[static restrict 72])
{
/* Add padding. */
SHA256_Pad(ctx, tmp32);
/* Write the hash. */
be32enc_vect(digest, ctx->state, 4);
}
/* Wrapper function for intermediate-values sanitization. */
void
SHA256_Final(uint8_t digest[32], SHA256_CTX * ctx)
{
uint32_t tmp32[72];
/* Call the real function. */
_SHA256_Final(digest, ctx, tmp32);
/* Clear the context state. */
insecure_memzero(ctx, sizeof(SHA256_CTX));
/* Clean the stack. */
insecure_memzero(tmp32, 288);
}
/**
* SHA256_Buf(in, len, digest):
* Compute the SHA256 hash of ${len} bytes from ${in} and write it to ${digest}.
*/
void
SHA256_Buf(const void * in, size_t len, uint8_t digest[32])
{
SHA256_CTX ctx;
uint32_t tmp32[72];
SHA256_Init(&ctx);
_SHA256_Update(&ctx, in, len, tmp32);
_SHA256_Final(digest, &ctx, tmp32);
/* Clean the stack. */
insecure_memzero(&ctx, sizeof(SHA256_CTX));
insecure_memzero(tmp32, 288);
}
/**
* HMAC_SHA256_Init(ctx, K, Klen):
* Initialize the HMAC-SHA256 context ${ctx} with ${Klen} bytes of key from
* ${K}.
*/
static void
_HMAC_SHA256_Init(HMAC_SHA256_CTX * ctx, const void * _K, size_t Klen,
uint32_t tmp32[static restrict 72], uint8_t pad[static restrict 64],
uint8_t khash[static restrict 32])
{
const uint8_t * K = _K;
size_t i;
/* If Klen > 64, the key is really SHA256(K). */
if (Klen > 64) {
SHA256_Init(&ctx->ictx);
_SHA256_Update(&ctx->ictx, K, Klen, tmp32);
_SHA256_Final(khash, &ctx->ictx, tmp32);
K = khash;
Klen = 32;
}
/* Inner SHA256 operation is SHA256(K xor [block of 0x36] || data). */
SHA256_Init(&ctx->ictx);
memset(pad, 0x36, 64);
for (i = 0; i < Klen; i++)
pad[i] ^= K[i];
_SHA256_Update(&ctx->ictx, pad, 64, tmp32);
/* Outer SHA256 operation is SHA256(K xor [block of 0x5c] || hash). */
SHA256_Init(&ctx->octx);
memset(pad, 0x5c, 64);
for (i = 0; i < Klen; i++)
pad[i] ^= K[i];
_SHA256_Update(&ctx->octx, pad, 64, tmp32);
}
/* Wrapper function for intermediate-values sanitization. */
void
HMAC_SHA256_Init(HMAC_SHA256_CTX * ctx, const void * _K, size_t Klen)
{
uint32_t tmp32[72];
uint8_t pad[64];
uint8_t khash[32];
/* Call the real function. */
_HMAC_SHA256_Init(ctx, _K, Klen, tmp32, pad, khash);
/* Clean the stack. */
insecure_memzero(tmp32, 288);
insecure_memzero(khash, 32);
insecure_memzero(pad, 64);
}
/**
* HMAC_SHA256_Update(ctx, in, len):
* Input ${len} bytes from ${in} into the HMAC-SHA256 context ${ctx}.
*/
static void
_HMAC_SHA256_Update(HMAC_SHA256_CTX * ctx, const void * in, size_t len,
uint32_t tmp32[static restrict 72])
{
/* Feed data to the inner SHA256 operation. */
_SHA256_Update(&ctx->ictx, in, len, tmp32);
}
/* Wrapper function for intermediate-values sanitization. */
void
HMAC_SHA256_Update(HMAC_SHA256_CTX * ctx, const void * in, size_t len)
{
uint32_t tmp32[72];
/* Call the real function. */
_HMAC_SHA256_Update(ctx, in, len, tmp32);
/* Clean the stack. */
insecure_memzero(tmp32, 288);
}
/**
* HMAC_SHA256_Final(digest, ctx):
* Output the HMAC-SHA256 of the data input to the context ${ctx} into the
* buffer ${digest}.
*/
static void
_HMAC_SHA256_Final(uint8_t digest[32], HMAC_SHA256_CTX * ctx,
uint32_t tmp32[static restrict 72], uint8_t ihash[static restrict 32])
{
/* Finish the inner SHA256 operation. */
_SHA256_Final(ihash, &ctx->ictx, tmp32);
/* Feed the inner hash to the outer SHA256 operation. */
_SHA256_Update(&ctx->octx, ihash, 32, tmp32);
/* Finish the outer SHA256 operation. */
_SHA256_Final(digest, &ctx->octx, tmp32);
}
/* Wrapper function for intermediate-values sanitization. */
void
HMAC_SHA256_Final(uint8_t digest[32], HMAC_SHA256_CTX * ctx)
{
uint32_t tmp32[72];
uint8_t ihash[32];
/* Call the real function. */
_HMAC_SHA256_Final(digest, ctx, tmp32, ihash);
/* Clean the stack. */
insecure_memzero(tmp32, 288);
insecure_memzero(ihash, 32);
}
/**
* HMAC_SHA256_Buf(K, Klen, in, len, digest):
* Compute the HMAC-SHA256 of ${len} bytes from ${in} using the key ${K} of
* length ${Klen}, and write the result to ${digest}.
*/
void
HMAC_SHA256_Buf(const void * K, size_t Klen, const void * in, size_t len,
uint8_t digest[32])
{
HMAC_SHA256_CTX ctx;
uint32_t tmp32[72];
uint8_t tmp8[96];
_HMAC_SHA256_Init(&ctx, K, Klen, tmp32, &tmp8[0], &tmp8[64]);
_HMAC_SHA256_Update(&ctx, in, len, tmp32);
_HMAC_SHA256_Final(digest, &ctx, tmp32, &tmp8[0]);
/* Clean the stack. */
insecure_memzero(&ctx, sizeof(HMAC_SHA256_CTX));
insecure_memzero(tmp32, 288);
insecure_memzero(tmp8, 96);
}
/* Add padding and terminating bit-count, but don't invoke Transform yet. */
static int
SHA256_Pad_Almost(SHA256_CTX * ctx, uint8_t len[static restrict 8],
uint32_t tmp32[static restrict 72])
{
uint32_t r;
r = (ctx->count >> 3) & 0x3f;
if (r >= 56)
return -1;
/*
* Convert length to a vector of bytes -- we do this now rather
* than later because the length will change after we pad.
*/
be64enc(len, ctx->count);
/* Add 1--56 bytes so that the resulting length is 56 mod 64. */
_SHA256_Update(ctx, PAD, 56 - r, tmp32);
/* Add the terminating bit-count. */
ctx->buf[63] = len[7];
_SHA256_Update(ctx, len, 7, tmp32);
return 0;
}
/**
* PBKDF2_SHA256(passwd, passwdlen, salt, saltlen, c, buf, dkLen):
* Compute PBKDF2(passwd, salt, c, dkLen) using HMAC-SHA256 as the PRF, and
* write the output to buf. The value dkLen must be at most 32 * (2^32 - 1).
*/
void
PBKDF2_SHA256(const uint8_t * passwd, size_t passwdlen, const uint8_t * salt,
size_t saltlen, uint64_t c, uint8_t * buf, size_t dkLen)
{
HMAC_SHA256_CTX Phctx, PShctx, hctx;
uint32_t tmp32[72];
union {
uint8_t tmp8[96];
uint32_t state[8];
} u;
size_t i;
uint8_t ivec[4];
uint8_t U[32];
uint8_t T[32];
uint64_t j;
int k;
size_t clen;
/* Sanity-check. */
assert(dkLen <= 32 * (size_t)(UINT32_MAX));
if (c == 1 && (dkLen & 31) == 0 && (saltlen & 63) <= 51) {
uint32_t oldcount;
uint8_t * ivecp;
/* Compute HMAC state after processing P and S. */
_HMAC_SHA256_Init(&hctx, passwd, passwdlen,
tmp32, &u.tmp8[0], &u.tmp8[64]);
_HMAC_SHA256_Update(&hctx, salt, saltlen, tmp32);
/* Prepare ictx padding. */
oldcount = hctx.ictx.count & (0x3f << 3);
_HMAC_SHA256_Update(&hctx, "\0\0\0", 4, tmp32);
if ((hctx.ictx.count & (0x3f << 3)) < oldcount ||
SHA256_Pad_Almost(&hctx.ictx, u.tmp8, tmp32))
goto generic; /* Can't happen due to saltlen check */
ivecp = hctx.ictx.buf + (oldcount >> 3);
/* Prepare octx padding. */
hctx.octx.count += 32 << 3;
SHA256_Pad_Almost(&hctx.octx, u.tmp8, tmp32);
/* Iterate through the blocks. */
for (i = 0; i * 32 < dkLen; i++) {
/* Generate INT(i + 1). */
be32enc(ivecp, (uint32_t)(i + 1));
/* Compute U_1 = PRF(P, S || INT(i)). */
memcpy(u.state, hctx.ictx.state, sizeof(u.state));
SHA256_Transform(u.state, hctx.ictx.buf,
&tmp32[0], &tmp32[64]);
be32enc_vect(hctx.octx.buf, u.state, 4);
memcpy(u.state, hctx.octx.state, sizeof(u.state));
SHA256_Transform(u.state, hctx.octx.buf,
&tmp32[0], &tmp32[64]);
be32enc_vect(&buf[i * 32], u.state, 4);
}
goto cleanup;
}
generic:
/* Compute HMAC state after processing P. */
_HMAC_SHA256_Init(&Phctx, passwd, passwdlen,
tmp32, &u.tmp8[0], &u.tmp8[64]);
/* Compute HMAC state after processing P and S. */
memcpy(&PShctx, &Phctx, sizeof(HMAC_SHA256_CTX));
_HMAC_SHA256_Update(&PShctx, salt, saltlen, tmp32);
/* Iterate through the blocks. */
for (i = 0; i * 32 < dkLen; i++) {
/* Generate INT(i + 1). */
be32enc(ivec, (uint32_t)(i + 1));
/* Compute U_1 = PRF(P, S || INT(i)). */
memcpy(&hctx, &PShctx, sizeof(HMAC_SHA256_CTX));
_HMAC_SHA256_Update(&hctx, ivec, 4, tmp32);
_HMAC_SHA256_Final(T, &hctx, tmp32, u.tmp8);
if (c > 1) {
/* T_i = U_1 ... */
memcpy(U, T, 32);
for (j = 2; j <= c; j++) {
/* Compute U_j. */
memcpy(&hctx, &Phctx, sizeof(HMAC_SHA256_CTX));
_HMAC_SHA256_Update(&hctx, U, 32, tmp32);
_HMAC_SHA256_Final(U, &hctx, tmp32, u.tmp8);
/* ... xor U_j ... */
for (k = 0; k < 32; k++)
T[k] ^= U[k];
}
}
/* Copy as many bytes as necessary into buf. */
clen = dkLen - i * 32;
if (clen > 32)
clen = 32;
memcpy(&buf[i * 32], T, clen);
}
/* Clean the stack. */
insecure_memzero(&Phctx, sizeof(HMAC_SHA256_CTX));
insecure_memzero(&PShctx, sizeof(HMAC_SHA256_CTX));
insecure_memzero(U, 32);
insecure_memzero(T, 32);
cleanup:
insecure_memzero(&hctx, sizeof(HMAC_SHA256_CTX));
insecure_memzero(tmp32, 288);
insecure_memzero(&u, sizeof(u));
}

129
src/crypto/randomx/panthera/sha256.h Normal file
View file

@ -0,0 +1,129 @@
/*-
* Copyright 2005-2016 Colin Percival
* All rights reserved.
*
* 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.
*/
#ifndef _SHA256_H_
#define _SHA256_H_
#include <stddef.h>
#include <stdint.h>
#ifdef __cplusplus
extern "C" {
#endif
/*
* Use #defines in order to avoid namespace collisions with anyone else's
* SHA256 code (e.g., the code in OpenSSL).
*/
#define SHA256_Init libcperciva_SHA256_Init
#define SHA256_Update libcperciva_SHA256_Update
#define SHA256_Final libcperciva_SHA256_Final
#define SHA256_Buf libcperciva_SHA256_Buf
#define SHA256_CTX libcperciva_SHA256_CTX
#define HMAC_SHA256_Init libcperciva_HMAC_SHA256_Init
#define HMAC_SHA256_Update libcperciva_HMAC_SHA256_Update
#define HMAC_SHA256_Final libcperciva_HMAC_SHA256_Final
#define HMAC_SHA256_Buf libcperciva_HMAC_SHA256_Buf
#define HMAC_SHA256_CTX libcperciva_HMAC_SHA256_CTX
/* Context structure for SHA256 operations. */
typedef struct {
uint32_t state[8];
uint64_t count;
uint8_t buf[64];
} SHA256_CTX;
/**
* SHA256_Init(ctx):
* Initialize the SHA256 context ${ctx}.
*/
void SHA256_Init(SHA256_CTX *);
/**
* SHA256_Update(ctx, in, len):
* Input ${len} bytes from ${in} into the SHA256 context ${ctx}.
*/
void SHA256_Update(SHA256_CTX *, const void *, size_t);
/**
* SHA256_Final(digest, ctx):
* Output the SHA256 hash of the data input to the context ${ctx} into the
* buffer ${digest}.
*/
void SHA256_Final(uint8_t[32], SHA256_CTX *);
/**
* SHA256_Buf(in, len, digest):
* Compute the SHA256 hash of ${len} bytes from ${in} and write it to ${digest}.
*/
void SHA256_Buf(const void *, size_t, uint8_t[32]);
/* Context structure for HMAC-SHA256 operations. */
typedef struct {
SHA256_CTX ictx;
SHA256_CTX octx;
} HMAC_SHA256_CTX;
/**
* HMAC_SHA256_Init(ctx, K, Klen):
* Initialize the HMAC-SHA256 context ${ctx} with ${Klen} bytes of key from
* ${K}.
*/
void HMAC_SHA256_Init(HMAC_SHA256_CTX *, const void *, size_t);
/**
* HMAC_SHA256_Update(ctx, in, len):
* Input ${len} bytes from ${in} into the HMAC-SHA256 context ${ctx}.
*/
void HMAC_SHA256_Update(HMAC_SHA256_CTX *, const void *, size_t);
/**
* HMAC_SHA256_Final(digest, ctx):
* Output the HMAC-SHA256 of the data input to the context ${ctx} into the
* buffer ${digest}.
*/
void HMAC_SHA256_Final(uint8_t[32], HMAC_SHA256_CTX *);
/**
* HMAC_SHA256_Buf(K, Klen, in, len, digest):
* Compute the HMAC-SHA256 of ${len} bytes from ${in} using the key ${K} of
* length ${Klen}, and write the result to ${digest}.
*/
void HMAC_SHA256_Buf(const void *, size_t, const void *, size_t, uint8_t[32]);
/**
* PBKDF2_SHA256(passwd, passwdlen, salt, saltlen, c, buf, dkLen):
* Compute PBKDF2(passwd, salt, c, dkLen) using HMAC-SHA256 as the PRF, and
* write the output to buf. The value dkLen must be at most 32 * (2^32 - 1).
*/
void PBKDF2_SHA256(const uint8_t *, size_t, const uint8_t *, size_t,
uint64_t, uint8_t *, size_t);
#ifdef __cplusplus
}
#endif
#endif /* !_SHA256_H_ */

94
src/crypto/randomx/panthera/sysendian.h Normal file
View file

@ -0,0 +1,94 @@
/*-
* Copyright 2007-2014 Colin Percival
* All rights reserved.
*
* 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.
*/
#ifndef _SYSENDIAN_H_
#define _SYSENDIAN_H_
#include <stdint.h>
/* Avoid namespace collisions with BSD <sys/endian.h>. */
#define be32dec libcperciva_be32dec
#define be32enc libcperciva_be32enc
#define be64enc libcperciva_be64enc
#define le32dec libcperciva_le32dec
#define le32enc libcperciva_le32enc
static inline uint32_t
be32dec(const void * pp)
{
const uint8_t * p = (uint8_t const *)pp;
return ((uint32_t)(p[3]) + ((uint32_t)(p[2]) << 8) +
((uint32_t)(p[1]) << 16) + ((uint32_t)(p[0]) << 24));
}
static inline void
be32enc(void * pp, uint32_t x)
{
uint8_t * p = (uint8_t *)pp;
p[3] = x & 0xff;
p[2] = (x >> 8) & 0xff;
p[1] = (x >> 16) & 0xff;
p[0] = (x >> 24) & 0xff;
}
static inline void
be64enc(void * pp, uint64_t x)
{
uint8_t * p = (uint8_t *)pp;
p[7] = x & 0xff;
p[6] = (x >> 8) & 0xff;
p[5] = (x >> 16) & 0xff;
p[4] = (x >> 24) & 0xff;
p[3] = (x >> 32) & 0xff;
p[2] = (x >> 40) & 0xff;
p[1] = (x >> 48) & 0xff;
p[0] = (x >> 56) & 0xff;
}
static inline uint32_t
le32dec(const void * pp)
{
const uint8_t * p = (uint8_t const *)pp;
return ((uint32_t)(p[0]) + ((uint32_t)(p[1]) << 8) +
((uint32_t)(p[2]) << 16) + ((uint32_t)(p[3]) << 24));
}
static inline void
le32enc(void * pp, uint32_t x)
{
uint8_t * p = (uint8_t *)pp;
p[0] = x & 0xff;
p[1] = (x >> 8) & 0xff;
p[2] = (x >> 16) & 0xff;
p[3] = (x >> 24) & 0xff;
}
#endif /* !_SYSENDIAN_H_ */

1153
src/crypto/randomx/panthera/yespower-opt.c Normal file
View file

File diff suppressed because it is too large Load diff

107
src/crypto/randomx/panthera/yespower-platform.c Normal file
View file

@ -0,0 +1,107 @@
/*-
* Copyright 2013-2018 Alexander Peslyak
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted.
*
* 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.
*/
#ifdef __unix__
#include <sys/mman.h>
#endif
#define HUGEPAGE_THRESHOLD (12 * 1024 * 1024)
#ifdef __x86_64__
#define HUGEPAGE_SIZE (2 * 1024 * 1024)
#else
#undef HUGEPAGE_SIZE
#endif
static void *alloc_region(yespower_region_t *region, size_t size)
{
size_t base_size = size;
uint8_t *base, *aligned;
#ifdef MAP_ANON
int flags =
#ifdef MAP_NOCORE
MAP_NOCORE |
#endif
MAP_ANON | MAP_PRIVATE;
#if defined(MAP_HUGETLB) && defined(HUGEPAGE_SIZE)
size_t new_size = size;
const size_t hugepage_mask = (size_t)HUGEPAGE_SIZE - 1;
if (size >= HUGEPAGE_THRESHOLD && size + hugepage_mask >= size) {
flags |= MAP_HUGETLB;
/*
* Linux's munmap() fails on MAP_HUGETLB mappings if size is not a multiple of
* huge page size, so let's round up to huge page size here.
*/
new_size = size + hugepage_mask;
new_size &= ~hugepage_mask;
}
base = mmap(NULL, new_size, PROT_READ | PROT_WRITE, flags, -1, 0);
if (base != MAP_FAILED) {
base_size = new_size;
} else if (flags & MAP_HUGETLB) {
flags &= ~MAP_HUGETLB;
base = mmap(NULL, size, PROT_READ | PROT_WRITE, flags, -1, 0);
}
#else
base = mmap(NULL, size, PROT_READ | PROT_WRITE, flags, -1, 0);
#endif
if (base == MAP_FAILED)
base = NULL;
aligned = base;
#elif defined(HAVE_POSIX_MEMALIGN)
if ((errno = posix_memalign((void **)&base, 64, size)) != 0)
base = NULL;
aligned = base;
#else
base = aligned = NULL;
if (size + 63 < size) {
errno = ENOMEM;
} else if ((base = malloc(size + 63)) != NULL) {
aligned = base + 63;
aligned -= (uintptr_t)aligned & 63;
}
#endif
region->base = base;
region->aligned = aligned;
region->base_size = base ? base_size : 0;
region->aligned_size = base ? size : 0;
return aligned;
}
static inline void init_region(yespower_region_t *region)
{
region->base = region->aligned = NULL;
region->base_size = region->aligned_size = 0;
}
static int free_region(yespower_region_t *region)
{
if (region->base) {
#ifdef MAP_ANON
if (munmap(region->base, region->base_size))
return -1;
#else
free(region->base);
#endif
}
init_region(region);
return 0;
}

582
src/crypto/randomx/panthera/yespower-ref.c Normal file
View file

@ -0,0 +1,582 @@
/*-
* Copyright 2009 Colin Percival
* Copyright 2013-2019 Alexander Peslyak
* All rights reserved.
*
* 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.
*
* This file was originally written by Colin Percival as part of the Tarsnap
* online backup system.
*
* This is a proof-of-work focused fork of yescrypt, including reference and
* cut-down implementation of the obsolete yescrypt 0.5 (based off its first
* submission to PHC back in 2014) and a new proof-of-work specific variation
* known as yespower 1.0. The former is intended as an upgrade for
* cryptocurrencies that already use yescrypt 0.5 and the latter may be used
* as a further upgrade (hard fork) by those and other cryptocurrencies. The
* version of algorithm to use is requested through parameters, allowing for
* both algorithms to co-exist in client and miner implementations (such as in
* preparation for a hard-fork).
*
* This is the reference implementation. Its purpose is to provide a simple
* human- and machine-readable specification that implementations intended
* for actual use should be tested against. It is deliberately mostly not
* optimized, and it is not meant to be used in production. Instead, use
* yespower-opt.c.
*/
#warning "This reference implementation is deliberately mostly not optimized. Use yespower-opt.c instead unless you're testing (against) the reference implementation on purpose."
#include <errno.h>
#include <stdint.h>
#include <stdlib.h>
#include <string.h>
#include "sha256.h"
#include "sysendian.h"
#include "yespower.h"
static void blkcpy(uint32_t *dst, const uint32_t *src, size_t count)
{
do {
*dst++ = *src++;
} while (--count);
}
static void blkxor(uint32_t *dst, const uint32_t *src, size_t count)
{
do {
*dst++ ^= *src++;
} while (--count);
}
/**
* salsa20(B):
* Apply the Salsa20 core to the provided block.
*/
static void salsa20(uint32_t B[16], uint32_t rounds)
{
uint32_t x[16];
size_t i;
/* SIMD unshuffle */
for (i = 0; i < 16; i++)
x[i * 5 % 16] = B[i];
for (i = 0; i < rounds; i += 2) {
#define R(a,b) (((a) << (b)) | ((a) >> (32 - (b))))
/* Operate on columns */
x[ 4] ^= R(x[ 0]+x[12], 7); x[ 8] ^= R(x[ 4]+x[ 0], 9);
x[12] ^= R(x[ 8]+x[ 4],13); x[ 0] ^= R(x[12]+x[ 8],18);
x[ 9] ^= R(x[ 5]+x[ 1], 7); x[13] ^= R(x[ 9]+x[ 5], 9);
x[ 1] ^= R(x[13]+x[ 9],13); x[ 5] ^= R(x[ 1]+x[13],18);
x[14] ^= R(x[10]+x[ 6], 7); x[ 2] ^= R(x[14]+x[10], 9);
x[ 6] ^= R(x[ 2]+x[14],13); x[10] ^= R(x[ 6]+x[ 2],18);
x[ 3] ^= R(x[15]+x[11], 7); x[ 7] ^= R(x[ 3]+x[15], 9);
x[11] ^= R(x[ 7]+x[ 3],13); x[15] ^= R(x[11]+x[ 7],18);
/* Operate on rows */
x[ 1] ^= R(x[ 0]+x[ 3], 7); x[ 2] ^= R(x[ 1]+x[ 0], 9);
x[ 3] ^= R(x[ 2]+x[ 1],13); x[ 0] ^= R(x[ 3]+x[ 2],18);
x[ 6] ^= R(x[ 5]+x[ 4], 7); x[ 7] ^= R(x[ 6]+x[ 5], 9);
x[ 4] ^= R(x[ 7]+x[ 6],13); x[ 5] ^= R(x[ 4]+x[ 7],18);
x[11] ^= R(x[10]+x[ 9], 7); x[ 8] ^= R(x[11]+x[10], 9);
x[ 9] ^= R(x[ 8]+x[11],13); x[10] ^= R(x[ 9]+x[ 8],18);
x[12] ^= R(x[15]+x[14], 7); x[13] ^= R(x[12]+x[15], 9);
x[14] ^= R(x[13]+x[12],13); x[15] ^= R(x[14]+x[13],18);
#undef R
}
/* SIMD shuffle */
for (i = 0; i < 16; i++)
B[i] += x[i * 5 % 16];
}
/**
* blockmix_salsa(B):
* Compute B = BlockMix_{salsa20, 1}(B). The input B must be 128 bytes in
* length.
*/
static void blockmix_salsa(uint32_t *B, uint32_t rounds)
{
uint32_t X[16];
size_t i;
/* 1: X <-- B_{2r - 1} */
blkcpy(X, &B[16], 16);
/* 2: for i = 0 to 2r - 1 do */
for (i = 0; i < 2; i++) {
/* 3: X <-- H(X xor B_i) */
blkxor(X, &B[i * 16], 16);
salsa20(X, rounds);
/* 4: Y_i <-- X */
/* 6: B' <-- (Y_0, Y_2 ... Y_{2r-2}, Y_1, Y_3 ... Y_{2r-1}) */
blkcpy(&B[i * 16], X, 16);
}
}
/*
* These are tunable, but they must meet certain constraints and are part of
* what defines a yespower version.
*/
#define PWXsimple 2
#define PWXgather 4
/* Version 0.5 */
#define PWXrounds_0_5 6
#define Swidth_0_5 8
/* Version 1.0 */
#define PWXrounds_1_0 3
#define Swidth_1_0 11
/* Derived values. Not tunable on their own. */
#define PWXbytes (PWXgather * PWXsimple * 8)
#define PWXwords (PWXbytes / sizeof(uint32_t))
#define rmin ((PWXbytes + 127) / 128)
/* Runtime derived values. Not tunable on their own. */
#define Swidth_to_Sbytes1(Swidth) ((1 << Swidth) * PWXsimple * 8)
#define Swidth_to_Smask(Swidth) (((1 << Swidth) - 1) * PWXsimple * 8)
typedef struct {
yespower_version_t version;
uint32_t salsa20_rounds;
uint32_t PWXrounds, Swidth, Sbytes, Smask;
uint32_t *S;
uint32_t (*S0)[2], (*S1)[2], (*S2)[2];
size_t w;
} pwxform_ctx_t;
/**
* pwxform(B):
* Transform the provided block using the provided S-boxes.
*/
static void pwxform(uint32_t *B, pwxform_ctx_t *ctx)
{
uint32_t (*X)[PWXsimple][2] = (uint32_t (*)[PWXsimple][2])B;
uint32_t (*S0)[2] = ctx->S0, (*S1)[2] = ctx->S1, (*S2)[2] = ctx->S2;
uint32_t Smask = ctx->Smask;
size_t w = ctx->w;
size_t i, j, k;
/* 1: for i = 0 to PWXrounds - 1 do */
for (i = 0; i < ctx->PWXrounds; i++) {
/* 2: for j = 0 to PWXgather - 1 do */
for (j = 0; j < PWXgather; j++) {
uint32_t xl = X[j][0][0];
uint32_t xh = X[j][0][1];
uint32_t (*p0)[2], (*p1)[2];
/* 3: p0 <-- (lo(B_{j,0}) & Smask) / (PWXsimple * 8) */
p0 = S0 + (xl & Smask) / sizeof(*S0);
/* 4: p1 <-- (hi(B_{j,0}) & Smask) / (PWXsimple * 8) */
p1 = S1 + (xh & Smask) / sizeof(*S1);
/* 5: for k = 0 to PWXsimple - 1 do */
for (k = 0; k < PWXsimple; k++) {
uint64_t x, s0, s1;
/* 6: B_{j,k} <-- (hi(B_{j,k}) * lo(B_{j,k}) + S0_{p0,k}) xor S1_{p1,k} */
s0 = ((uint64_t)p0[k][1] << 32) + p0[k][0];
s1 = ((uint64_t)p1[k][1] << 32) + p1[k][0];
xl = X[j][k][0];
xh = X[j][k][1];
x = (uint64_t)xh * xl;
x += s0;
x ^= s1;
X[j][k][0] = x;
X[j][k][1] = x >> 32;
}
if (ctx->version != YESPOWER_0_5 &&
(i == 0 || j < PWXgather / 2)) {
if (j & 1) {
for (k = 0; k < PWXsimple; k++) {
S1[w][0] = X[j][k][0];
S1[w][1] = X[j][k][1];
w++;
}
} else {
for (k = 0; k < PWXsimple; k++) {
S0[w + k][0] = X[j][k][0];
S0[w + k][1] = X[j][k][1];
}
}
}
}
}
if (ctx->version != YESPOWER_0_5) {
/* 14: (S0, S1, S2) <-- (S2, S0, S1) */
ctx->S0 = S2;
ctx->S1 = S0;
ctx->S2 = S1;
/* 15: w <-- w mod 2^Swidth */
ctx->w = w & ((1 << ctx->Swidth) * PWXsimple - 1);
}
}
/**
* blockmix_pwxform(B, ctx, r):
* Compute B = BlockMix_pwxform{salsa20, ctx, r}(B). The input B must be
* 128r bytes in length.
*/
static void blockmix_pwxform(uint32_t *B, pwxform_ctx_t *ctx, size_t r)
{
uint32_t X[PWXwords];
size_t r1, i;
/* Convert 128-byte blocks to PWXbytes blocks */
/* 1: r_1 <-- 128r / PWXbytes */
r1 = 128 * r / PWXbytes;
/* 2: X <-- B'_{r_1 - 1} */
blkcpy(X, &B[(r1 - 1) * PWXwords], PWXwords);
/* 3: for i = 0 to r_1 - 1 do */
for (i = 0; i < r1; i++) {
/* 4: if r_1 > 1 */
if (r1 > 1) {
/* 5: X <-- X xor B'_i */
blkxor(X, &B[i * PWXwords], PWXwords);
}
/* 7: X <-- pwxform(X) */
pwxform(X, ctx);
/* 8: B'_i <-- X */
blkcpy(&B[i * PWXwords], X, PWXwords);
}
/* 10: i <-- floor((r_1 - 1) * PWXbytes / 64) */
i = (r1 - 1) * PWXbytes / 64;
/* 11: B_i <-- H(B_i) */
salsa20(&B[i * 16], ctx->salsa20_rounds);
#if 1 /* No-op with our current pwxform settings, but do it to make sure */
/* 12: for i = i + 1 to 2r - 1 do */
for (i++; i < 2 * r; i++) {
/* 13: B_i <-- H(B_i xor B_{i-1}) */
blkxor(&B[i * 16], &B[(i - 1) * 16], 16);
salsa20(&B[i * 16], ctx->salsa20_rounds);
}
#endif
}
/**
* integerify(B, r):
* Return the result of parsing B_{2r-1} as a little-endian integer.
*/
static uint32_t integerify(const uint32_t *B, size_t r)
{
/*
* Our 32-bit words are in host byte order. Also, they are SIMD-shuffled, but
* we only care about the least significant 32 bits anyway.
*/
const uint32_t *X = &B[(2 * r - 1) * 16];
return X[0];
}
/**
* p2floor(x):
* Largest power of 2 not greater than argument.
*/
static uint32_t p2floor(uint32_t x)
{
uint32_t y;
while ((y = x & (x - 1)))
x = y;
return x;
}
/**
* wrap(x, i):
* Wrap x to the range 0 to i-1.
*/
static uint32_t wrap(uint32_t x, uint32_t i)
{
uint32_t n = p2floor(i);
return (x & (n - 1)) + (i - n);
}
/**
* smix1(B, r, N, V, X, ctx):
* Compute first loop of B = SMix_r(B, N). The input B must be 128r bytes in
* length; the temporary storage V must be 128rN bytes in length; the temporary
* storage X must be 128r bytes in length.
*/
static void smix1(uint32_t *B, size_t r, uint32_t N,
uint32_t *V, uint32_t *X, pwxform_ctx_t *ctx)
{
size_t s = 32 * r;
uint32_t i, j;
size_t k;
/* 1: X <-- B */
for (k = 0; k < 2 * r; k++)
for (i = 0; i < 16; i++)
X[k * 16 + i] = le32dec(&B[k * 16 + (i * 5 % 16)]);
if (ctx->version != YESPOWER_0_5) {
for (k = 1; k < r; k++) {
blkcpy(&X[k * 32], &X[(k - 1) * 32], 32);
blockmix_pwxform(&X[k * 32], ctx, 1);
}
}
/* 2: for i = 0 to N - 1 do */
for (i = 0; i < N; i++) {
/* 3: V_i <-- X */
blkcpy(&V[i * s], X, s);
if (i > 1) {
/* j <-- Wrap(Integerify(X), i) */
j = wrap(integerify(X, r), i);
/* X <-- X xor V_j */
blkxor(X, &V[j * s], s);
}
/* 4: X <-- H(X) */
if (V != ctx->S)
blockmix_pwxform(X, ctx, r);
else
blockmix_salsa(X, ctx->salsa20_rounds);
}
/* B' <-- X */
for (k = 0; k < 2 * r; k++)
for (i = 0; i < 16; i++)
le32enc(&B[k * 16 + (i * 5 % 16)], X[k * 16 + i]);
}
/**
* smix2(B, r, N, Nloop, V, X, ctx):
* Compute second loop of B = SMix_r(B, N). The input B must be 128r bytes in
* length; the temporary storage V must be 128rN bytes in length; the temporary
* storage X must be 128r bytes in length. The value N must be a power of 2
* greater than 1.
*/
static void smix2(uint32_t *B, size_t r, uint32_t N, uint32_t Nloop,
uint32_t *V, uint32_t *X, pwxform_ctx_t *ctx)
{
size_t s = 32 * r;
uint32_t i, j;
size_t k;
/* X <-- B */
for (k = 0; k < 2 * r; k++)
for (i = 0; i < 16; i++)
X[k * 16 + i] = le32dec(&B[k * 16 + (i * 5 % 16)]);
/* 6: for i = 0 to N - 1 do */
for (i = 0; i < Nloop; i++) {
/* 7: j <-- Integerify(X) mod N */
j = integerify(X, r) & (N - 1);
/* 8.1: X <-- X xor V_j */
blkxor(X, &V[j * s], s);
/* V_j <-- X */
if (Nloop != 2)
blkcpy(&V[j * s], X, s);
/* 8.2: X <-- H(X) */
blockmix_pwxform(X, ctx, r);
}
/* 10: B' <-- X */
for (k = 0; k < 2 * r; k++)
for (i = 0; i < 16; i++)
le32enc(&B[k * 16 + (i * 5 % 16)], X[k * 16 + i]);
}
/**
* smix(B, r, N, p, t, V, X, ctx):
* Compute B = SMix_r(B, N). The input B must be 128rp bytes in length; the
* temporary storage V must be 128rN bytes in length; the temporary storage
* X must be 128r bytes in length. The value N must be a power of 2 and at
* least 16.
*/
static void smix(uint32_t *B, size_t r, uint32_t N,
uint32_t *V, uint32_t *X, pwxform_ctx_t *ctx)
{
uint32_t Nloop_all = (N + 2) / 3; /* 1/3, round up */
uint32_t Nloop_rw = Nloop_all;
Nloop_all++; Nloop_all &= ~(uint32_t)1; /* round up to even */
if (ctx->version == YESPOWER_0_5) {
Nloop_rw &= ~(uint32_t)1; /* round down to even */
} else {
Nloop_rw++; Nloop_rw &= ~(uint32_t)1; /* round up to even */
}
smix1(B, 1, ctx->Sbytes / 128, ctx->S, X, ctx);
smix1(B, r, N, V, X, ctx);
smix2(B, r, N, Nloop_rw /* must be > 2 */, V, X, ctx);
smix2(B, r, N, Nloop_all - Nloop_rw /* 0 or 2 */, V, X, ctx);
}
/**
* yespower(local, src, srclen, params, dst):
* Compute yespower(src[0 .. srclen - 1], N, r), to be checked for "< target".
*
* Return 0 on success; or -1 on error.
*/
int yespower(yespower_local_t *local,
const uint8_t *src, size_t srclen,
const yespower_params_t *params, yespower_binary_t *dst)
{
yespower_version_t version = params->version;
uint32_t N = params->N;
uint32_t r = params->r;
const uint8_t *pers = params->pers;
size_t perslen = params->perslen;
int retval = -1;
size_t B_size, V_size;
uint32_t *B, *V, *X, *S;
pwxform_ctx_t ctx;
uint32_t sha256[8];
memset(dst, 0xff, sizeof(*dst));
/* Sanity-check parameters */
if ((version != YESPOWER_0_5 && version != YESPOWER_1_0) ||
N < 1024 || N > 512 * 1024 || r < 8 || r > 32 ||
(N & (N - 1)) != 0 || r < rmin ||
(!pers && perslen)) {
errno = EINVAL;
return -1;
}
/* Allocate memory */
B_size = (size_t)128 * r;
V_size = B_size * N;
if ((V = malloc(V_size)) == NULL)
return -1;
if ((B = malloc(B_size)) == NULL)
goto free_V;
if ((X = malloc(B_size)) == NULL)
goto free_B;
ctx.version = version;
if (version == YESPOWER_0_5) {
ctx.salsa20_rounds = 8;
ctx.PWXrounds = PWXrounds_0_5;
ctx.Swidth = Swidth_0_5;
ctx.Sbytes = 2 * Swidth_to_Sbytes1(ctx.Swidth);
} else {
ctx.salsa20_rounds = 2;
ctx.PWXrounds = PWXrounds_1_0;
ctx.Swidth = Swidth_1_0;
ctx.Sbytes = 3 * Swidth_to_Sbytes1(ctx.Swidth);
}
if ((S = malloc(ctx.Sbytes)) == NULL)
goto free_X;
ctx.S = S;
ctx.S0 = (uint32_t (*)[2])S;
ctx.S1 = ctx.S0 + (1 << ctx.Swidth) * PWXsimple;
ctx.S2 = ctx.S1 + (1 << ctx.Swidth) * PWXsimple;
ctx.Smask = Swidth_to_Smask(ctx.Swidth);
ctx.w = 0;
SHA256_Buf(src, srclen, (uint8_t *)sha256);
if (version != YESPOWER_0_5) {
if (pers) {
src = pers;
srclen = perslen;
} else {
srclen = 0;
}
}
/* 1: (B_0 ... B_{p-1}) <-- PBKDF2(P, S, 1, p * MFLen) */
PBKDF2_SHA256((uint8_t *)sha256, sizeof(sha256),
src, srclen, 1, (uint8_t *)B, B_size);
blkcpy(sha256, B, sizeof(sha256) / sizeof(sha256[0]));
/* 3: B_i <-- MF(B_i, N) */
smix(B, r, N, V, X, &ctx);
if (version == YESPOWER_0_5) {
/* 5: DK <-- PBKDF2(P, B, 1, dkLen) */
PBKDF2_SHA256((uint8_t *)sha256, sizeof(sha256),
(uint8_t *)B, B_size, 1, (uint8_t *)dst, sizeof(*dst));
if (pers) {
HMAC_SHA256_Buf(dst, sizeof(*dst), pers, perslen,
(uint8_t *)sha256);
SHA256_Buf(sha256, sizeof(sha256), (uint8_t *)dst);
}
} else {
HMAC_SHA256_Buf((uint8_t *)B + B_size - 64, 64,
sha256, sizeof(sha256), (uint8_t *)dst);
}
/* Success! */
retval = 0;
/* Free memory */
free(S);
free_X:
free(X);
free_B:
free(B);
free_V:
free(V);
return retval;
}
int yespower_tls(const uint8_t *src, size_t srclen,
const yespower_params_t *params, yespower_binary_t *dst)
{
/* The reference implementation doesn't use thread-local storage */
return yespower(NULL, src, srclen, params, dst);
}
int yespower_init_local(yespower_local_t *local)
{
/* The reference implementation doesn't use the local structure */
local->base = local->aligned = NULL;
local->base_size = local->aligned_size = 0;
return 0;
}
int yespower_free_local(yespower_local_t *local)
{
/* The reference implementation frees its memory in yespower() */
(void)local; /* unused */
return 0;
}

130
src/crypto/randomx/panthera/yespower.h Normal file
View file

@ -0,0 +1,130 @@
/*-
* Copyright 2009 Colin Percival
* Copyright 2013-2018 Alexander Peslyak
* All rights reserved.
*
* 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.
*
* This file was originally written by Colin Percival as part of the Tarsnap
* online backup system.
*/
#ifndef _YESPOWER_H_
#define _YESPOWER_H_
#include <stdint.h>
#include <stdlib.h> /* for size_t */
#ifdef __cplusplus
extern "C" {
#endif
/**
* Internal type used by the memory allocator. Please do not use it directly.
* Use yespower_local_t instead.
*/
typedef struct {
void *base, *aligned;
size_t base_size, aligned_size;
} yespower_region_t;
/**
* Type for thread-local (RAM) data structure.
*/
typedef yespower_region_t yespower_local_t;
/*
* Type for yespower algorithm version numbers.
*/
typedef enum { YESPOWER_0_5 = 5, YESPOWER_1_0 = 10 } yespower_version_t;
/**
* yespower parameters combined into one struct.
*/
typedef struct {
yespower_version_t version;
uint32_t N, r;
const uint8_t *pers;
size_t perslen;
} yespower_params_t;
/**
* A 256-bit yespower hash.
*/
typedef struct {
unsigned char uc[32];
} yespower_binary_t;
/**
* yespower_init_local(local):
* Initialize the thread-local (RAM) data structure. Actual memory allocation
* is currently fully postponed until a call to yespower().
*
* Return 0 on success; or -1 on error.
*
* MT-safe as long as local is local to the thread.
*/
extern int yespower_init_local(yespower_local_t *local);
/**
* yespower_free_local(local):
* Free memory that may have been allocated for an initialized thread-local
* (RAM) data structure.
*
* Return 0 on success; or -1 on error.
*
* MT-safe as long as local is local to the thread.
*/
extern int yespower_free_local(yespower_local_t *local);
/**
* yespower(local, src, srclen, params, dst):
* Compute yespower(src[0 .. srclen - 1], N, r), to be checked for "< target".
* local is the thread-local data structure, allowing to preserve and reuse a
* memory allocation across calls, thereby reducing processing overhead.
*
* Return 0 on success; or -1 on error.
*
* local must be initialized with yespower_init_local().
*
* MT-safe as long as local and dst are local to the thread.
*/
extern int yespower(yespower_local_t *local,
const uint8_t *src, size_t srclen,
const yespower_params_t *params, yespower_binary_t *dst);
/**
* yespower_tls(src, srclen, params, dst):
* Compute yespower(src[0 .. srclen - 1], N, r), to be checked for "< target".
* The memory allocation is maintained internally using thread-local storage.
*
* Return 0 on success; or -1 on error.
*
* MT-safe as long as dst is local to the thread.
*/
extern int yespower_tls(const uint8_t *src, size_t srclen,
const yespower_params_t *params, yespower_binary_t *dst);
#ifdef __cplusplus
}
#endif
#endif /* !_YESPOWER_H_ */

37
src/crypto/randomx/randomx.cpp
View file

@ -49,6 +49,7 @@ OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
extern "C" {
#include "crypto/randomx/defyx/yescrypt.h"
#include "crypto/randomx/panthera/yespower.h"
#include "crypto/randomx/defyx/KangarooTwelve.h"
}
@ -131,6 +132,10 @@ RandomX_ConfigurationScala::RandomX_ConfigurationScala()
RANDOMX_FREQ_CBRANCH = 16;
}
RandomX_ConfigurationScala2::RandomX_ConfigurationScala2()
{
}
RandomX_ConfigurationBase::RandomX_ConfigurationBase()
: ArgonMemory(262144)
, ArgonIterations(3)
@ -335,6 +340,7 @@ RandomX_ConfigurationArqma RandomX_ArqmaConfig;
RandomX_ConfigurationSafex RandomX_SafexConfig;
RandomX_ConfigurationKeva RandomX_KevaConfig;
RandomX_ConfigurationScala RandomX_ScalaConfig;
RandomX_ConfigurationScala2 RandomX_Scala2Config;
alignas(64) RandomX_ConfigurationBase RandomX_CurrentConfig;
@ -359,6 +365,13 @@ int rx_sipesh_k12(void *out, size_t outlen, const void *in, size_t inlen)
return retval;
}
int rx_yespower_k12(void *out, size_t outlen, const void *in, size_t inlen)
{
rx_blake2b(out, outlen, in, inlen, 0, 0);
yespower_hash(out, outlen, out);
return KangarooTwelve((const unsigned char *)in, inlen, (unsigned char *)out, 32, 0, 0);
}
extern "C" {
randomx_cache *randomx_create_cache(randomx_flags flags, uint8_t *memory) {
@ -564,10 +577,10 @@ extern "C" {
assert(inputSize == 0 || input != nullptr);
assert(output != nullptr);
alignas(16) uint64_t tempHash[8];
if (algo == xmrig::Algorithm::RX_DEFYX) {
rx_sipesh_k12(tempHash, sizeof(tempHash), input, inputSize);
} else {
rx_blake2b(tempHash, sizeof(tempHash), input, inputSize, nullptr, 0);
switch (algo) {
case xmrig::Algorithm::RX_DEFYX: rx_sipesh_k12(tempHash, sizeof(tempHash), input, inputSize); break;
case xmrig::Algorithm::RX_XLA: rx_yespower_k12(tempHash, sizeof(tempHash), input, inputSize); break;
default: rx_blake2b(tempHash, sizeof(tempHash), input, inputSize, nullptr, 0);
}
machine->initScratchpad(&tempHash);
machine->resetRoundingMode();
@ -580,10 +593,10 @@ extern "C" {
}
void randomx_calculate_hash_first(randomx_vm* machine, uint64_t (&tempHash)[8], const void* input, size_t inputSize, const xmrig::Algorithm algo) {
if (algo == xmrig::Algorithm::RX_DEFYX) {
rx_sipesh_k12(tempHash, sizeof(tempHash), input, inputSize);
} else {
rx_blake2b(tempHash, sizeof(tempHash), input, inputSize, nullptr, 0);
switch (algo) {
case xmrig::Algorithm::RX_DEFYX: rx_sipesh_k12(tempHash, sizeof(tempHash), input, inputSize); break;
case xmrig::Algorithm::RX_XLA: rx_yespower_k12(tempHash, sizeof(tempHash), input, inputSize); break;
default: rx_blake2b(tempHash, sizeof(tempHash), input, inputSize, nullptr, 0);
}
machine->initScratchpad(tempHash);
}
@ -597,10 +610,10 @@ extern "C" {
machine->run(&tempHash);
// Finish current hash and fill the scratchpad for the next hash at the same time
if (algo == xmrig::Algorithm::RX_DEFYX) {
rx_sipesh_k12(tempHash, sizeof(tempHash), nextInput, nextInputSize);
} else {
rx_blake2b(tempHash, sizeof(tempHash), nextInput, nextInputSize, nullptr, 0);
switch (algo) {
case xmrig::Algorithm::RX_DEFYX: rx_sipesh_k12(tempHash, sizeof(tempHash), nextInput, nextInputSize); break;
case xmrig::Algorithm::RX_XLA: rx_yespower_k12(tempHash, sizeof(tempHash), nextInput, nextInputSize); break;
default: rx_blake2b(tempHash, sizeof(tempHash), nextInput, nextInputSize, nullptr, 0);
}
machine->hashAndFill(output, RANDOMX_HASH_SIZE, tempHash);
}

2
src/crypto/randomx/randomx.h
View file

@ -186,6 +186,7 @@ struct RandomX_ConfigurationArqma : public RandomX_ConfigurationBase { RandomX_C
struct RandomX_ConfigurationSafex : public RandomX_ConfigurationBase { RandomX_ConfigurationSafex(); };
struct RandomX_ConfigurationKeva : public RandomX_ConfigurationBase { RandomX_ConfigurationKeva(); };
struct RandomX_ConfigurationScala : public RandomX_ConfigurationBase { RandomX_ConfigurationScala(); };
struct RandomX_ConfigurationScala2 : public RandomX_ConfigurationScala { RandomX_ConfigurationScala2(); };
extern RandomX_ConfigurationMonero RandomX_MoneroConfig;
extern RandomX_ConfigurationWownero RandomX_WowneroConfig;
@ -194,6 +195,7 @@ extern RandomX_ConfigurationArqma RandomX_ArqmaConfig;
extern RandomX_ConfigurationSafex RandomX_SafexConfig;
extern RandomX_ConfigurationKeva RandomX_KevaConfig;
extern RandomX_ConfigurationScala RandomX_ScalaConfig;
extern RandomX_ConfigurationScala2 RandomX_Scala2Config;
extern RandomX_ConfigurationBase RandomX_CurrentConfig;

3
src/crypto/rx/RxAlgo.cpp
View file

@ -58,6 +58,9 @@ const RandomX_ConfigurationBase *xmrig::RxAlgo::base(Algorithm::Id algorithm)
case Algorithm::RX_DEFYX:
return &RandomX_ScalaConfig;
case Algorithm::RX_XLA:
return &RandomX_Scala2Config;
default:
break;
}