Argon2 bits - Not exposed in the API yet

This commit is contained in:
Frank Denis
2015-12-29 11:24:11 +01:00
parent 9abc0fdbd0
commit da927a985f
14 changed files with 2488 additions and 0 deletions
+13
View File
@@ -51,6 +51,17 @@ libsodium_la_SOURCES = \
crypto_onetimeauth/poly1305/donna/poly1305_donna32.h \
crypto_onetimeauth/poly1305/donna/poly1305_donna64.h \
crypto_onetimeauth/poly1305/donna/poly1305_donna.c \
crypto_pwhash/argon2/argon2-core.c \
crypto_pwhash/argon2/argon2-core.h \
crypto_pwhash/argon2/argon2-encoding.c \
crypto_pwhash/argon2/argon2-encoding.h \
crypto_pwhash/argon2/argon2-fill-block-ref.c \
crypto_pwhash/argon2/argon2-impl.h \
crypto_pwhash/argon2/argon2.c \
crypto_pwhash/argon2/argon2.h \
crypto_pwhash/argon2/blake2b-long.c \
crypto_pwhash/argon2/blake2b-long.h \
crypto_pwhash/argon2/blamka-round-ref.h \
crypto_pwhash/scryptsalsa208sha256/crypto_scrypt-common.c \
crypto_pwhash/scryptsalsa208sha256/crypto_scrypt.h \
crypto_pwhash/scryptsalsa208sha256/scrypt_platform.c \
@@ -224,6 +235,8 @@ libssse3_la_CPPFLAGS = $(libsodium_la_CPPFLAGS) \
@CFLAGS_SSE2@ @CFLAGS_SSSE3@
libssse3_la_SOURCES = \
crypto_generichash/blake2/ref/blake2b-compress-ssse3.c \
crypto_pwhash/argon2/argon2-fill-block-ssse3.c \
crypto_pwhash/argon2/blamka-round-ssse3.h \
crypto_stream/chacha20/vec/stream_chacha20_vec.h \
crypto_stream/chacha20/vec/stream_chacha20_vec.c
@@ -0,0 +1,506 @@
/*
* Argon2 source code package
*
* Written by Daniel Dinu and Dmitry Khovratovich, 2015
*
* This work is licensed under a Creative Commons CC0 1.0 License/Waiver.
*
* You should have received a copy of the CC0 Public Domain Dedication along
* with
* this software. If not, see
* <http://creativecommons.org/publicdomain/zero/1.0/>.
*/
#include <inttypes.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include "crypto_generichash_blake2b.h"
#include "runtime.h"
#include "utils.h"
#include "argon2-core.h"
#include "argon2-impl.h"
#include "blake2b-long.h"
static fill_segment_fn fill_segment = fill_segment_ref;
/***************Instance and Position constructors**********/
void init_block_value(block *b, uint8_t in) {
memset(b->v, in, sizeof(b->v));
}
void copy_block(block *dst, const block *src) {
memcpy(dst->v, src->v, sizeof(uint64_t) * ARGON2_QWORDS_IN_BLOCK);
}
void xor_block(block *dst, const block *src) {
int i;
for (i = 0; i < ARGON2_QWORDS_IN_BLOCK; ++i) {
dst->v[i] ^= src->v[i];
}
}
static void load_block(block *dst, const void *input) {
unsigned i;
for (i = 0; i < ARGON2_QWORDS_IN_BLOCK; ++i) {
dst->v[i] = load64((const uint8_t *)input + i * sizeof(dst->v[i]));
}
}
static void store_block(void *output, const block *src) {
unsigned i;
for (i = 0; i < ARGON2_QWORDS_IN_BLOCK; ++i) {
store64((uint8_t *)output + i * sizeof(src->v[i]), src->v[i]);
}
}
/***************Memory allocators*****************/
int allocate_memory(block **memory, uint32_t m_cost) {
if (memory != NULL) {
size_t memory_size = sizeof(block) * m_cost;
if (m_cost == 0 ||
memory_size / m_cost !=
sizeof(block)) { /*1. Check for multiplication overflow*/
return ARGON2_MEMORY_ALLOCATION_ERROR;
}
*memory = (block *)malloc(memory_size); /*2. Try to allocate*/
if (!*memory) {
return ARGON2_MEMORY_ALLOCATION_ERROR;
}
return ARGON2_OK;
} else {
return ARGON2_MEMORY_ALLOCATION_ERROR;
}
}
/*********Memory functions*/
void clear_memory(argon2_instance_t *instance, int clear) {
if (instance->memory != NULL && clear) {
sodium_memzero(instance->memory,
sizeof(block) * instance->memory_blocks);
}
}
void free_memory(block *memory) { free(memory); }
void finalize(const argon2_context *context, argon2_instance_t *instance) {
if (context != NULL && instance != NULL) {
block blockhash;
uint32_t l;
copy_block(&blockhash, instance->memory + instance->lane_length - 1);
/* XOR the last blocks */
for (l = 1; l < instance->lanes; ++l) {
uint32_t last_block_in_lane =
l * instance->lane_length + (instance->lane_length - 1);
xor_block(&blockhash, instance->memory + last_block_in_lane);
}
/* Hash the result */
{
uint8_t blockhash_bytes[ARGON2_BLOCK_SIZE];
store_block(blockhash_bytes, &blockhash);
blake2b_long(context->out, context->outlen, blockhash_bytes,
ARGON2_BLOCK_SIZE);
sodium_memzero(blockhash.v,
ARGON2_BLOCK_SIZE); /* clear blockhash */
sodium_memzero(blockhash_bytes,
ARGON2_BLOCK_SIZE); /* clear blockhash_bytes */
}
/* Clear memory */
clear_memory(instance, context->flags & ARGON2_FLAG_CLEAR_PASSWORD);
/* Deallocate the memory */
if (NULL != context->free_cbk) {
context->free_cbk((uint8_t *)instance->memory,
instance->memory_blocks * sizeof(block));
} else {
free_memory(instance->memory);
}
}
}
uint32_t index_alpha(const argon2_instance_t *instance,
const argon2_position_t *position, uint32_t pseudo_rand,
int same_lane) {
/*
* Pass 0:
* This lane : all already finished segments plus already constructed
* blocks in this segment
* Other lanes : all already finished segments
* Pass 1+:
* This lane : (SYNC_POINTS - 1) last segments plus already constructed
* blocks in this segment
* Other lanes : (SYNC_POINTS - 1) last segments
*/
uint32_t reference_area_size;
uint64_t relative_position;
uint32_t start_position, absolute_position;
if (0 == position->pass) {
/* First pass */
if (0 == position->slice) {
/* First slice */
reference_area_size =
position->index - 1; /* all but the previous */
} else {
if (same_lane) {
/* The same lane => add current segment */
reference_area_size =
position->slice * instance->segment_length +
position->index - 1;
} else {
reference_area_size =
position->slice * instance->segment_length +
((position->index == 0) ? (-1) : 0);
}
}
} else {
/* Second pass */
if (same_lane) {
reference_area_size = instance->lane_length -
instance->segment_length + position->index -
1;
} else {
reference_area_size = instance->lane_length -
instance->segment_length +
((position->index == 0) ? (-1) : 0);
}
}
/* 1.2.4. Mapping pseudo_rand to 0..<reference_area_size-1> and produce
* relative position */
relative_position = pseudo_rand;
relative_position = relative_position * relative_position >> 32;
relative_position = reference_area_size - 1 -
(reference_area_size * relative_position >> 32);
/* 1.2.5 Computing starting position */
start_position = 0;
if (0 != position->pass) {
start_position = (position->slice == ARGON2_SYNC_POINTS - 1)
? 0
: (position->slice + 1) * instance->segment_length;
}
/* 1.2.6. Computing absolute position */
absolute_position = (start_position + relative_position) %
instance->lane_length; /* absolute position */
return absolute_position;
}
void fill_memory_blocks(argon2_instance_t *instance) {
uint32_t r, s;
if (instance == NULL || instance->lanes == 0) {
return;
}
for (r = 0; r < instance->passes; ++r) {
for (s = 0; s < ARGON2_SYNC_POINTS; ++s) {
uint32_t l;
for (l = 0; l < instance->lanes; ++l) {
argon2_position_t position;
position.pass = r;
position.lane = l;
position.slice = (uint8_t)s;
position.index = 0;
fill_segment(instance, position);
}
}
}
}
int validate_inputs(const argon2_context *context) {
if (NULL == context) {
return ARGON2_INCORRECT_PARAMETER;
}
if (NULL == context->out) {
return ARGON2_OUTPUT_PTR_NULL;
}
/* Validate output length */
if (ARGON2_MIN_OUTLEN > context->outlen) {
return ARGON2_OUTPUT_TOO_SHORT;
}
if (ARGON2_MAX_OUTLEN < context->outlen) {
return ARGON2_OUTPUT_TOO_LONG;
}
/* Validate password length */
if (NULL == context->pwd) {
if (0 != context->pwdlen) {
return ARGON2_PWD_PTR_MISMATCH;
}
} else {
if (ARGON2_MIN_PWD_LENGTH > context->pwdlen) {
return ARGON2_PWD_TOO_SHORT;
}
if (ARGON2_MAX_PWD_LENGTH < context->pwdlen) {
return ARGON2_PWD_TOO_LONG;
}
}
/* Validate salt length */
if (NULL == context->salt) {
if (0 != context->saltlen) {
return ARGON2_SALT_PTR_MISMATCH;
}
} else {
if (ARGON2_MIN_SALT_LENGTH > context->saltlen) {
return ARGON2_SALT_TOO_SHORT;
}
if (ARGON2_MAX_SALT_LENGTH < context->saltlen) {
return ARGON2_SALT_TOO_LONG;
}
}
/* Validate secret length */
if (NULL == context->secret) {
if (0 != context->secretlen) {
return ARGON2_SECRET_PTR_MISMATCH;
}
} else {
if (ARGON2_MIN_SECRET > context->secretlen) {
return ARGON2_SECRET_TOO_SHORT;
}
if (ARGON2_MAX_SECRET < context->secretlen) {
return ARGON2_SECRET_TOO_LONG;
}
}
/* Validate associated data */
if (NULL == context->ad) {
if (0 != context->adlen) {
return ARGON2_AD_PTR_MISMATCH;
}
} else {
if (ARGON2_MIN_AD_LENGTH > context->adlen) {
return ARGON2_AD_TOO_SHORT;
}
if (ARGON2_MAX_AD_LENGTH < context->adlen) {
return ARGON2_AD_TOO_LONG;
}
}
/* Validate memory cost */
if (ARGON2_MIN_MEMORY > context->m_cost) {
return ARGON2_MEMORY_TOO_LITTLE;
}
if (ARGON2_MAX_MEMORY < context->m_cost) {
return ARGON2_MEMORY_TOO_MUCH;
}
if (context->m_cost < 8*context->lanes) {
return ARGON2_MEMORY_TOO_LITTLE;
}
/* Validate time cost */
if (ARGON2_MIN_TIME > context->t_cost) {
return ARGON2_TIME_TOO_SMALL;
}
if (ARGON2_MAX_TIME < context->t_cost) {
return ARGON2_TIME_TOO_LARGE;
}
/* Validate lanes */
if (ARGON2_MIN_LANES > context->lanes) {
return ARGON2_LANES_TOO_FEW;
}
if (ARGON2_MAX_LANES < context->lanes) {
return ARGON2_LANES_TOO_MANY;
}
/* Validate threads */
if (ARGON2_MIN_THREADS > context->threads) {
return ARGON2_THREADS_TOO_FEW;
}
if (ARGON2_MAX_THREADS < context->threads) {
return ARGON2_THREADS_TOO_MANY;
}
if (NULL != context->allocate_cbk && NULL == context->free_cbk) {
return ARGON2_FREE_MEMORY_CBK_NULL;
}
if (NULL == context->allocate_cbk && NULL != context->free_cbk) {
return ARGON2_ALLOCATE_MEMORY_CBK_NULL;
}
return ARGON2_OK;
}
void fill_first_blocks(uint8_t *blockhash, const argon2_instance_t *instance) {
uint32_t l;
/* Make the first and second block in each lane as G(H0||i||0) or
G(H0||i||1) */
uint8_t blockhash_bytes[ARGON2_BLOCK_SIZE];
for (l = 0; l < instance->lanes; ++l) {
store32(blockhash + ARGON2_PREHASH_DIGEST_LENGTH, 0);
store32(blockhash + ARGON2_PREHASH_DIGEST_LENGTH + 4, l);
blake2b_long(blockhash_bytes, ARGON2_BLOCK_SIZE, blockhash,
ARGON2_PREHASH_SEED_LENGTH);
load_block(&instance->memory[l * instance->lane_length + 0],
blockhash_bytes);
store32(blockhash + ARGON2_PREHASH_DIGEST_LENGTH, 1);
blake2b_long(blockhash_bytes, ARGON2_BLOCK_SIZE, blockhash,
ARGON2_PREHASH_SEED_LENGTH);
load_block(&instance->memory[l * instance->lane_length + 1],
blockhash_bytes);
}
sodium_memzero(blockhash_bytes, ARGON2_BLOCK_SIZE);
}
void initial_hash(uint8_t *blockhash, argon2_context *context,
argon2_type type) {
crypto_generichash_blake2b_state BlakeHash;
uint8_t value[4U /* sizeof(uint32_t) */];
if (NULL == context || NULL == blockhash) {
return;
}
crypto_generichash_blake2b_init(&BlakeHash, NULL, 0U,
ARGON2_PREHASH_DIGEST_LENGTH);
store32(&value, context->lanes);
crypto_generichash_blake2b_update(&BlakeHash, (const uint8_t *)&value, sizeof(value));
store32(&value, context->outlen);
crypto_generichash_blake2b_update(&BlakeHash, (const uint8_t *)&value, sizeof(value));
store32(&value, context->m_cost);
crypto_generichash_blake2b_update(&BlakeHash, (const uint8_t *)&value, sizeof(value));
store32(&value, context->t_cost);
crypto_generichash_blake2b_update(&BlakeHash, (const uint8_t *)&value, sizeof(value));
store32(&value, ARGON2_VERSION_NUMBER);
crypto_generichash_blake2b_update(&BlakeHash, (const uint8_t *)&value, sizeof(value));
store32(&value, (uint32_t)type);
crypto_generichash_blake2b_update(&BlakeHash, (const uint8_t *)&value, sizeof(value));
store32(&value, context->pwdlen);
crypto_generichash_blake2b_update(&BlakeHash, (const uint8_t *)&value, sizeof(value));
if (context->pwd != NULL) {
crypto_generichash_blake2b_update(&BlakeHash, (const uint8_t *)context->pwd,
context->pwdlen);
if (context->flags & ARGON2_FLAG_CLEAR_PASSWORD) {
sodium_memzero(context->pwd, context->pwdlen);
context->pwdlen = 0;
}
}
store32(&value, context->saltlen);
crypto_generichash_blake2b_update(&BlakeHash, (const uint8_t *)&value, sizeof(value));
if (context->salt != NULL) {
crypto_generichash_blake2b_update(&BlakeHash, (const uint8_t *)context->salt,
context->saltlen);
}
store32(&value, context->secretlen);
crypto_generichash_blake2b_update(&BlakeHash, (const uint8_t *)&value, sizeof(value));
if (context->secret != NULL) {
crypto_generichash_blake2b_update(&BlakeHash, (const uint8_t *)context->secret,
context->secretlen);
if (context->flags & ARGON2_FLAG_CLEAR_SECRET) {
sodium_memzero(context->secret, context->secretlen);
context->secretlen = 0;
}
}
store32(&value, context->adlen);
crypto_generichash_blake2b_update(&BlakeHash, (const uint8_t *)&value, sizeof(value));
if (context->ad != NULL) {
crypto_generichash_blake2b_update(&BlakeHash, (const uint8_t *)context->ad,
context->adlen);
}
crypto_generichash_blake2b_final(&BlakeHash, blockhash, ARGON2_PREHASH_DIGEST_LENGTH);
}
int initialize(argon2_instance_t *instance, argon2_context *context) {
uint8_t blockhash[ARGON2_PREHASH_SEED_LENGTH];
int result = ARGON2_OK;
if (instance == NULL || context == NULL)
return ARGON2_INCORRECT_PARAMETER;
/* 1. Memory allocation */
if (NULL != context->allocate_cbk) {
uint8_t *p;
result = context->allocate_cbk(&p, instance->memory_blocks *
ARGON2_BLOCK_SIZE);
if (ARGON2_OK != result) {
return result;
}
memcpy(&(instance->memory), p, sizeof(instance->memory));
} else {
result = allocate_memory(&(instance->memory), instance->memory_blocks);
if (ARGON2_OK != result) {
return result;
}
}
/* 2. Initial hashing */
/* H_0 + 8 extra bytes to produce the first blocks */
/* uint8_t blockhash[ARGON2_PREHASH_SEED_LENGTH]; */
/* Hashing all inputs */
initial_hash(blockhash, context, instance->type);
/* Zeroing 8 extra bytes */
sodium_memzero(blockhash + ARGON2_PREHASH_DIGEST_LENGTH,
ARGON2_PREHASH_SEED_LENGTH - ARGON2_PREHASH_DIGEST_LENGTH);
/* 3. Creating first blocks, we always have at least two blocks in a slice
*/
fill_first_blocks(blockhash, instance);
/* Clearing the hash */
sodium_memzero(blockhash, ARGON2_PREHASH_SEED_LENGTH);
return ARGON2_OK;
}
int argon2_pick_best_implementation(void)
{
#if (defined(HAVE_EMMINTRIN_H) && defined(HAVE_TMMINTRIN_H)) || \
(defined(_MSC_VER) && (defined(_M_X64) || defined(_M_AMD64)))
if (sodium_runtime_has_ssse3()) {
fill_segment_fn fill_segment = fill_segment_ssse3;
return 0;
}
#endif
fill_segment_fn fill_segment = fill_segment_ref;
return 0;
}
@@ -0,0 +1,209 @@
/*
* Argon2 source code package
*
* Written by Daniel Dinu and Dmitry Khovratovich, 2015
*
* This work is licensed under a Creative Commons CC0 1.0 License/Waiver.
*
* You should have received a copy of the CC0 Public Domain Dedication along
* with
* this software. If not, see
* <http://creativecommons.org/publicdomain/zero/1.0/>.
*/
#ifndef ARGON2_CORE_H
#define ARGON2_CORE_H
#include "argon2.h"
/*************************Argon2 internal
* constants**************************************************/
enum argon2_core_constants {
/* Version of the algorithm */
ARGON2_VERSION_NUMBER = 0x10,
/* Memory block size in bytes */
ARGON2_BLOCK_SIZE = 1024,
ARGON2_QWORDS_IN_BLOCK = ARGON2_BLOCK_SIZE / 8,
ARGON2_OWORDS_IN_BLOCK = ARGON2_BLOCK_SIZE / 16,
/* Number of pseudo-random values generated by one call to Blake in Argon2i
to
generate reference block positions */
ARGON2_ADDRESSES_IN_BLOCK = 128,
/* Pre-hashing digest length and its extension*/
ARGON2_PREHASH_DIGEST_LENGTH = 64,
ARGON2_PREHASH_SEED_LENGTH = 72
};
/*************************Argon2 internal data
* types**************************************************/
/*
* Structure for the (1KB) memory block implemented as 128 64-bit words.
* Memory blocks can be copied, XORed. Internal words can be accessed by [] (no
* bounds checking).
*/
typedef struct block_ { uint64_t v[ARGON2_QWORDS_IN_BLOCK]; } block;
/*****************Functions that work with the block******************/
/* Initialize each byte of the block with @in */
void init_block_value(block *b, uint8_t in);
/* Copy block @src to block @dst */
void copy_block(block *dst, const block *src);
/* XOR @src onto @dst bytewise */
void xor_block(block *dst, const block *src);
/*
* Argon2 instance: memory pointer, number of passes, amount of memory, type,
* and derived values.
* Used to evaluate the number and location of blocks to construct in each
* thread
*/
typedef struct Argon2_instance_t {
block *memory; /* Memory pointer */
uint32_t passes; /* Number of passes */
uint32_t memory_blocks; /* Number of blocks in memory */
uint32_t segment_length;
uint32_t lane_length;
uint32_t lanes;
uint32_t threads;
argon2_type type;
int print_internals; /* whether to print the memory blocks */
} argon2_instance_t;
/*
* Argon2 position: where we construct the block right now. Used to distribute
* work between threads.
*/
typedef struct Argon2_position_t {
uint32_t pass;
uint32_t lane;
uint8_t slice;
uint32_t index;
} argon2_position_t;
/*Struct that holds the inputs for thread handling FillSegment*/
typedef struct Argon2_thread_data {
argon2_instance_t *instance_ptr;
argon2_position_t pos;
} argon2_thread_data;
/*************************Argon2 core
* functions**************************************************/
/* Allocates memory to the given pointer
* @param memory pointer to the pointer to the memory
* @param m_cost number of blocks to allocate in the memory
* @return ARGON2_OK if @memory is a valid pointer and memory is allocated
*/
int allocate_memory(block **memory, uint32_t m_cost);
/* Clears memory
* @param instance pointer to the current instance
* @param clear_memory indicates if we clear the memory with zeros.
*/
void clear_memory(argon2_instance_t *instance, int clear);
/* Deallocates memory
* @param memory pointer to the blocks
*/
void free_memory(block *memory);
/*
* Computes absolute position of reference block in the lane following a skewed
* distribution and using a pseudo-random value as input
* @param instance Pointer to the current instance
* @param position Pointer to the current position
* @param pseudo_rand 32-bit pseudo-random value used to determine the position
* @param same_lane Indicates if the block will be taken from the current lane.
* If so we can reference the current segment
* @pre All pointers must be valid
*/
uint32_t index_alpha(const argon2_instance_t *instance,
const argon2_position_t *position, uint32_t pseudo_rand,
int same_lane);
/*
* Function that validates all inputs against predefined restrictions and return
* an error code
* @param context Pointer to current Argon2 context
* @return ARGON2_OK if everything is all right, otherwise one of error codes
* (all defined in <argon2.h>
*/
int validate_inputs(const argon2_context *context);
/*
* Hashes all the inputs into @a blockhash[PREHASH_DIGEST_LENGTH], clears
* password and secret if needed
* @param context Pointer to the Argon2 internal structure containing memory
* pointer, and parameters for time and space requirements.
* @param blockhash Buffer for pre-hashing digest
* @param type Argon2 type
* @pre @a blockhash must have at least @a PREHASH_DIGEST_LENGTH bytes
* allocated
*/
void initial_hash(uint8_t *blockhash, argon2_context *context,
argon2_type type);
/*
* Function creates first 2 blocks per lane
* @param instance Pointer to the current instance
* @param blockhash Pointer to the pre-hashing digest
* @pre blockhash must point to @a PREHASH_SEED_LENGTH allocated values
*/
void fill_first_blocks(uint8_t *blockhash, const argon2_instance_t *instance);
/*
* Function allocates memory, hashes the inputs with Blake, and creates first
* two blocks. Returns the pointer to the main memory with 2 blocks per lane
* initialized
* @param context Pointer to the Argon2 internal structure containing memory
* pointer, and parameters for time and space requirements.
* @param instance Current Argon2 instance
* @return Zero if successful, -1 if memory failed to allocate. @context->state
* will be modified if successful.
*/
int initialize(argon2_instance_t *instance, argon2_context *context);
/*
* XORing the last block of each lane, hashing it, making the tag. Deallocates
* the memory.
* @param context Pointer to current Argon2 context (use only the out parameters
* from it)
* @param instance Pointer to current instance of Argon2
* @pre instance->state must point to necessary amount of memory
* @pre context->out must point to outlen bytes of memory
* @pre if context->free_cbk is not NULL, it should point to a function that
* deallocates memory
*/
void finalize(const argon2_context *context, argon2_instance_t *instance);
/*
* Function that fills the segment using previous segments also from other
* threads
* @param instance Pointer to the current instance
* @param position Current position
* @pre all block pointers must be valid
*/
typedef void (*fill_segment_fn)(const argon2_instance_t *instance,
argon2_position_t position);
int argon2_pick_best_implementation(void);
void fill_segment_ssse3(const argon2_instance_t *instance,
argon2_position_t position);
void fill_segment_ref(const argon2_instance_t *instance,
argon2_position_t position);
/*
* Function that fills the entire memory t_cost times based on the first two
* blocks in each lane
* @param instance Pointer to the current instance
*/
void fill_memory_blocks(argon2_instance_t *instance);
#endif
@@ -0,0 +1,445 @@
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <limits.h>
#include "argon2-encoding.h"
/*
* Example code for a decoder and encoder of "hash strings", with Argon2i
* parameters.
*
* This code comprises three sections:
*
* -- The first section contains generic Base64 encoding and decoding
* functions. It is conceptually applicable to any hash function
* implementation that uses Base64 to encode and decode parameters,
* salts and outputs. It could be made into a library, provided that
* the relevant functions are made public (non-static) and be given
* reasonable names to avoid collisions with other functions.
*
* -- The second section is specific to Argon2i. It encodes and decodes
* the parameters, salts and outputs. It does not compute the hash
* itself.
*
* -- The third section is test code, with a main() function. With
* this section, the whole file compiles as a stand-alone program
* that exercises the encoding and decoding functions with some
* test vectors.
*
* The code was originally written by Thomas Pornin <pornin@bolet.org>,
* to whom comments and remarks may be sent. It is released under what
* should amount to Public Domain or its closest equivalent; the
* following mantra is supposed to incarnate that fact with all the
* proper legal rituals:
*
* ---------------------------------------------------------------------
* This file is provided under the terms of Creative Commons CC0 1.0
* Public Domain Dedication. To the extent possible under law, the
* author (Thomas Pornin) has waived all copyright and related or
* neighboring rights to this file. This work is published from: Canada.
* ---------------------------------------------------------------------
*
* Copyright (c) 2015 Thomas Pornin
*/
/* ==================================================================== */
/*
* Common code; could be shared between different hash functions.
*
* Note: the Base64 functions below assume that uppercase letters (resp.
* lowercase letters) have consecutive numerical codes, that fit on 8
* bits. All modern systems use ASCII-compatible charsets, where these
* properties are true. If you are stuck with a dinosaur of a system
* that still defaults to EBCDIC then you already have much bigger
* interoperability issues to deal with.
*/
/*
* Some macros for constant-time comparisons. These work over values in
* the 0..255 range. Returned value is 0x00 on "false", 0xFF on "true".
*/
#define EQ(x, y) ((((0U-((unsigned)(x) ^ (unsigned)(y))) >> 8) & 0xFF) ^ 0xFF)
#define GT(x, y) ((((unsigned)(y) - (unsigned)(x)) >> 8) & 0xFF)
#define GE(x, y) (GT(y, x) ^ 0xFF)
#define LT(x, y) GT(y, x)
#define LE(x, y) GE(y, x)
/*
* Convert value x (0..63) to corresponding Base64 character.
*/
static int b64_byte_to_char(unsigned x) {
return (LT(x, 26) & (x + 'A')) |
(GE(x, 26) & LT(x, 52) & (x + ('a' - 26))) |
(GE(x, 52) & LT(x, 62) & (x + ('0' - 52))) | (EQ(x, 62) & '+') |
(EQ(x, 63) & '/');
}
/*
* Convert character c to the corresponding 6-bit value. If character c
* is not a Base64 character, then 0xFF (255) is returned.
*/
static unsigned b64_char_to_byte(int c) {
unsigned x;
x = (GE(c, 'A') & LE(c, 'Z') & (c - 'A')) |
(GE(c, 'a') & LE(c, 'z') & (c - ('a' - 26))) |
(GE(c, '0') & LE(c, '9') & (c - ('0' - 52))) | (EQ(c, '+') & 62) |
(EQ(c, '/') & 63);
return x | (EQ(x, 0) & (EQ(c, 'A') ^ 0xFF));
}
/*
* Convert some bytes to Base64. 'dst_len' is the length (in characters)
* of the output buffer 'dst'; if that buffer is not large enough to
* receive the result (including the terminating 0), then (size_t)-1
* is returned. Otherwise, the zero-terminated Base64 string is written
* in the buffer, and the output length (counted WITHOUT the terminating
* zero) is returned.
*/
static size_t to_base64(char *dst, size_t dst_len, const void *src,
size_t src_len) {
size_t olen;
const unsigned char *buf;
unsigned acc, acc_len;
olen = (src_len / 3) << 2;
switch (src_len % 3) {
case 2:
olen++;
/* fall through */
case 1:
olen += 2;
break;
}
if (dst_len <= olen) {
return (size_t)-1;
}
acc = 0;
acc_len = 0;
buf = (const unsigned char *)src;
while (src_len-- > 0) {
acc = (acc << 8) + (*buf++);
acc_len += 8;
while (acc_len >= 6) {
acc_len -= 6;
*dst++ = (char) b64_byte_to_char((acc >> acc_len) & 0x3F);
}
}
if (acc_len > 0) {
*dst++ = (char) b64_byte_to_char((acc << (6 - acc_len)) & 0x3F);
}
*dst++ = 0;
return olen;
}
/*
* Decode Base64 chars into bytes. The '*dst_len' value must initially
* contain the length of the output buffer '*dst'; when the decoding
* ends, the actual number of decoded bytes is written back in
* '*dst_len'.
*
* Decoding stops when a non-Base64 character is encountered, or when
* the output buffer capacity is exceeded. If an error occurred (output
* buffer is too small, invalid last characters leading to unprocessed
* buffered bits), then NULL is returned; otherwise, the returned value
* points to the first non-Base64 character in the source stream, which
* may be the terminating zero.
*/
static const char *from_base64(void *dst, size_t *dst_len, const char *src) {
size_t len;
unsigned char *buf;
unsigned acc, acc_len;
buf = (unsigned char *)dst;
len = 0;
acc = 0;
acc_len = 0;
for (;;) {
unsigned d;
d = b64_char_to_byte(*src);
if (d == 0xFF) {
break;
}
src++;
acc = (acc << 6) + d;
acc_len += 6;
if (acc_len >= 8) {
acc_len -= 8;
if ((len++) >= *dst_len) {
return NULL;
}
*buf++ = (acc >> acc_len) & 0xFF;
}
}
/*
* If the input length is equal to 1 modulo 4 (which is
* invalid), then there will remain 6 unprocessed bits;
* otherwise, only 0, 2 or 4 bits are buffered. The buffered
* bits must also all be zero.
*/
if (acc_len > 4 || (acc & (((unsigned)1 << acc_len) - 1)) != 0) {
return NULL;
}
*dst_len = len;
return src;
}
/*
* Decode decimal integer from 'str'; the value is written in '*v'.
* Returned value is a pointer to the next non-decimal character in the
* string. If there is no digit at all, or the value encoding is not
* minimal (extra leading zeros), or the value does not fit in an
* 'unsigned long', then NULL is returned.
*/
static const char *decode_decimal(const char *str, unsigned long *v) {
const char *orig;
unsigned long acc;
acc = 0;
for (orig = str;; str++) {
int c;
c = *str;
if (c < '0' || c > '9') {
break;
}
c -= '0';
if (acc > (ULONG_MAX / 10)) {
return NULL;
}
acc *= 10;
if ((unsigned long)c > (ULONG_MAX - acc)) {
return NULL;
}
acc += (unsigned long)c;
}
if (str == orig || (*orig == '0' && str != (orig + 1))) {
return NULL;
}
*v = acc;
return str;
}
/* ==================================================================== */
/*
* Code specific to Argon2i.
*
* The code below applies the following format:
*
* $argon2i$m=<num>,t=<num>,p=<num>[,keyid=<bin>][,data=<bin>][$<bin>[$<bin>]]
*
* where <num> is a decimal integer (positive, fits in an 'unsigned long')
* and <bin> is Base64-encoded data (no '=' padding characters, no newline
* or whitespace). The "keyid" is a binary identifier for a key (up to 8
* bytes); "data" is associated data (up to 32 bytes). When the 'keyid'
* (resp. the 'data') is empty, then it is ommitted from the output.
*
* The last two binary chunks (encoded in Base64) are, in that order,
* the salt and the output. Both are optional, but you cannot have an
* output without a salt. The binary salt length is between 8 and 48 bytes.
* The output length is always exactly 32 bytes.
*/
/*
* Decode an Argon2i hash string into the provided structure 'ctx'.
* Returned value is 1 on success, 0 on error.
*/
int decode_string(argon2_context *ctx, const char *str, argon2_type type) {
#define CC(prefix) \
do { \
size_t cc_len = strlen(prefix); \
if (strncmp(str, prefix, cc_len) != 0) { \
return 0; \
} \
str += cc_len; \
} while ((void)0, 0)
#define CC_opt(prefix, code) \
do { \
size_t cc_len = strlen(prefix); \
if (strncmp(str, prefix, cc_len) == 0) { \
str += cc_len; \
{ code; } \
} \
} while ((void)0, 0)
#define DECIMAL(x) \
do { \
unsigned long dec_x; \
str = decode_decimal(str, &dec_x); \
if (str == NULL) { \
return 0; \
} \
(x) = dec_x; \
} while ((void)0, 0)
#define BIN(buf, max_len, len) \
do { \
size_t bin_len = (max_len); \
str = from_base64(buf, &bin_len, str); \
if (str == NULL || bin_len > UINT32_MAX) { \
return 0; \
} \
(len) = (uint32_t)bin_len; \
} while ((void)0, 0)
size_t maxadlen = ctx->adlen;
size_t maxsaltlen = ctx->saltlen;
size_t maxoutlen = ctx->outlen;
ctx->adlen = 0;
ctx->saltlen = 0;
ctx->outlen = 0;
if (type == Argon2_i)
CC("$argon2i");
else
return 0;
CC("$m=");
DECIMAL(ctx->m_cost);
CC(",t=");
DECIMAL(ctx->t_cost);
CC(",p=");
DECIMAL(ctx->lanes);
ctx->threads = ctx->lanes;
/*
* Both m and t must be no more than 2^32-1. The tests below
* use a shift by 30 bits to avoid a direct comparison with
* 0xFFFFFFFF, which may trigger a spurious compiler warning
* on machines where 'unsigned long' is a 32-bit type.
*/
if (ctx->m_cost < 1 || (ctx->m_cost >> 30) > 3) {
return 0;
}
if (ctx->t_cost < 1 || (ctx->t_cost >> 30) > 3) {
return 0;
}
/*
* The parallelism p must be between 1 and 255. The memory cost
* parameter, expressed in kilobytes, must be at least 8 times
* the value of p.
*/
if (ctx->lanes < 1 || ctx->lanes > 255) {
return 0;
}
if (ctx->m_cost < (ctx->lanes << 3)) {
return 0;
}
CC_opt(",data=", BIN(ctx->ad, maxadlen, ctx->adlen));
if (*str == 0) {
return 1;
}
CC("$");
BIN(ctx->salt, maxsaltlen, ctx->saltlen);
if (ctx->saltlen < 8) {
return 0;
}
if (*str == 0) {
return 1;
}
CC("$");
BIN(ctx->out, maxoutlen, ctx->outlen);
if (ctx->outlen < 12) {
return 0;
}
return *str == 0;
#undef CC
#undef CC_opt
#undef DECIMAL
#undef BIN
}
#define U32_STR_MAXSIZE 11U
static void u32_to_string(char *str, uint32_t x) {
char tmp[U32_STR_MAXSIZE - 1U];
size_t i;
i = sizeof tmp;
do {
tmp[--i] = (x % (uint32_t) 10U) + '0';
x /= (uint32_t) 10U;
} while (x != 0U && i != 0U);
memcpy(str, &tmp[i], (sizeof tmp) - i);
str[(sizeof tmp) - i] = 0;
}
/*
* encode an argon2i hash string into the provided buffer. 'dst_len'
* contains the size, in characters, of the 'dst' buffer; if 'dst_len'
* is less than the number of required characters (including the
* terminating 0), then this function returns 0.
*
* if pp->output_len is 0, then the hash string will be a salt string
* (no output). if pp->salt_len is also 0, then the string will be a
* parameter-only string (no salt and no output).
*
* on success, 1 is returned.
*/
int encode_string(char *dst, size_t dst_len, argon2_context *ctx,
argon2_type type) {
#define SS(str) \
do { \
size_t pp_len = strlen(str); \
if (pp_len >= dst_len) { \
return 0; \
} \
memcpy(dst, str, pp_len + 1); \
dst += pp_len; \
dst_len -= pp_len; \
} while ((void)0, 0)
#define SX(x) \
do { \
char tmp[U32_STR_MAXSIZE]; \
u32_to_string(tmp, x); \
SS(tmp); \
} while ((void)0, 0)
#define SB(buf, len) \
do { \
size_t sb_len = to_base64(dst, dst_len, buf, len); \
if (sb_len == (size_t)-1) { \
return 0; \
} \
dst += sb_len; \
dst_len -= sb_len; \
} while ((void)0, 0)
if (type == Argon2_i)
SS("$argon2i$m=");
else
return 0;
SX(ctx->m_cost);
SS(",t=");
SX(ctx->t_cost);
SS(",p=");
SX(ctx->lanes);
if (ctx->adlen > 0) {
SS(",data=");
SB(ctx->ad, ctx->adlen);
}
if (ctx->saltlen == 0)
return 1;
SS("$");
SB(ctx->salt, ctx->saltlen);
if (ctx->outlen == 0)
return 1;
SS("$");
SB(ctx->out, ctx->outlen);
return 1;
#undef SS
#undef SX
#undef SB
}
@@ -0,0 +1,10 @@
#ifndef ENCODING_H
#define ENCODING_H
#include "argon2.h"
int encode_string(char *dst, size_t dst_len, argon2_context *ctx,
argon2_type type);
int decode_string(argon2_context *ctx, const char *str, argon2_type type);
#endif
@@ -0,0 +1,188 @@
/*
* Argon2 source code package
*
* Written by Daniel Dinu and Dmitry Khovratovich, 2015
*
* This work is licensed under a Creative Commons CC0 1.0 License/Waiver.
*
* You should have received a copy of the CC0 Public Domain Dedication along
* with
* this software. If not, see
* <http://creativecommons.org/publicdomain/zero/1.0/>.
*/
#include <stdint.h>
#include <stdlib.h>
#include <string.h>
#include "argon2.h"
#include "argon2-core.h"
#include "argon2-impl.h"
#include "blamka-round-ref.h"
/*
* Function fills a new memory block
* @param prev_block Pointer to the previous block
* @param ref_block Pointer to the reference block
* @param next_block Pointer to the block to be constructed
* @pre all block pointers must be valid
*/
static void fill_block(const block *prev_block, const block *ref_block,
block *next_block) {
block blockR, block_tmp;
unsigned i;
copy_block(&blockR, ref_block);
xor_block(&blockR, prev_block);
copy_block(&block_tmp, &blockR);
/* Apply Blake2 on columns of 64-bit words: (0,1,...,15) , then
(16,17,..31)... finally (112,113,...127) */
for (i = 0; i < 8; ++i) {
BLAKE2_ROUND_NOMSG(
blockR.v[16 * i], blockR.v[16 * i + 1], blockR.v[16 * i + 2],
blockR.v[16 * i + 3], blockR.v[16 * i + 4], blockR.v[16 * i + 5],
blockR.v[16 * i + 6], blockR.v[16 * i + 7], blockR.v[16 * i + 8],
blockR.v[16 * i + 9], blockR.v[16 * i + 10], blockR.v[16 * i + 11],
blockR.v[16 * i + 12], blockR.v[16 * i + 13], blockR.v[16 * i + 14],
blockR.v[16 * i + 15]);
}
/* Apply Blake2 on rows of 64-bit words: (0,1,16,17,...112,113), then
(2,3,18,19,...,114,115).. finally (14,15,30,31,...,126,127) */
for (i = 0; i < 8; i++) {
BLAKE2_ROUND_NOMSG(
blockR.v[2 * i], blockR.v[2 * i + 1], blockR.v[2 * i + 16],
blockR.v[2 * i + 17], blockR.v[2 * i + 32], blockR.v[2 * i + 33],
blockR.v[2 * i + 48], blockR.v[2 * i + 49], blockR.v[2 * i + 64],
blockR.v[2 * i + 65], blockR.v[2 * i + 80], blockR.v[2 * i + 81],
blockR.v[2 * i + 96], blockR.v[2 * i + 97], blockR.v[2 * i + 112],
blockR.v[2 * i + 113]);
}
copy_block(next_block, &block_tmp);
xor_block(next_block, &blockR);
}
/*
* Generate pseudo-random values to reference blocks in the segment and puts
* them into the array
* @param instance Pointer to the current instance
* @param position Pointer to the current position
* @param pseudo_rands Pointer to the array of 64-bit values
* @pre pseudo_rands must point to @a instance->segment_length allocated values
*/
static void generate_addresses(const argon2_instance_t *instance,
const argon2_position_t *position,
uint64_t *pseudo_rands) {
block zero_block, input_block, address_block;
uint32_t i;
init_block_value(&zero_block, 0);
init_block_value(&input_block, 0);
init_block_value(&address_block, 0);
if (instance != NULL && position != NULL) {
input_block.v[0] = position->pass;
input_block.v[1] = position->lane;
input_block.v[2] = position->slice;
input_block.v[3] = instance->memory_blocks;
input_block.v[4] = instance->passes;
input_block.v[5] = instance->type;
for (i = 0; i < instance->segment_length; ++i) {
if (i % ARGON2_ADDRESSES_IN_BLOCK == 0) {
input_block.v[6]++;
fill_block(&zero_block, &input_block, &address_block);
fill_block(&zero_block, &address_block, &address_block);
}
pseudo_rands[i] = address_block.v[i % ARGON2_ADDRESSES_IN_BLOCK];
}
}
}
void fill_segment_ref(const argon2_instance_t *instance,
argon2_position_t position) {
block *ref_block = NULL, *curr_block = NULL;
uint64_t pseudo_rand, ref_index, ref_lane;
uint32_t prev_offset, curr_offset;
uint32_t starting_index;
uint32_t i;
int data_independent_addressing = (instance->type == Argon2_i);
/* Pseudo-random values that determine the reference block position */
uint64_t *pseudo_rands = NULL;
if (instance == NULL) {
return;
}
pseudo_rands =
(uint64_t *)malloc(sizeof(uint64_t) * (instance->segment_length));
if (pseudo_rands == NULL) {
return;
}
if (data_independent_addressing) {
generate_addresses(instance, &position, pseudo_rands);
}
starting_index = 0;
if ((0 == position.pass) && (0 == position.slice)) {
starting_index = 2; /* we have already generated the first two blocks */
}
/* Offset of the current block */
curr_offset = position.lane * instance->lane_length +
position.slice * instance->segment_length + starting_index;
if (0 == curr_offset % instance->lane_length) {
/* Last block in this lane */
prev_offset = curr_offset + instance->lane_length - 1;
} else {
/* Previous block */
prev_offset = curr_offset - 1;
}
for (i = starting_index; i < instance->segment_length;
++i, ++curr_offset, ++prev_offset) {
/*1.1 Rotating prev_offset if needed */
if (curr_offset % instance->lane_length == 1) {
prev_offset = curr_offset - 1;
}
/* 1.2 Computing the index of the reference block */
/* 1.2.1 Taking pseudo-random value from the previous block */
if (data_independent_addressing) {
pseudo_rand = pseudo_rands[i];
} else {
pseudo_rand = instance->memory[prev_offset].v[0];
}
/* 1.2.2 Computing the lane of the reference block */
ref_lane = ((pseudo_rand >> 32)) % instance->lanes;
if ((position.pass == 0) && (position.slice == 0)) {
/* Can not reference other lanes yet */
ref_lane = position.lane;
}
/* 1.2.3 Computing the number of possible reference block within the
* lane.
*/
position.index = i;
ref_index = index_alpha(instance, &position, pseudo_rand & 0xFFFFFFFF,
ref_lane == position.lane);
/* 2 Creating a new block */
ref_block =
instance->memory + instance->lane_length * ref_lane + ref_index;
curr_block = instance->memory + curr_offset;
fill_block(instance->memory + prev_offset, ref_block, curr_block);
}
free(pseudo_rands);
}
@@ -0,0 +1,182 @@
/*
* Argon2 source code package
*
* Written by Daniel Dinu and Dmitry Khovratovich, 2015
*
* This work is licensed under a Creative Commons CC0 1.0 License/Waiver.
*
* You should have received a copy of the CC0 Public Domain Dedication along
* with
* this software. If not, see
* <http://creativecommons.org/publicdomain/zero/1.0/>.
*/
#include <stdint.h>
#include <stdlib.h>
#include <string.h>
#if (defined(HAVE_EMMINTRIN_H) && defined(HAVE_TMMINTRIN_H)) || \
(defined(_MSC_VER) && (defined(_M_X64) || defined(_M_AMD64)))
#pragma GCC target("sse2")
#pragma GCC target("ssse3")
#ifdef _MSC_VER
# include <intrin.h> /* for _mm_set_epi64x */
#endif
#include <emmintrin.h>
#include <tmmintrin.h>
#include "argon2.h"
#include "argon2-core.h"
#include "argon2-impl.h"
#include "blamka-round-ssse3.h"
static void fill_block(__m128i *state, const uint8_t *ref_block, uint8_t *next_block) {
__m128i block_XY[ARGON2_OWORDS_IN_BLOCK];
uint32_t i;
for (i = 0; i < ARGON2_OWORDS_IN_BLOCK; i++) {
block_XY[i] = state[i] = _mm_xor_si128(
state[i], _mm_loadu_si128((__m128i const *)(&ref_block[16 * i])));
}
for (i = 0; i < 8; ++i) {
BLAKE2_ROUND(state[8 * i + 0], state[8 * i + 1], state[8 * i + 2],
state[8 * i + 3], state[8 * i + 4], state[8 * i + 5],
state[8 * i + 6], state[8 * i + 7]);
}
for (i = 0; i < 8; ++i) {
BLAKE2_ROUND(state[8 * 0 + i], state[8 * 1 + i], state[8 * 2 + i],
state[8 * 3 + i], state[8 * 4 + i], state[8 * 5 + i],
state[8 * 6 + i], state[8 * 7 + i]);
}
for (i = 0; i < ARGON2_OWORDS_IN_BLOCK; i++) {
state[i] = _mm_xor_si128(state[i], block_XY[i]);
_mm_storeu_si128((__m128i *)(&next_block[16 * i]), state[i]);
}
}
static void generate_addresses(const argon2_instance_t *instance,
const argon2_position_t *position,
uint64_t *pseudo_rands) {
block address_block, input_block;
uint32_t i;
init_block_value(&address_block, 0);
init_block_value(&input_block, 0);
if (instance != NULL && position != NULL) {
input_block.v[0] = position->pass;
input_block.v[1] = position->lane;
input_block.v[2] = position->slice;
input_block.v[3] = instance->memory_blocks;
input_block.v[4] = instance->passes;
input_block.v[5] = instance->type;
for (i = 0; i < instance->segment_length; ++i) {
if (i % ARGON2_ADDRESSES_IN_BLOCK == 0) {
__m128i zero_block[ARGON2_OWORDS_IN_BLOCK];
__m128i zero2_block[ARGON2_OWORDS_IN_BLOCK];
memset(zero_block, 0, sizeof(zero_block));
memset(zero2_block, 0, sizeof(zero2_block));
input_block.v[6]++;
fill_block(zero_block, (uint8_t *)&input_block.v,
(uint8_t *)&address_block.v);
fill_block(zero2_block, (uint8_t *)&address_block.v,
(uint8_t *)&address_block.v);
}
pseudo_rands[i] = address_block.v[i % ARGON2_ADDRESSES_IN_BLOCK];
}
}
}
void fill_segment_ssse3(const argon2_instance_t *instance,
argon2_position_t position) {
block *ref_block = NULL, *curr_block = NULL;
uint64_t pseudo_rand, ref_index, ref_lane;
uint32_t prev_offset, curr_offset;
uint32_t starting_index, i;
__m128i state[64];
int data_independent_addressing = (instance->type == Argon2_i);
/* Pseudo-random values that determine the reference block position */
uint64_t *pseudo_rands = NULL;
if (instance == NULL) {
return;
}
pseudo_rands =
(uint64_t *)malloc(sizeof(uint64_t) * instance->segment_length);
if (pseudo_rands == NULL) {
return;
}
if (data_independent_addressing) {
generate_addresses(instance, &position, pseudo_rands);
}
starting_index = 0;
if ((0 == position.pass) && (0 == position.slice)) {
starting_index = 2; /* we have already generated the first two blocks */
}
/* Offset of the current block */
curr_offset = position.lane * instance->lane_length +
position.slice * instance->segment_length + starting_index;
if (0 == curr_offset % instance->lane_length) {
/* Last block in this lane */
prev_offset = curr_offset + instance->lane_length - 1;
} else {
/* Previous block */
prev_offset = curr_offset - 1;
}
memcpy(state, ((instance->memory + prev_offset)->v), ARGON2_BLOCK_SIZE);
for (i = starting_index; i < instance->segment_length;
++i, ++curr_offset, ++prev_offset) {
/*1.1 Rotating prev_offset if needed */
if (curr_offset % instance->lane_length == 1) {
prev_offset = curr_offset - 1;
}
/* 1.2 Computing the index of the reference block */
/* 1.2.1 Taking pseudo-random value from the previous block */
if (data_independent_addressing) {
pseudo_rand = pseudo_rands[i];
} else {
pseudo_rand = instance->memory[prev_offset].v[0];
}
/* 1.2.2 Computing the lane of the reference block */
ref_lane = ((pseudo_rand >> 32)) % instance->lanes;
if ((position.pass == 0) && (position.slice == 0)) {
/* Can not reference other lanes yet */
ref_lane = position.lane;
}
/* 1.2.3 Computing the number of possible reference block within the
* lane.
*/
position.index = i;
ref_index = index_alpha(instance, &position, pseudo_rand & 0xFFFFFFFF,
ref_lane == position.lane);
/* 2 Creating a new block */
ref_block =
instance->memory + instance->lane_length * ref_lane + ref_index;
curr_block = instance->memory + curr_offset;
fill_block(state, (uint8_t *)ref_block->v, (uint8_t *)curr_block->v);
}
free(pseudo_rands);
}
#endif
@@ -0,0 +1,129 @@
/*
BLAKE2 reference source code package - reference C implementations
Written in 2012 by Samuel Neves <sneves@dei.uc.pt>
To the extent possible under law, the author(s) have dedicated all copyright
and related and neighboring rights to this software to the public domain
worldwide. This software is distributed without any warranty.
You should have received a copy of the CC0 Public Domain Dedication along with
this software. If not, see <http://creativecommons.org/publicdomain/zero/1.0/>.
*/
#ifndef blake2_impl_H
#define blake2_impl_H
#include <stdint.h>
#include <string.h>
static inline uint32_t load32( const void *src )
{
#ifdef NATIVE_LITTLE_ENDIAN
uint32_t w;
memcpy(&w, src, sizeof w);
return w;
#else
const uint8_t *p = ( const uint8_t * )src;
uint32_t w = *p++;
w |= ( uint32_t )( *p++ ) << 8;
w |= ( uint32_t )( *p++ ) << 16;
w |= ( uint32_t )( *p++ ) << 24;
return w;
#endif
}
static inline uint64_t load64( const void *src )
{
#ifdef NATIVE_LITTLE_ENDIAN
uint64_t w;
memcpy(&w, src, sizeof w);
return w;
#else
const uint8_t *p = ( const uint8_t * )src;
uint64_t w = *p++;
w |= ( uint64_t )( *p++ ) << 8;
w |= ( uint64_t )( *p++ ) << 16;
w |= ( uint64_t )( *p++ ) << 24;
w |= ( uint64_t )( *p++ ) << 32;
w |= ( uint64_t )( *p++ ) << 40;
w |= ( uint64_t )( *p++ ) << 48;
w |= ( uint64_t )( *p++ ) << 56;
return w;
#endif
}
static inline void store32( void *dst, uint32_t w )
{
#ifdef NATIVE_LITTLE_ENDIAN
memcpy(dst, &w, sizeof w);
#else
uint8_t *p = ( uint8_t * )dst;
*p++ = ( uint8_t )w; w >>= 8;
*p++ = ( uint8_t )w; w >>= 8;
*p++ = ( uint8_t )w; w >>= 8;
*p++ = ( uint8_t )w;
#endif
}
static inline void store64( void *dst, uint64_t w )
{
#ifdef NATIVE_LITTLE_ENDIAN
memcpy(dst, &w, sizeof w);
#else
uint8_t *p = ( uint8_t * )dst;
*p++ = ( uint8_t )w; w >>= 8;
*p++ = ( uint8_t )w; w >>= 8;
*p++ = ( uint8_t )w; w >>= 8;
*p++ = ( uint8_t )w; w >>= 8;
*p++ = ( uint8_t )w; w >>= 8;
*p++ = ( uint8_t )w; w >>= 8;
*p++ = ( uint8_t )w; w >>= 8;
*p++ = ( uint8_t )w;
#endif
}
static inline uint64_t load48( const void *src )
{
const uint8_t *p = ( const uint8_t * )src;
uint64_t w = *p++;
w |= ( uint64_t )( *p++ ) << 8;
w |= ( uint64_t )( *p++ ) << 16;
w |= ( uint64_t )( *p++ ) << 24;
w |= ( uint64_t )( *p++ ) << 32;
w |= ( uint64_t )( *p++ ) << 40;
return w;
}
static inline void store48( void *dst, uint64_t w )
{
uint8_t *p = ( uint8_t * )dst;
*p++ = ( uint8_t )w; w >>= 8;
*p++ = ( uint8_t )w; w >>= 8;
*p++ = ( uint8_t )w; w >>= 8;
*p++ = ( uint8_t )w; w >>= 8;
*p++ = ( uint8_t )w; w >>= 8;
*p++ = ( uint8_t )w;
}
static inline uint32_t rotl32( const uint32_t w, const unsigned c )
{
return ( w << c ) | ( w >> ( 32 - c ) );
}
static inline uint64_t rotl64( const uint64_t w, const unsigned c )
{
return ( w << c ) | ( w >> ( 64 - c ) );
}
static inline uint32_t rotr32( const uint32_t w, const unsigned c )
{
return ( w >> c ) | ( w << ( 32 - c ) );
}
static inline uint64_t rotr64( const uint64_t w, const unsigned c )
{
return ( w >> c ) | ( w << ( 64 - c ) );
}
#endif
+249
View File
@@ -0,0 +1,249 @@
/*
* Argon2 source code package
*
* Written by Daniel Dinu and Dmitry Khovratovich, 2015
*
* This work is licensed under a Creative Commons CC0 1.0 License/Waiver.
*
* You should have received a copy of the CC0 Public Domain Dedication along
* with
* this software. If not, see
* <http://creativecommons.org/publicdomain/zero/1.0/>.
*/
#include <stdint.h>
#include <string.h>
#include <stdlib.h>
#include <stdio.h>
#include <limits.h>
#include "utils.h"
#include "argon2.h"
#include "argon2-encoding.h"
#include "argon2-core.h"
int argon2_core(argon2_context *context, argon2_type type) {
/* 1. Validate all inputs */
int result = validate_inputs(context);
uint32_t memory_blocks, segment_length;
argon2_instance_t instance;
if (ARGON2_OK != result) {
return result;
}
if (Argon2_i != type) {
return ARGON2_INCORRECT_TYPE;
}
/* 2. Align memory size */
/* Minimum memory_blocks = 8L blocks, where L is the number of lanes */
memory_blocks = context->m_cost;
if (memory_blocks < 2 * ARGON2_SYNC_POINTS * context->lanes) {
memory_blocks = 2 * ARGON2_SYNC_POINTS * context->lanes;
}
segment_length = memory_blocks / (context->lanes * ARGON2_SYNC_POINTS);
/* Ensure that all segments have equal length */
memory_blocks = segment_length * (context->lanes * ARGON2_SYNC_POINTS);
instance.memory = NULL;
instance.passes = context->t_cost;
instance.memory_blocks = memory_blocks;
instance.segment_length = segment_length;
instance.lane_length = segment_length * ARGON2_SYNC_POINTS;
instance.lanes = context->lanes;
instance.threads = context->threads;
instance.type = type;
/* 3. Initialization: Hashing inputs, allocating memory, filling first
* blocks
*/
result = initialize(&instance, context);
if (ARGON2_OK != result) {
return result;
}
/* 4. Filling memory */
fill_memory_blocks(&instance);
/* 5. Finalization */
finalize(context, &instance);
return ARGON2_OK;
}
int argon2_hash(const uint32_t t_cost, const uint32_t m_cost,
const uint32_t parallelism, const void *pwd,
const size_t pwdlen, const void *salt, const size_t saltlen,
void *hash, const size_t hashlen, char *encoded,
const size_t encodedlen, argon2_type type) {
argon2_context context;
int result;
uint8_t *out;
/* Detect and reject overflowing sizes */
/* TODO: This should probably be fixed in the function signature */
if (pwdlen > UINT32_MAX) {
return ARGON2_PWD_TOO_LONG;
}
if (hashlen > UINT32_MAX) {
return ARGON2_OUTPUT_TOO_LONG;
}
if (saltlen > UINT32_MAX) {
return ARGON2_SALT_TOO_LONG;
}
out = (uint8_t *) malloc(hashlen);
if (!out) {
return ARGON2_MEMORY_ALLOCATION_ERROR;
}
context.out = (uint8_t *)out;
context.outlen = (uint32_t)hashlen;
context.pwd = (uint8_t *)pwd;
context.pwdlen = (uint32_t)pwdlen;
context.salt = (uint8_t *)salt;
context.saltlen = (uint32_t)saltlen;
context.secret = NULL;
context.secretlen = 0;
context.ad = NULL;
context.adlen = 0;
context.t_cost = t_cost;
context.m_cost = m_cost;
context.lanes = parallelism;
context.threads = parallelism;
context.allocate_cbk = NULL;
context.free_cbk = NULL;
context.flags = ARGON2_DEFAULT_FLAGS;
result = argon2_core(&context, type);
if (result != ARGON2_OK) {
memset(out, 0x00, hashlen);
free(out);
return result;
}
/* if raw hash requested, write it */
if (hash) {
memcpy(hash, out, hashlen);
}
/* if encoding requested, write it */
if (encoded && encodedlen) {
if (!encode_string(encoded, encodedlen, &context, type)) {
memset(out, 0x00, hashlen);
memset(encoded, 0x00, encodedlen);
free(out);
return ARGON2_ENCODING_FAIL;
}
}
free(out);
return ARGON2_OK;
}
int argon2i_hash_encoded(const uint32_t t_cost, const uint32_t m_cost,
const uint32_t parallelism, const void *pwd,
const size_t pwdlen, const void *salt,
const size_t saltlen, const size_t hashlen,
char *encoded, const size_t encodedlen) {
return argon2_hash(t_cost, m_cost, parallelism, pwd, pwdlen, salt, saltlen,
NULL, hashlen, encoded, encodedlen, Argon2_i);
}
int argon2i_hash_raw(const uint32_t t_cost, const uint32_t m_cost,
const uint32_t parallelism, const void *pwd,
const size_t pwdlen, const void *salt,
const size_t saltlen, void *hash, const size_t hashlen) {
return argon2_hash(t_cost, m_cost, parallelism, pwd, pwdlen, salt, saltlen,
hash, hashlen, NULL, 0, Argon2_i);
}
int argon2_verify(const char *encoded, const void *pwd, const size_t pwdlen,
argon2_type type) {
argon2_context ctx;
uint8_t *out;
int ret;
/* max values, to be updated in decode_string */
ctx.adlen = 512;
ctx.saltlen = 512;
ctx.outlen = 512;
ctx.ad = (uint8_t *) malloc(ctx.adlen);
ctx.salt = (uint8_t *) malloc(ctx.saltlen);
ctx.out = (uint8_t *) malloc(ctx.outlen);
if (!ctx.out || !ctx.salt || !ctx.ad) {
free(ctx.ad);
free(ctx.salt);
free(ctx.out);
return ARGON2_MEMORY_ALLOCATION_ERROR;
}
out = (uint8_t *) malloc(ctx.outlen);
if (!out) {
free(ctx.ad);
free(ctx.salt);
free(ctx.out);
return ARGON2_MEMORY_ALLOCATION_ERROR;
}
if(decode_string(&ctx, encoded, type) != 1) {
free(ctx.ad);
free(ctx.salt);
free(ctx.out);
free(out);
return ARGON2_DECODING_FAIL;
}
ret = argon2_hash(ctx.t_cost, ctx.m_cost, ctx.threads, pwd, pwdlen, ctx.salt,
ctx.saltlen, out, ctx.outlen, NULL, 0, type);
free(ctx.ad);
free(ctx.salt);
if (ret != ARGON2_OK || sodium_memcmp(out, ctx.out, ctx.outlen) != 0) {
free(out);
free(ctx.out);
return ARGON2_DECODING_FAIL;
}
free(out);
free(ctx.out);
return ARGON2_OK;
}
int argon2i_verify(const char *encoded, const void *pwd, const size_t pwdlen) {
return argon2_verify(encoded, pwd, pwdlen, Argon2_i);
}
int argon2i(argon2_context *context) {
return argon2_core(context, Argon2_i);
}
int verify_i(argon2_context *context, const char *hash) {
int result;
if (0 == context->outlen || NULL == hash) {
return ARGON2_OUT_PTR_MISMATCH;
}
result = argon2_core(context, Argon2_i);
if (ARGON2_OK != result) {
return result;
}
return 0 == memcmp(hash, context->out, context->outlen);
}
+287
View File
@@ -0,0 +1,287 @@
/*
* Argon2 source code package
*
* Written by Daniel Dinu and Dmitry Khovratovich, 2015
*
* This work is licensed under a Creative Commons CC0 1.0 License/Waiver.
*
* You should have received a copy of the CC0 Public Domain Dedication along
* with
* this software. If not, see
* <http://creativecommons.org/publicdomain/zero/1.0/>.
*/
#ifndef ARGON2_H
#define ARGON2_H
#include <stdint.h>
#include <stddef.h>
#include <limits.h>
#if defined(__cplusplus)
extern "C" {
#endif
/*************************Argon2 input parameter
* restrictions**************************************************/
/* Minimum and maximum number of lanes (degree of parallelism) */
#define ARGON2_MIN_LANES UINT32_C(1)
#define ARGON2_MAX_LANES UINT32_C(0xFFFFFF)
/* Minimum and maximum number of threads */
#define ARGON2_MIN_THREADS UINT32_C(1)
#define ARGON2_MAX_THREADS UINT32_C(0xFFFFFF)
/* Number of synchronization points between lanes per pass */
#define ARGON2_SYNC_POINTS UINT32_C(4)
/* Minimum and maximum digest size in bytes */
#define ARGON2_MIN_OUTLEN UINT32_C(4)
#define ARGON2_MAX_OUTLEN UINT32_C(0xFFFFFFFF)
/* Minimum and maximum number of memory blocks (each of BLOCK_SIZE bytes) */
#define ARGON2_MIN_MEMORY (2 * ARGON2_SYNC_POINTS) /* 2 blocks per slice */
#define ARGON2_MIN(a, b) ((a) < (b) ? (a) : (b))
/* Max memory size is half the addressing space, topping at 2^32 blocks (4 TB)
*/
#define ARGON2_MAX_MEMORY_BITS \
ARGON2_MIN(UINT32_C(32), (sizeof(void *) * CHAR_BIT - 10 - 1))
#define ARGON2_MAX_MEMORY \
ARGON2_MIN(UINT32_C(0xFFFFFFFF), UINT64_C(1) << ARGON2_MAX_MEMORY_BITS)
/* Minimum and maximum number of passes */
#define ARGON2_MIN_TIME UINT32_C(1)
#define ARGON2_MAX_TIME UINT32_C(0xFFFFFFFF)
/* Minimum and maximum password length in bytes */
#define ARGON2_MIN_PWD_LENGTH UINT32_C(0)
#define ARGON2_MAX_PWD_LENGTH UINT32_C(0xFFFFFFFF)
/* Minimum and maximum associated data length in bytes */
#define ARGON2_MIN_AD_LENGTH UINT32_C(0)
#define ARGON2_MAX_AD_LENGTH UINT32_C(0xFFFFFFFF)
/* Minimum and maximum salt length in bytes */
#define ARGON2_MIN_SALT_LENGTH UINT32_C(8)
#define ARGON2_MAX_SALT_LENGTH UINT32_C(0xFFFFFFFF)
/* Minimum and maximum key length in bytes */
#define ARGON2_MIN_SECRET UINT32_C(0)
#define ARGON2_MAX_SECRET UINT32_C(0xFFFFFFFF)
#define ARGON2_FLAG_CLEAR_PASSWORD (UINT32_C(1) << 0)
#define ARGON2_FLAG_CLEAR_SECRET (UINT32_C(1) << 1)
#define ARGON2_FLAG_CLEAR_MEMORY (UINT32_C(1) << 2)
#define ARGON2_DEFAULT_FLAGS (ARGON2_FLAG_CLEAR_MEMORY)
/* Error codes */
typedef enum Argon2_ErrorCodes {
ARGON2_OK = 0,
ARGON2_OUTPUT_PTR_NULL = 1,
ARGON2_OUTPUT_TOO_SHORT = 2,
ARGON2_OUTPUT_TOO_LONG = 3,
ARGON2_PWD_TOO_SHORT = 4,
ARGON2_PWD_TOO_LONG = 5,
ARGON2_SALT_TOO_SHORT = 6,
ARGON2_SALT_TOO_LONG = 7,
ARGON2_AD_TOO_SHORT = 8,
ARGON2_AD_TOO_LONG = 9,
ARGON2_SECRET_TOO_SHORT = 10,
ARGON2_SECRET_TOO_LONG = 11,
ARGON2_TIME_TOO_SMALL = 12,
ARGON2_TIME_TOO_LARGE = 13,
ARGON2_MEMORY_TOO_LITTLE = 14,
ARGON2_MEMORY_TOO_MUCH = 15,
ARGON2_LANES_TOO_FEW = 16,
ARGON2_LANES_TOO_MANY = 17,
ARGON2_PWD_PTR_MISMATCH = 18, /* NULL ptr with non-zero length */
ARGON2_SALT_PTR_MISMATCH = 19, /* NULL ptr with non-zero length */
ARGON2_SECRET_PTR_MISMATCH = 20, /* NULL ptr with non-zero length */
ARGON2_AD_PTR_MISMATCH = 21, /* NULL ptr with non-zero length */
ARGON2_MEMORY_ALLOCATION_ERROR = 22,
ARGON2_FREE_MEMORY_CBK_NULL = 23,
ARGON2_ALLOCATE_MEMORY_CBK_NULL = 24,
ARGON2_INCORRECT_PARAMETER = 25,
ARGON2_INCORRECT_TYPE = 26,
ARGON2_OUT_PTR_MISMATCH = 27,
ARGON2_THREADS_TOO_FEW = 28,
ARGON2_THREADS_TOO_MANY = 29,
ARGON2_MISSING_ARGS = 30,
ARGON2_ENCODING_FAIL = 31,
ARGON2_DECODING_FAIL = 32,
ARGON2_ERROR_CODES_LENGTH /* Do NOT remove; Do NOT add error codes after
this
error code */
} argon2_error_codes;
/* Memory allocator types --- for external allocation */
typedef int (*allocate_fptr)(uint8_t **memory, size_t bytes_to_allocate);
typedef void (*deallocate_fptr)(uint8_t *memory, size_t bytes_to_allocate);
/* Argon2 external data structures */
/*
*****Context: structure to hold Argon2 inputs:
* output array and its length,
* password and its length,
* salt and its length,
* secret and its length,
* associated data and its length,
* number of passes, amount of used memory (in KBytes, can be rounded up a bit)
* number of parallel threads that will be run.
* All the parameters above affect the output hash value.
* Additionally, two function pointers can be provided to allocate and
deallocate the memory (if NULL, memory will be allocated internally).
* Also, three flags indicate whether to erase password, secret as soon as they
are pre-hashed (and thus not needed anymore), and the entire memory
****************************
Simplest situation: you have output array out[8], password is stored in
pwd[32], salt is stored in salt[16], you do not have keys nor associated data.
You need to spend 1 GB of RAM and you run 5 passes of Argon2 with 4 parallel
lanes.
You want to erase the password, but you're OK with last pass not being erased.
You want to use the default memory allocator.
Then you initialize
Argon2_Context(out,8,pwd,32,salt,16,NULL,0,NULL,0,5,1<<20,4,4,NULL,NULL,true,false,false,false).
*/
typedef struct Argon2_Context {
uint8_t *out; /* output array */
uint32_t outlen; /* digest length */
uint8_t *pwd; /* password array */
uint32_t pwdlen; /* password length */
uint8_t *salt; /* salt array */
uint32_t saltlen; /* salt length */
uint8_t *secret; /* key array */
uint32_t secretlen; /* key length */
uint8_t *ad; /* associated data array */
uint32_t adlen; /* associated data length */
uint32_t t_cost; /* number of passes */
uint32_t m_cost; /* amount of memory requested (KB) */
uint32_t lanes; /* number of lanes */
uint32_t threads; /* maximum number of threads */
allocate_fptr allocate_cbk; /* pointer to memory allocator */
deallocate_fptr free_cbk; /* pointer to memory deallocator */
uint32_t flags; /* array of bool options */
} argon2_context;
/* Argon2 primitive type */
typedef enum Argon2_type { Argon2_i = 1 } argon2_type;
/*
* Function that performs memory-hard hashing with certain degree of parallelism
* @param context Pointer to the Argon2 internal structure
* @return Error code if smth is wrong, ARGON2_OK otherwise
*/
int argon2_core(argon2_context *context, argon2_type type);
/**
* Hashes a password with Argon2i, producing an encoded hash
* @param t_cost Number of iterations
* @param m_cost Sets memory usage to 2^m_cost kibibytes
* @param parallelism Number of threads and compute lanes
* @param pwd Pointer to password
* @param pwdlen Password size in bytes
* @param salt Pointer to salt
* @param saltlen Salt size in bytes
* @param hashlen Desired length of the hash in bytes
* @param encoded Buffer where to write the encoded hash
* @param encodedlen Size of the buffer (thus max size of the encoded hash)
* @pre Different parallelism levels will give different results
* @pre Returns ARGON2_OK if successful
*/
int argon2i_hash_encoded(const uint32_t t_cost, const uint32_t m_cost,
const uint32_t parallelism, const void *pwd,
const size_t pwdlen, const void *salt,
const size_t saltlen, const size_t hashlen,
char *encoded, const size_t encodedlen);
/**
* Hashes a password with Argon2i, producing a raw hash
* @param t_cost Number of iterations
* @param m_cost Sets memory usage to 2^m_cost kibibytes
* @param parallelism Number of threads and compute lanes
* @param pwd Pointer to password
* @param pwdlen Password size in bytes
* @param salt Pointer to salt
* @param saltlen Salt size in bytes
* @param hash Buffer where to write the raw hash
* @param hashlen Desired length of the hash in bytes
* @pre Different parallelism levels will give different results
* @pre Returns ARGON2_OK if successful
*/
int argon2i_hash_raw(const uint32_t t_cost, const uint32_t m_cost,
const uint32_t parallelism, const void *pwd,
const size_t pwdlen, const void *salt,
const size_t saltlen, void *hash, const size_t hashlen);
/* generic function underlying the above ones */
int argon2_hash(const uint32_t t_cost, const uint32_t m_cost,
const uint32_t parallelism, const void *pwd,
const size_t pwdlen, const void *salt, const size_t saltlen,
void *hash, const size_t hashlen, char *encoded,
const size_t encodedlen, argon2_type type);
/**
* Verifies a password against an encoded string
* @param encoded String encoding parameters, salt, hash
* @param pwd Pointer to password
* @pre Returns ARGON2_OK if successful
*/
int argon2i_verify(const char *encoded, const void *pwd, const size_t pwdlen);
/* generic function underlying the above ones */
int argon2_verify(const char *encoded, const void *pwd, const size_t pwdlen,
argon2_type type);
/*
* * **************Argon2i: Version of Argon2 that picks memory blocks
*independent on the password and salt. Good for side-channels,
******************* but worse w.r.t. tradeoff attacks if
*******************only one pass is used***************
* @param context Pointer to current Argon2 context
* @return Zero if successful, a non zero error code otherwise
*/
int argon2i(argon2_context *context);
/*
* Verify if a given password is correct for Argon2i hashing
* @param context Pointer to current Argon2 context
* @param hash The password hash to verify. The length of the hash is
* specified by the context outlen member
* @return Zero if successful, a non zero error code otherwise
*/
int verify_i(argon2_context *context, const char *hash);
#if defined(__cplusplus)
}
#endif
#endif
@@ -0,0 +1,78 @@
#include <limits.h>
#include <stdint.h>
#include <stdlib.h>
#include <string.h>
#include "crypto_generichash_blake2b.h"
#include "utils.h"
#include "argon2-impl.h"
int blake2b_long(void *pout, size_t outlen, const void *in, size_t inlen) {
uint8_t *out = (uint8_t *)pout;
crypto_generichash_blake2b_state blake_state;
uint8_t outlen_bytes[4 /* sizeof(uint32_t) */] = {0};
int ret = -1;
if (outlen > UINT32_MAX) {
goto fail;
}
/* Ensure little-endian byte order! */
store32(outlen_bytes, (uint32_t)outlen);
#define TRY(statement) \
do { \
ret = statement; \
if (ret < 0) { \
goto fail; \
} \
} while ((void)0, 0)
if (outlen <= crypto_generichash_blake2b_BYTES_MAX) {
TRY(crypto_generichash_blake2b_init(&blake_state, NULL, 0U, outlen));
TRY(crypto_generichash_blake2b_update(&blake_state, outlen_bytes,
sizeof(outlen_bytes)));
TRY(crypto_generichash_blake2b_update(&blake_state,
(const unsigned char *) in,
inlen));
TRY(crypto_generichash_blake2b_final(&blake_state, out, outlen));
} else {
uint32_t toproduce;
uint8_t out_buffer[crypto_generichash_blake2b_BYTES_MAX];
uint8_t in_buffer[crypto_generichash_blake2b_BYTES_MAX];
TRY(crypto_generichash_blake2b_init(&blake_state, NULL, 0U,
crypto_generichash_blake2b_BYTES_MAX));
TRY(crypto_generichash_blake2b_update(&blake_state, outlen_bytes,
sizeof(outlen_bytes)));
TRY(crypto_generichash_blake2b_update(&blake_state,
(const unsigned char *) in,
inlen));
TRY(crypto_generichash_blake2b_final(&blake_state, out_buffer,
crypto_generichash_blake2b_BYTES_MAX));
memcpy(out, out_buffer, crypto_generichash_blake2b_BYTES_MAX / 2);
out += crypto_generichash_blake2b_BYTES_MAX / 2;
toproduce = (uint32_t)outlen - crypto_generichash_blake2b_BYTES_MAX / 2;
while (toproduce > crypto_generichash_blake2b_BYTES_MAX) {
memcpy(in_buffer, out_buffer, crypto_generichash_blake2b_BYTES_MAX);
TRY(crypto_generichash_blake2b(out_buffer, crypto_generichash_blake2b_BYTES_MAX,
in_buffer,
crypto_generichash_blake2b_BYTES_MAX,
NULL, 0U));
memcpy(out, out_buffer, crypto_generichash_blake2b_BYTES_MAX / 2);
out += crypto_generichash_blake2b_BYTES_MAX / 2;
toproduce -= crypto_generichash_blake2b_BYTES_MAX / 2;
}
memcpy(in_buffer, out_buffer, crypto_generichash_blake2b_BYTES_MAX);
TRY(crypto_generichash_blake2b(out_buffer, toproduce, in_buffer,
crypto_generichash_blake2b_BYTES_MAX,
NULL, 0U));
memcpy(out, out_buffer, toproduce);
}
fail:
sodium_memzero(&blake_state, sizeof(blake_state));
return ret;
#undef TRY
}
@@ -0,0 +1,8 @@
#ifndef blake2_long_H
#define blake2_long_H
#include <stddef.h>
int blake2b_long(void *pout, size_t outlen, const void *in, size_t inlen);
#endif
@@ -0,0 +1,38 @@
#ifndef BLAKE_ROUND_MKA_H
#define BLAKE_ROUND_MKA_H
#include "argon2-impl.h"
/*designed by the Lyra PHC team */
static inline uint64_t fBlaMka(uint64_t x, uint64_t y) {
const uint64_t m = UINT64_C(0xFFFFFFFF);
const uint64_t xy = (x & m) * (y & m);
return x + y + 2 * xy;
}
#define G(a, b, c, d) \
do { \
a = fBlaMka(a, b); \
d = rotr64(d ^ a, 32); \
c = fBlaMka(c, d); \
b = rotr64(b ^ c, 24); \
a = fBlaMka(a, b); \
d = rotr64(d ^ a, 16); \
c = fBlaMka(c, d); \
b = rotr64(b ^ c, 63); \
} while ((void)0, 0)
#define BLAKE2_ROUND_NOMSG(v0, v1, v2, v3, v4, v5, v6, v7, v8, v9, v10, v11, \
v12, v13, v14, v15) \
do { \
G(v0, v4, v8, v12); \
G(v1, v5, v9, v13); \
G(v2, v6, v10, v14); \
G(v3, v7, v11, v15); \
G(v0, v5, v10, v15); \
G(v1, v6, v11, v12); \
G(v2, v7, v8, v13); \
G(v3, v4, v9, v14); \
} while ((void)0, 0)
#endif
@@ -0,0 +1,146 @@
#ifndef BLAKE_ROUND_MKA_OPT_SSSE3_H
#define BLAKE_ROUND_MKA_OPT_SSSE3_H
#include "argon2-impl.h"
#define r16 \
(_mm_setr_epi8(2, 3, 4, 5, 6, 7, 0, 1, 10, 11, 12, 13, 14, 15, 8, 9))
#define r24 \
(_mm_setr_epi8(3, 4, 5, 6, 7, 0, 1, 2, 11, 12, 13, 14, 15, 8, 9, 10))
#define _mm_roti_epi64(x, c) \
(-(c) == 32) \
? _mm_shuffle_epi32((x), _MM_SHUFFLE(2, 3, 0, 1)) \
: (-(c) == 24) \
? _mm_shuffle_epi8((x), r24) \
: (-(c) == 16) \
? _mm_shuffle_epi8((x), r16) \
: (-(c) == 63) \
? _mm_xor_si128(_mm_srli_epi64((x), -(c)), \
_mm_add_epi64((x), (x))) \
: _mm_xor_si128(_mm_srli_epi64((x), -(c)), \
_mm_slli_epi64((x), 64 - (-(c))))
static inline __m128i fBlaMka(__m128i x, __m128i y) {
const __m128i z = _mm_mul_epu32(x, y);
return _mm_add_epi64(_mm_add_epi64(x, y), _mm_add_epi64(z, z));
}
#define G1(A0, B0, C0, D0, A1, B1, C1, D1) \
do { \
A0 = fBlaMka(A0, B0); \
A1 = fBlaMka(A1, B1); \
\
D0 = _mm_xor_si128(D0, A0); \
D1 = _mm_xor_si128(D1, A1); \
\
D0 = _mm_roti_epi64(D0, -32); \
D1 = _mm_roti_epi64(D1, -32); \
\
C0 = fBlaMka(C0, D0); \
C1 = fBlaMka(C1, D1); \
\
B0 = _mm_xor_si128(B0, C0); \
B1 = _mm_xor_si128(B1, C1); \
\
B0 = _mm_roti_epi64(B0, -24); \
B1 = _mm_roti_epi64(B1, -24); \
} while ((void)0, 0)
#define G2(A0, B0, C0, D0, A1, B1, C1, D1) \
do { \
A0 = fBlaMka(A0, B0); \
A1 = fBlaMka(A1, B1); \
\
D0 = _mm_xor_si128(D0, A0); \
D1 = _mm_xor_si128(D1, A1); \
\
D0 = _mm_roti_epi64(D0, -16); \
D1 = _mm_roti_epi64(D1, -16); \
\
C0 = fBlaMka(C0, D0); \
C1 = fBlaMka(C1, D1); \
\
B0 = _mm_xor_si128(B0, C0); \
B1 = _mm_xor_si128(B1, C1); \
\
B0 = _mm_roti_epi64(B0, -63); \
B1 = _mm_roti_epi64(B1, -63); \
} while ((void)0, 0)
#if defined(__SSSE3__)
#define DIAGONALIZE(A0, B0, C0, D0, A1, B1, C1, D1) \
do { \
__m128i t0 = _mm_alignr_epi8(B1, B0, 8); \
__m128i t1 = _mm_alignr_epi8(B0, B1, 8); \
B0 = t0; \
B1 = t1; \
\
t0 = C0; \
C0 = C1; \
C1 = t0; \
\
t0 = _mm_alignr_epi8(D1, D0, 8); \
t1 = _mm_alignr_epi8(D0, D1, 8); \
D0 = t1; \
D1 = t0; \
} while ((void)0, 0)
#define UNDIAGONALIZE(A0, B0, C0, D0, A1, B1, C1, D1) \
do { \
__m128i t0 = _mm_alignr_epi8(B0, B1, 8); \
__m128i t1 = _mm_alignr_epi8(B1, B0, 8); \
B0 = t0; \
B1 = t1; \
\
t0 = C0; \
C0 = C1; \
C1 = t0; \
\
t0 = _mm_alignr_epi8(D0, D1, 8); \
t1 = _mm_alignr_epi8(D1, D0, 8); \
D0 = t1; \
D1 = t0; \
} while ((void)0, 0)
#else /* SSE2 */
#define DIAGONALIZE(A0, B0, C0, D0, A1, B1, C1, D1) \
do { \
__m128i t0 = D0; \
__m128i t1 = B0; \
D0 = C0; \
C0 = C1; \
C1 = D0; \
D0 = _mm_unpackhi_epi64(D1, _mm_unpacklo_epi64(t0, t0)); \
D1 = _mm_unpackhi_epi64(t0, _mm_unpacklo_epi64(D1, D1)); \
B0 = _mm_unpackhi_epi64(B0, _mm_unpacklo_epi64(B1, B1)); \
B1 = _mm_unpackhi_epi64(B1, _mm_unpacklo_epi64(t1, t1)); \
} while ((void)0, 0)
#define UNDIAGONALIZE(A0, B0, C0, D0, A1, B1, C1, D1) \
do { \
__m128i t0, t1; \
t0 = C0; \
C0 = C1; \
C1 = t0; \
t0 = B0; \
t1 = D0; \
B0 = _mm_unpackhi_epi64(B1, _mm_unpacklo_epi64(B0, B0)); \
B1 = _mm_unpackhi_epi64(t0, _mm_unpacklo_epi64(B1, B1)); \
D0 = _mm_unpackhi_epi64(D0, _mm_unpacklo_epi64(D1, D1)); \
D1 = _mm_unpackhi_epi64(D1, _mm_unpacklo_epi64(t1, t1)); \
} while ((void)0, 0)
#endif
#define BLAKE2_ROUND(A0, A1, B0, B1, C0, C1, D0, D1) \
do { \
G1(A0, B0, C0, D0, A1, B1, C1, D1); \
G2(A0, B0, C0, D0, A1, B1, C1, D1); \
\
DIAGONALIZE(A0, B0, C0, D0, A1, B1, C1, D1); \
\
G1(A0, B0, C0, D0, A1, B1, C1, D1); \
G2(A0, B0, C0, D0, A1, B1, C1, D1); \
\
UNDIAGONALIZE(A0, B0, C0, D0, A1, B1, C1, D1); \
} while ((void)0, 0)
#endif