mirror of
https://github.com/ossrs/srs.git
synced 2025-03-09 15:49:59 +00:00
Upgrade libsrtp from 2.0.0 to 2.3.0, with source code. 4.0.79
This commit is contained in:
parent
3749d4d833
commit
8089fc004c
111 changed files with 45307 additions and 5 deletions
490
trunk/3rdparty/libsrtp-2-fit/crypto/math/datatypes.c
vendored
Normal file
490
trunk/3rdparty/libsrtp-2-fit/crypto/math/datatypes.c
vendored
Normal file
|
@ -0,0 +1,490 @@
|
|||
/*
|
||||
* datatypes.c
|
||||
*
|
||||
* data types for finite fields and functions for input, output, and
|
||||
* manipulation
|
||||
*
|
||||
* David A. McGrew
|
||||
* Cisco Systems, Inc.
|
||||
*/
|
||||
/*
|
||||
*
|
||||
* Copyright (c) 2001-2017 Cisco Systems, Inc.
|
||||
* All rights reserved.
|
||||
*
|
||||
* Redistribution and use in source and binary forms, with or without
|
||||
* modification, are permitted provided that the following conditions
|
||||
* are met:
|
||||
*
|
||||
* Redistributions of source code must retain the above copyright
|
||||
* notice, this list of conditions and the following disclaimer.
|
||||
*
|
||||
* 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.
|
||||
*
|
||||
* Neither the name of the Cisco Systems, Inc. nor the names of its
|
||||
* contributors may be used to endorse or promote products derived
|
||||
* from this software without specific prior written permission.
|
||||
*
|
||||
* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS 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
|
||||
* COPYRIGHT HOLDERS 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 HAVE_CONFIG_H
|
||||
#include <config.h>
|
||||
#endif
|
||||
|
||||
#ifdef OPENSSL
|
||||
#include <openssl/crypto.h>
|
||||
#endif
|
||||
|
||||
#include "datatypes.h"
|
||||
|
||||
static const int8_t octet_weight[256] = {
|
||||
0, 1, 1, 2, 1, 2, 2, 3, 1, 2, 2, 3, 2, 3, 3, 4, 1, 2, 2, 3, 2, 3, 3, 4,
|
||||
2, 3, 3, 4, 3, 4, 4, 5, 1, 2, 2, 3, 2, 3, 3, 4, 2, 3, 3, 4, 3, 4, 4, 5,
|
||||
2, 3, 3, 4, 3, 4, 4, 5, 3, 4, 4, 5, 4, 5, 5, 6, 1, 2, 2, 3, 2, 3, 3, 4,
|
||||
2, 3, 3, 4, 3, 4, 4, 5, 2, 3, 3, 4, 3, 4, 4, 5, 3, 4, 4, 5, 4, 5, 5, 6,
|
||||
2, 3, 3, 4, 3, 4, 4, 5, 3, 4, 4, 5, 4, 5, 5, 6, 3, 4, 4, 5, 4, 5, 5, 6,
|
||||
4, 5, 5, 6, 5, 6, 6, 7, 1, 2, 2, 3, 2, 3, 3, 4, 2, 3, 3, 4, 3, 4, 4, 5,
|
||||
2, 3, 3, 4, 3, 4, 4, 5, 3, 4, 4, 5, 4, 5, 5, 6, 2, 3, 3, 4, 3, 4, 4, 5,
|
||||
3, 4, 4, 5, 4, 5, 5, 6, 3, 4, 4, 5, 4, 5, 5, 6, 4, 5, 5, 6, 5, 6, 6, 7,
|
||||
2, 3, 3, 4, 3, 4, 4, 5, 3, 4, 4, 5, 4, 5, 5, 6, 3, 4, 4, 5, 4, 5, 5, 6,
|
||||
4, 5, 5, 6, 5, 6, 6, 7, 3, 4, 4, 5, 4, 5, 5, 6, 4, 5, 5, 6, 5, 6, 6, 7,
|
||||
4, 5, 5, 6, 5, 6, 6, 7, 5, 6, 6, 7, 6, 7, 7, 8
|
||||
};
|
||||
|
||||
int octet_get_weight(uint8_t octet)
|
||||
{
|
||||
return (int)octet_weight[octet];
|
||||
}
|
||||
|
||||
/*
|
||||
* bit_string is a buffer that is used to hold output strings, e.g.
|
||||
* for printing.
|
||||
*/
|
||||
|
||||
/* the value MAX_PRINT_STRING_LEN is defined in datatypes.h */
|
||||
|
||||
char bit_string[MAX_PRINT_STRING_LEN];
|
||||
|
||||
uint8_t srtp_nibble_to_hex_char(uint8_t nibble)
|
||||
{
|
||||
char buf[16] = { '0', '1', '2', '3', '4', '5', '6', '7',
|
||||
'8', '9', 'a', 'b', 'c', 'd', 'e', 'f' };
|
||||
return buf[nibble & 0xF];
|
||||
}
|
||||
|
||||
char *srtp_octet_string_hex_string(const void *s, int length)
|
||||
{
|
||||
const uint8_t *str = (const uint8_t *)s;
|
||||
int i;
|
||||
|
||||
/* double length, since one octet takes two hex characters */
|
||||
length *= 2;
|
||||
|
||||
/* truncate string if it would be too long */
|
||||
if (length > MAX_PRINT_STRING_LEN)
|
||||
length = MAX_PRINT_STRING_LEN - 2;
|
||||
|
||||
for (i = 0; i < length; i += 2) {
|
||||
bit_string[i] = srtp_nibble_to_hex_char(*str >> 4);
|
||||
bit_string[i + 1] = srtp_nibble_to_hex_char(*str++ & 0xF);
|
||||
}
|
||||
bit_string[i] = 0; /* null terminate string */
|
||||
return bit_string;
|
||||
}
|
||||
|
||||
char *v128_hex_string(v128_t *x)
|
||||
{
|
||||
int i, j;
|
||||
|
||||
for (i = j = 0; i < 16; i++) {
|
||||
bit_string[j++] = srtp_nibble_to_hex_char(x->v8[i] >> 4);
|
||||
bit_string[j++] = srtp_nibble_to_hex_char(x->v8[i] & 0xF);
|
||||
}
|
||||
|
||||
bit_string[j] = 0; /* null terminate string */
|
||||
return bit_string;
|
||||
}
|
||||
|
||||
char *v128_bit_string(v128_t *x)
|
||||
{
|
||||
int j, i;
|
||||
uint32_t mask;
|
||||
|
||||
for (j = i = 0; j < 4; j++) {
|
||||
for (mask = 0x80000000; mask > 0; mask >>= 1) {
|
||||
if (x->v32[j] & mask)
|
||||
bit_string[i] = '1';
|
||||
else
|
||||
bit_string[i] = '0';
|
||||
++i;
|
||||
}
|
||||
}
|
||||
bit_string[128] = 0; /* null terminate string */
|
||||
|
||||
return bit_string;
|
||||
}
|
||||
|
||||
void v128_copy_octet_string(v128_t *x, const uint8_t s[16])
|
||||
{
|
||||
#ifdef ALIGNMENT_32BIT_REQUIRED
|
||||
if ((((uint32_t)&s[0]) & 0x3) != 0)
|
||||
#endif
|
||||
{
|
||||
x->v8[0] = s[0];
|
||||
x->v8[1] = s[1];
|
||||
x->v8[2] = s[2];
|
||||
x->v8[3] = s[3];
|
||||
x->v8[4] = s[4];
|
||||
x->v8[5] = s[5];
|
||||
x->v8[6] = s[6];
|
||||
x->v8[7] = s[7];
|
||||
x->v8[8] = s[8];
|
||||
x->v8[9] = s[9];
|
||||
x->v8[10] = s[10];
|
||||
x->v8[11] = s[11];
|
||||
x->v8[12] = s[12];
|
||||
x->v8[13] = s[13];
|
||||
x->v8[14] = s[14];
|
||||
x->v8[15] = s[15];
|
||||
}
|
||||
#ifdef ALIGNMENT_32BIT_REQUIRED
|
||||
else {
|
||||
v128_t *v = (v128_t *)&s[0];
|
||||
|
||||
v128_copy(x, v);
|
||||
}
|
||||
#endif
|
||||
}
|
||||
|
||||
#ifndef DATATYPES_USE_MACROS /* little functions are not macros */
|
||||
|
||||
void v128_set_to_zero(v128_t *x)
|
||||
{
|
||||
_v128_set_to_zero(x);
|
||||
}
|
||||
|
||||
void v128_copy(v128_t *x, const v128_t *y)
|
||||
{
|
||||
_v128_copy(x, y);
|
||||
}
|
||||
|
||||
void v128_xor(v128_t *z, v128_t *x, v128_t *y)
|
||||
{
|
||||
_v128_xor(z, x, y);
|
||||
}
|
||||
|
||||
void v128_and(v128_t *z, v128_t *x, v128_t *y)
|
||||
{
|
||||
_v128_and(z, x, y);
|
||||
}
|
||||
|
||||
void v128_or(v128_t *z, v128_t *x, v128_t *y)
|
||||
{
|
||||
_v128_or(z, x, y);
|
||||
}
|
||||
|
||||
void v128_complement(v128_t *x)
|
||||
{
|
||||
_v128_complement(x);
|
||||
}
|
||||
|
||||
int v128_is_eq(const v128_t *x, const v128_t *y)
|
||||
{
|
||||
return _v128_is_eq(x, y);
|
||||
}
|
||||
|
||||
int v128_xor_eq(v128_t *x, const v128_t *y)
|
||||
{
|
||||
return _v128_xor_eq(x, y);
|
||||
}
|
||||
|
||||
int v128_get_bit(const v128_t *x, int i)
|
||||
{
|
||||
return _v128_get_bit(x, i);
|
||||
}
|
||||
|
||||
void v128_set_bit(v128_t *x, int i)
|
||||
{
|
||||
_v128_set_bit(x, i);
|
||||
}
|
||||
|
||||
void v128_clear_bit(v128_t *x, int i)
|
||||
{
|
||||
_v128_clear_bit(x, i);
|
||||
}
|
||||
|
||||
void v128_set_bit_to(v128_t *x, int i, int y)
|
||||
{
|
||||
_v128_set_bit_to(x, i, y);
|
||||
}
|
||||
|
||||
#endif /* DATATYPES_USE_MACROS */
|
||||
|
||||
void v128_right_shift(v128_t *x, int shift)
|
||||
{
|
||||
const int base_index = shift >> 5;
|
||||
const int bit_index = shift & 31;
|
||||
int i, from;
|
||||
uint32_t b;
|
||||
|
||||
if (shift > 127) {
|
||||
v128_set_to_zero(x);
|
||||
return;
|
||||
}
|
||||
|
||||
if (bit_index == 0) {
|
||||
/* copy each word from left size to right side */
|
||||
x->v32[4 - 1] = x->v32[4 - 1 - base_index];
|
||||
for (i = 4 - 1; i > base_index; i--)
|
||||
x->v32[i - 1] = x->v32[i - 1 - base_index];
|
||||
|
||||
} else {
|
||||
/* set each word to the "or" of the two bit-shifted words */
|
||||
for (i = 4; i > base_index; i--) {
|
||||
from = i - 1 - base_index;
|
||||
b = x->v32[from] << bit_index;
|
||||
if (from > 0)
|
||||
b |= x->v32[from - 1] >> (32 - bit_index);
|
||||
x->v32[i - 1] = b;
|
||||
}
|
||||
}
|
||||
|
||||
/* now wrap up the final portion */
|
||||
for (i = 0; i < base_index; i++)
|
||||
x->v32[i] = 0;
|
||||
}
|
||||
|
||||
void v128_left_shift(v128_t *x, int shift)
|
||||
{
|
||||
int i;
|
||||
const int base_index = shift >> 5;
|
||||
const int bit_index = shift & 31;
|
||||
|
||||
if (shift > 127) {
|
||||
v128_set_to_zero(x);
|
||||
return;
|
||||
}
|
||||
|
||||
if (bit_index == 0) {
|
||||
for (i = 0; i < 4 - base_index; i++)
|
||||
x->v32[i] = x->v32[i + base_index];
|
||||
} else {
|
||||
for (i = 0; i < 4 - base_index - 1; i++)
|
||||
x->v32[i] = (x->v32[i + base_index] >> bit_index) ^
|
||||
(x->v32[i + base_index + 1] << (32 - bit_index));
|
||||
x->v32[4 - base_index - 1] = x->v32[4 - 1] >> bit_index;
|
||||
}
|
||||
|
||||
/* now wrap up the final portion */
|
||||
for (i = 4 - base_index; i < 4; i++)
|
||||
x->v32[i] = 0;
|
||||
}
|
||||
|
||||
/* functions manipulating bitvector_t */
|
||||
|
||||
#ifndef DATATYPES_USE_MACROS /* little functions are not macros */
|
||||
|
||||
int bitvector_get_bit(const bitvector_t *v, int bit_index)
|
||||
{
|
||||
return _bitvector_get_bit(v, bit_index);
|
||||
}
|
||||
|
||||
void bitvector_set_bit(bitvector_t *v, int bit_index)
|
||||
{
|
||||
_bitvector_set_bit(v, bit_index);
|
||||
}
|
||||
|
||||
void bitvector_clear_bit(bitvector_t *v, int bit_index)
|
||||
{
|
||||
_bitvector_clear_bit(v, bit_index);
|
||||
}
|
||||
|
||||
#endif /* DATATYPES_USE_MACROS */
|
||||
|
||||
int bitvector_alloc(bitvector_t *v, unsigned long length)
|
||||
{
|
||||
unsigned long l;
|
||||
|
||||
/* Round length up to a multiple of bits_per_word */
|
||||
length =
|
||||
(length + bits_per_word - 1) & ~(unsigned long)((bits_per_word - 1));
|
||||
|
||||
l = length / bits_per_word * bytes_per_word;
|
||||
|
||||
/* allocate memory, then set parameters */
|
||||
if (l == 0) {
|
||||
v->word = NULL;
|
||||
v->length = 0;
|
||||
return -1;
|
||||
} else {
|
||||
v->word = (uint32_t *)srtp_crypto_alloc(l);
|
||||
if (v->word == NULL) {
|
||||
v->length = 0;
|
||||
return -1;
|
||||
}
|
||||
}
|
||||
v->length = length;
|
||||
|
||||
/* initialize bitvector to zero */
|
||||
bitvector_set_to_zero(v);
|
||||
|
||||
return 0;
|
||||
}
|
||||
|
||||
void bitvector_dealloc(bitvector_t *v)
|
||||
{
|
||||
if (v->word != NULL)
|
||||
srtp_crypto_free(v->word);
|
||||
v->word = NULL;
|
||||
v->length = 0;
|
||||
}
|
||||
|
||||
void bitvector_set_to_zero(bitvector_t *x)
|
||||
{
|
||||
/* C99 guarantees that memset(0) will set the value 0 for uint32_t */
|
||||
memset(x->word, 0, x->length >> 3);
|
||||
}
|
||||
|
||||
char *bitvector_bit_string(bitvector_t *x, char *buf, int len)
|
||||
{
|
||||
int j, i;
|
||||
uint32_t mask;
|
||||
|
||||
for (j = i = 0; j < (int)(x->length >> 5) && i < len - 1; j++) {
|
||||
for (mask = 0x80000000; mask > 0; mask >>= 1) {
|
||||
if (x->word[j] & mask)
|
||||
buf[i] = '1';
|
||||
else
|
||||
buf[i] = '0';
|
||||
++i;
|
||||
if (i >= len - 1)
|
||||
break;
|
||||
}
|
||||
}
|
||||
buf[i] = 0; /* null terminate string */
|
||||
|
||||
return buf;
|
||||
}
|
||||
|
||||
void bitvector_left_shift(bitvector_t *x, int shift)
|
||||
{
|
||||
int i;
|
||||
const int base_index = shift >> 5;
|
||||
const int bit_index = shift & 31;
|
||||
const int word_length = x->length >> 5;
|
||||
|
||||
if (shift >= (int)x->length) {
|
||||
bitvector_set_to_zero(x);
|
||||
return;
|
||||
}
|
||||
|
||||
if (bit_index == 0) {
|
||||
for (i = 0; i < word_length - base_index; i++)
|
||||
x->word[i] = x->word[i + base_index];
|
||||
} else {
|
||||
for (i = 0; i < word_length - base_index - 1; i++)
|
||||
x->word[i] = (x->word[i + base_index] >> bit_index) ^
|
||||
(x->word[i + base_index + 1] << (32 - bit_index));
|
||||
x->word[word_length - base_index - 1] =
|
||||
x->word[word_length - 1] >> bit_index;
|
||||
}
|
||||
|
||||
/* now wrap up the final portion */
|
||||
for (i = word_length - base_index; i < word_length; i++)
|
||||
x->word[i] = 0;
|
||||
}
|
||||
|
||||
int srtp_octet_string_is_eq(uint8_t *a, uint8_t *b, int len)
|
||||
{
|
||||
uint8_t *end = b + len;
|
||||
uint8_t accumulator = 0;
|
||||
|
||||
/*
|
||||
* We use this somewhat obscure implementation to try to ensure the running
|
||||
* time only depends on len, even accounting for compiler optimizations.
|
||||
* The accumulator ends up zero iff the strings are equal.
|
||||
*/
|
||||
while (b < end)
|
||||
accumulator |= (*a++ ^ *b++);
|
||||
|
||||
/* Return 1 if *not* equal. */
|
||||
return accumulator != 0;
|
||||
}
|
||||
|
||||
void srtp_cleanse(void *s, size_t len)
|
||||
{
|
||||
volatile unsigned char *p = (volatile unsigned char *)s;
|
||||
while (len--)
|
||||
*p++ = 0;
|
||||
}
|
||||
|
||||
void octet_string_set_to_zero(void *s, size_t len)
|
||||
{
|
||||
#if defined(OPENSSL) && !defined(OPENSSL_CLEANSE_BROKEN)
|
||||
OPENSSL_cleanse(s, len);
|
||||
#else
|
||||
srtp_cleanse(s, len);
|
||||
#endif
|
||||
}
|
||||
|
||||
#ifdef TESTAPP_SOURCE
|
||||
|
||||
static const char b64chars[] = "ABCDEFGHIJKLMNOPQRSTUVWXYZ"
|
||||
"abcdefghijklmnopqrstuvwxyz0123456789+/";
|
||||
|
||||
static int base64_block_to_octet_triple(char *out, char *in)
|
||||
{
|
||||
unsigned char sextets[4] = { 0 };
|
||||
int j = 0;
|
||||
int i;
|
||||
|
||||
for (i = 0; i < 4; i++) {
|
||||
char *p = strchr(b64chars, in[i]);
|
||||
if (p != NULL)
|
||||
sextets[i] = p - b64chars;
|
||||
else
|
||||
j++;
|
||||
}
|
||||
|
||||
out[0] = (sextets[0] << 2) | (sextets[1] >> 4);
|
||||
if (j < 2)
|
||||
out[1] = (sextets[1] << 4) | (sextets[2] >> 2);
|
||||
if (j < 1)
|
||||
out[2] = (sextets[2] << 6) | sextets[3];
|
||||
return j;
|
||||
}
|
||||
|
||||
int base64_string_to_octet_string(char *out, int *pad, char *in, int len)
|
||||
{
|
||||
int k = 0;
|
||||
int i = 0;
|
||||
int j = 0;
|
||||
if (len % 4 != 0)
|
||||
return 0;
|
||||
|
||||
while (i < len && j == 0) {
|
||||
j = base64_block_to_octet_triple(out + k, in + i);
|
||||
k += 3;
|
||||
i += 4;
|
||||
}
|
||||
*pad = j;
|
||||
return i;
|
||||
}
|
||||
|
||||
#endif
|
213
trunk/3rdparty/libsrtp-2-fit/crypto/math/stat.c
vendored
Normal file
213
trunk/3rdparty/libsrtp-2-fit/crypto/math/stat.c
vendored
Normal file
|
@ -0,0 +1,213 @@
|
|||
/*
|
||||
* stats.c
|
||||
*
|
||||
* statistical tests
|
||||
*
|
||||
* David A. McGrew
|
||||
* Cisco Systems, Inc.
|
||||
*/
|
||||
|
||||
/*
|
||||
*
|
||||
* Copyright (c) 2001-2017, Cisco Systems, Inc.
|
||||
* All rights reserved.
|
||||
*
|
||||
* Redistribution and use in source and binary forms, with or without
|
||||
* modification, are permitted provided that the following conditions
|
||||
* are met:
|
||||
*
|
||||
* Redistributions of source code must retain the above copyright
|
||||
* notice, this list of conditions and the following disclaimer.
|
||||
*
|
||||
* 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.
|
||||
*
|
||||
* Neither the name of the Cisco Systems, Inc. nor the names of its
|
||||
* contributors may be used to endorse or promote products derived
|
||||
* from this software without specific prior written permission.
|
||||
*
|
||||
* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS 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
|
||||
* COPYRIGHT HOLDERS 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 HAVE_CONFIG_H
|
||||
#include <config.h>
|
||||
#endif
|
||||
|
||||
#include "stat.h"
|
||||
|
||||
srtp_debug_module_t srtp_mod_stat = {
|
||||
0, /* debugging is off by default */
|
||||
(char *)"stat test" /* printable module name */
|
||||
};
|
||||
|
||||
/*
|
||||
* each test assumes that 20,000 bits (2500 octets) of data is
|
||||
* provided as input
|
||||
*/
|
||||
|
||||
#define STAT_TEST_DATA_LEN 2500
|
||||
|
||||
srtp_err_status_t stat_test_monobit(uint8_t *data)
|
||||
{
|
||||
uint8_t *data_end = data + STAT_TEST_DATA_LEN;
|
||||
uint16_t ones_count;
|
||||
|
||||
ones_count = 0;
|
||||
while (data < data_end) {
|
||||
ones_count += octet_get_weight(*data);
|
||||
data++;
|
||||
}
|
||||
|
||||
debug_print(srtp_mod_stat, "bit count: %d", ones_count);
|
||||
|
||||
if ((ones_count < 9725) || (ones_count > 10275))
|
||||
return srtp_err_status_algo_fail;
|
||||
|
||||
return srtp_err_status_ok;
|
||||
}
|
||||
|
||||
srtp_err_status_t stat_test_poker(uint8_t *data)
|
||||
{
|
||||
int i;
|
||||
uint8_t *data_end = data + STAT_TEST_DATA_LEN;
|
||||
double poker;
|
||||
uint16_t f[16] = { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 };
|
||||
|
||||
while (data < data_end) {
|
||||
f[*data & 0x0f]++; /* increment freq. count for low nibble */
|
||||
f[(*data) >> 4]++; /* increment freq. count for high nibble */
|
||||
data++;
|
||||
}
|
||||
|
||||
poker = 0.0;
|
||||
for (i = 0; i < 16; i++)
|
||||
poker += (double)f[i] * f[i];
|
||||
|
||||
poker *= (16.0 / 5000.0);
|
||||
poker -= 5000.0;
|
||||
|
||||
debug_print(srtp_mod_stat, "poker test: %f\n", poker);
|
||||
|
||||
if ((poker < 2.16) || (poker > 46.17))
|
||||
return srtp_err_status_algo_fail;
|
||||
|
||||
return srtp_err_status_ok;
|
||||
}
|
||||
|
||||
/*
|
||||
* runs[i] holds the number of runs of size (i-1)
|
||||
*/
|
||||
|
||||
srtp_err_status_t stat_test_runs(uint8_t *data)
|
||||
{
|
||||
uint8_t *data_end = data + STAT_TEST_DATA_LEN;
|
||||
uint16_t runs[6] = { 0, 0, 0, 0, 0, 0 };
|
||||
uint16_t gaps[6] = { 0, 0, 0, 0, 0, 0 };
|
||||
uint16_t lo_value[6] = { 2315, 1114, 527, 240, 103, 103 };
|
||||
uint16_t hi_value[6] = { 2685, 1386, 723, 384, 209, 209 };
|
||||
int state = 0;
|
||||
uint16_t mask;
|
||||
int i;
|
||||
|
||||
/*
|
||||
* the state variable holds the number of bits in the
|
||||
* current run (or gap, if negative)
|
||||
*/
|
||||
|
||||
while (data < data_end) {
|
||||
/* loop over the bits of this byte */
|
||||
for (mask = 1; mask < 256; mask <<= 1) {
|
||||
if (*data & mask) {
|
||||
/* next bit is a one */
|
||||
if (state > 0) {
|
||||
/* prefix is a run, so increment the run-count */
|
||||
state++;
|
||||
|
||||
/* check for long runs */
|
||||
if (state > 25) {
|
||||
debug_print(srtp_mod_stat, ">25 runs: %d", state);
|
||||
return srtp_err_status_algo_fail;
|
||||
}
|
||||
|
||||
} else if (state < 0) {
|
||||
/* prefix is a gap */
|
||||
if (state < -25) {
|
||||
debug_print(srtp_mod_stat, ">25 gaps: %d", state);
|
||||
return srtp_err_status_algo_fail; /* long-runs test
|
||||
failed */
|
||||
}
|
||||
if (state < -6) {
|
||||
state = -6; /* group together gaps > 5 */
|
||||
}
|
||||
gaps[-1 - state]++; /* increment gap count */
|
||||
state = 1; /* set state at one set bit */
|
||||
} else {
|
||||
/* state is zero; this happens only at initialization */
|
||||
state = 1;
|
||||
}
|
||||
} else {
|
||||
/* next bit is a zero */
|
||||
if (state > 0) {
|
||||
/* prefix is a run */
|
||||
if (state > 25) {
|
||||
debug_print(srtp_mod_stat, ">25 runs (2): %d", state);
|
||||
return srtp_err_status_algo_fail; /* long-runs test
|
||||
failed */
|
||||
}
|
||||
if (state > 6) {
|
||||
state = 6; /* group together runs > 5 */
|
||||
}
|
||||
runs[state - 1]++; /* increment run count */
|
||||
state = -1; /* set state at one zero bit */
|
||||
} else if (state < 0) {
|
||||
/* prefix is a gap, so increment gap-count (decrement state)
|
||||
*/
|
||||
state--;
|
||||
|
||||
/* check for long gaps */
|
||||
if (state < -25) {
|
||||
debug_print(srtp_mod_stat, ">25 gaps (2): %d", state);
|
||||
return srtp_err_status_algo_fail;
|
||||
}
|
||||
|
||||
} else {
|
||||
/* state is zero; this happens only at initialization */
|
||||
state = -1;
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
/* move along to next octet */
|
||||
data++;
|
||||
}
|
||||
|
||||
if (srtp_mod_stat.on) {
|
||||
debug_print0(srtp_mod_stat, "runs test");
|
||||
for (i = 0; i < 6; i++)
|
||||
debug_print(srtp_mod_stat, " runs[]: %d", runs[i]);
|
||||
for (i = 0; i < 6; i++)
|
||||
debug_print(srtp_mod_stat, " gaps[]: %d", gaps[i]);
|
||||
}
|
||||
|
||||
/* check run and gap counts against the fixed limits */
|
||||
for (i = 0; i < 6; i++)
|
||||
if ((runs[i] < lo_value[i]) || (runs[i] > hi_value[i]) ||
|
||||
(gaps[i] < lo_value[i]) || (gaps[i] > hi_value[i]))
|
||||
return srtp_err_status_algo_fail;
|
||||
|
||||
return srtp_err_status_ok;
|
||||
}
|
Loading…
Add table
Add a link
Reference in a new issue