mirror of
https://github.com/cryb-to/cryb-to.git
synced 2024-11-25 15:15:42 +00:00
319 lines
7.5 KiB
C
319 lines
7.5 KiB
C
/*-
|
||
* Copyright (c) 2015-2016 Dag-Erling Smørgrav
|
||
* 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.
|
||
* 3. The name of the author may not be used to endorse or promote
|
||
* products derived from this software without specific prior written
|
||
* permission.
|
||
*
|
||
* 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 "cryb/impl.h"
|
||
|
||
#include <sys/types.h>
|
||
|
||
#include <stdint.h>
|
||
#include <stdio.h>
|
||
|
||
#include <cryb/endian.h>
|
||
#include <cryb/hash.h>
|
||
|
||
#include <cryb/test.h>
|
||
|
||
#define FLETCHER16_INVALID 0xffffU
|
||
#define FLETCHER32_INVALID 0xffffffffLU
|
||
#define FLETCHER64_INVALID 0xffffffffffffffffLLU
|
||
|
||
static uint8_t ones8[65536];
|
||
static uint16_t ones16[65536];
|
||
static uint32_t ones32[65536];
|
||
|
||
struct t_case {
|
||
const char *desc;
|
||
const void *data;
|
||
size_t len;
|
||
uint16_t sum16;
|
||
uint32_t sum32;
|
||
uint64_t sum64;
|
||
};
|
||
|
||
/***************************************************************************
|
||
* Test cases
|
||
*
|
||
* Most of these test cases use input sequences where each word has the
|
||
* value 1. To understand the test cases, it helps to realize that (until
|
||
* it overflows) the low word of the Fletcher checksum of such a sequence
|
||
* of length n is equal to n, while the high word is equal to the nth
|
||
* triangular number, i.e. (n*(n+1))/2+n. Also, remember that the input
|
||
* width for Fletcher is half the checksum width.
|
||
*/
|
||
#define TRI(n) (n * (n + 1) / 2)
|
||
static struct t_case t_cases[] = {
|
||
/* trivial cases */
|
||
{
|
||
.desc = "all-zeroes message",
|
||
.data = t_zero,
|
||
.len = sizeof t_zero,
|
||
.sum16 = 0x00,
|
||
.sum32 = 0x0000,
|
||
.sum64 = 0x00000000,
|
||
},
|
||
{
|
||
.desc = "length 0",
|
||
.data = ones8,
|
||
.len = 0,
|
||
.sum16 = 0x00,
|
||
.sum32 = 0x0000,
|
||
.sum64 = 0x00000000,
|
||
},
|
||
|
||
/* padding */
|
||
{
|
||
.desc = "length 1",
|
||
.data = ones8,
|
||
.len = 1,
|
||
.sum16 = 0x0101,
|
||
.sum32 = 0x01000100,
|
||
.sum64 = 0x0100000001000000,
|
||
},
|
||
{
|
||
.desc = "length 2",
|
||
.data = ones8,
|
||
.len = 2,
|
||
.sum16 = 0x0302,
|
||
.sum32 = 0x01010101,
|
||
.sum64 = 0x0101000001010000,
|
||
},
|
||
{
|
||
.desc = "length 3",
|
||
.data = ones8,
|
||
.len = 3,
|
||
.sum16 = 0x0603,
|
||
.sum32 = 0x03020201,
|
||
.sum64 = 0x0101010001010100,
|
||
},
|
||
{
|
||
.desc = "length 4",
|
||
.data = ones8,
|
||
.len = 4,
|
||
.sum16 = 0x0a04,
|
||
.sum32 = 0x03030202,
|
||
.sum64 = 0x0101010101010101,
|
||
},
|
||
{
|
||
.desc = "length 5",
|
||
.data = ones8,
|
||
.len = 5,
|
||
.sum16 = 0x0f05,
|
||
.sum32 = 0x06050302,
|
||
.sum64 = 0x0302020202010101,
|
||
},
|
||
{
|
||
.desc = "length 6",
|
||
.data = ones8,
|
||
.len = 6,
|
||
.sum16 = 0x1506,
|
||
.sum32 = 0x06060303,
|
||
.sum64 = 0x0303020202020101,
|
||
},
|
||
{
|
||
.desc = "length 7",
|
||
.data = ones8,
|
||
.len = 7,
|
||
.sum16 = 0x1c07,
|
||
.sum32 = 0x0a090403,
|
||
.sum64 = 0x0303030202020201,
|
||
},
|
||
{
|
||
.desc = "length 8",
|
||
.data = ones8,
|
||
.len = 8,
|
||
.sum16 = 0x2408,
|
||
.sum32 = 0x0a0a0404,
|
||
.sum64 = 0x0303030302020202,
|
||
},
|
||
|
||
/* overflow (16-bit checksum) */
|
||
{
|
||
.desc = "high overflow - 1",
|
||
.data = ones8,
|
||
.len = 22,
|
||
.sum16 = (TRI(22) % 255) << 8 | 22,
|
||
.sum32 = FLETCHER32_INVALID,
|
||
.sum64 = FLETCHER64_INVALID,
|
||
},
|
||
{
|
||
.desc = "high overflow",
|
||
.data = ones8,
|
||
.len = 23,
|
||
.sum16 = (TRI(23) % 255) << 8 | 23,
|
||
.sum32 = FLETCHER32_INVALID,
|
||
.sum64 = FLETCHER64_INVALID,
|
||
},
|
||
{
|
||
.desc = "low overflow - 1",
|
||
.data = ones8,
|
||
.len = 254,
|
||
.sum16 = (TRI(254) % 255) << 8 | 254,
|
||
.sum32 = FLETCHER32_INVALID,
|
||
.sum64 = FLETCHER64_INVALID,
|
||
},
|
||
{
|
||
.desc = "low overflow",
|
||
.data = ones8,
|
||
.len = 255,
|
||
.sum16 = (TRI(254) % 255) << 8 | 0,
|
||
.sum32 = FLETCHER32_INVALID,
|
||
.sum64 = FLETCHER64_INVALID,
|
||
},
|
||
{
|
||
.desc = "low overflow + 1",
|
||
.data = ones8,
|
||
.len = 256,
|
||
.sum16 = ((TRI(254) + 1) % 255) << 8 | 1,
|
||
.sum32 = FLETCHER32_INVALID,
|
||
.sum64 = FLETCHER64_INVALID,
|
||
},
|
||
|
||
/* overflow (32-bit checksum) */
|
||
{
|
||
.desc = "high overflow - 1",
|
||
.data = ones16,
|
||
.len = 362 * sizeof ones16[0],
|
||
.sum16 = FLETCHER16_INVALID,
|
||
.sum32 = (TRI(362U) % 65535U) << 16 | 362U,
|
||
.sum64 = FLETCHER64_INVALID,
|
||
},
|
||
{
|
||
.desc = "high overflow",
|
||
.data = ones16,
|
||
.len = 363 * sizeof ones16[0],
|
||
.sum16 = FLETCHER16_INVALID,
|
||
.sum32 = (TRI(363U) % 65535U) << 16 | 363U,
|
||
.sum64 = FLETCHER64_INVALID,
|
||
},
|
||
{
|
||
.desc = "low overflow - 1",
|
||
.data = ones16,
|
||
.len = 65534 * sizeof ones16[0],
|
||
.sum16 = FLETCHER16_INVALID,
|
||
.sum32 = (TRI(65534U) % 65535U) << 16 | 65534U,
|
||
.sum64 = FLETCHER64_INVALID,
|
||
},
|
||
{
|
||
.desc = "low overflow",
|
||
.data = ones16,
|
||
.len = 65535 * sizeof ones16[0],
|
||
.sum16 = FLETCHER16_INVALID,
|
||
.sum32 = (TRI(65534U) % 65535U) << 16 | 0U,
|
||
.sum64 = FLETCHER64_INVALID,
|
||
},
|
||
{
|
||
.desc = "low overflow + 1",
|
||
.data = ones16,
|
||
.len = 65536 * sizeof ones16[0],
|
||
.sum16 = FLETCHER16_INVALID,
|
||
.sum32 = ((TRI(65534U) + 1) % 65535U) << 16 | 1U,
|
||
.sum64 = FLETCHER64_INVALID,
|
||
},
|
||
|
||
/* overflow (64-bit checksum */
|
||
/*
|
||
* This is currently not realistic. We need to change the
|
||
* fletcher*() API to take a seed so it can do partial /
|
||
* incremental checksums, and provide a precomputed seed just
|
||
* below the overflow point.
|
||
*/
|
||
};
|
||
|
||
/***************************************************************************
|
||
* Test function
|
||
*/
|
||
static int
|
||
t_fletcher16(char **desc, void *arg)
|
||
{
|
||
struct t_case *t = arg;
|
||
|
||
(void)asprintf(desc, "(16-bit) %s", t->desc);
|
||
return (t_compare_x16(t->sum16, fletcher16_hash(t->data, t->len)));
|
||
}
|
||
|
||
static int
|
||
t_fletcher32(char **desc, void *arg)
|
||
{
|
||
struct t_case *t = arg;
|
||
|
||
(void)asprintf(desc, "(32-bit) %s", t->desc);
|
||
return (t_compare_x32(t->sum32, fletcher32_hash(t->data, t->len)));
|
||
}
|
||
|
||
static int
|
||
t_fletcher64(char **desc, void *arg)
|
||
{
|
||
struct t_case *t = arg;
|
||
|
||
(void)asprintf(desc, "(64-bit) %s", t->desc);
|
||
return (t_compare_x64(t->sum64, fletcher64_hash(t->data, t->len)));
|
||
}
|
||
|
||
|
||
/***************************************************************************
|
||
* Boilerplate
|
||
*/
|
||
|
||
static int
|
||
t_prepare(int argc, char *argv[])
|
||
{
|
||
unsigned int i, n;
|
||
|
||
(void)argc;
|
||
(void)argv;
|
||
ones8[0] = 1;
|
||
be16enc(&ones16[0], 1);
|
||
be32enc(&ones32[0], 1);
|
||
for (i = 1; i < 65536; ++i) {
|
||
ones8[i] = ones8[0];
|
||
ones16[i] = ones16[0];
|
||
ones32[i] = ones32[0];
|
||
}
|
||
n = sizeof t_cases / sizeof t_cases[0];
|
||
for (i = 0; i < n; ++i)
|
||
if (t_cases[i].sum16 != 0xffffU)
|
||
t_add_test(t_fletcher16, &t_cases[i], "%s",
|
||
t_cases[i].desc);
|
||
for (i = 0; i < n; ++i)
|
||
if (t_cases[i].sum32 != 0xffffffffLU)
|
||
t_add_test(t_fletcher32, &t_cases[i], "%s",
|
||
t_cases[i].desc);
|
||
for (i = 0; i < n; ++i)
|
||
if (t_cases[i].sum64 != 0xffffffffffffffffLLU)
|
||
t_add_test(t_fletcher64, &t_cases[i], "%s",
|
||
t_cases[i].desc);
|
||
return (0);
|
||
}
|
||
|
||
int
|
||
main(int argc, char *argv[])
|
||
{
|
||
|
||
t_main(t_prepare, NULL, argc, argv);
|
||
}
|