如何将表驱动的CRC实现转换为按位实现？

Question

我有一些我需要测试的CRC方法.这些方法不是我自己写的.我能看到的最好方法是使用不同的技术实现CRC算法,然后比较结果.

目前,所有CRC例程都是以表驱动格式编写的.我似乎无法找到任何教程解释如何将代码转换为按位实现.我确信这是我对CRC的工作方式缺乏了解.

如果某人可以提供相同CRC例程的示例,则可以采用两种不同的方式实现.我在互联网上搜索过,并没有找到太多帮助我的问题.

以下是我需要测试的一种CRC算法的示例代码:

// This table implements the EPO CRC16 algorithm with a polynomial of 0x17531
static const uint16 s_EPOCRC16Table[] =
{
   0x0000U, 0x7531U, 0xEA62U, 0x9F53U, 0xA1F5U, 0xD4C4U, 0x4B97U, 0x3EA6U,
   0x36DBU, 0x43EAU, 0xDCB9U, 0xA988U, 0x972EU, 0xE21FU, 0x7D4CU, 0x087DU,
   0x6DB6U, 0x1887U, 0x87D4U, 0xF2E5U, 0xCC43U, 0xB972U, 0x2621U, 0x5310U,
   0x5B6DU, 0x2E5CU, 0xB10FU, 0xC43EU, 0xFA98U, 0x8FA9U, 0x10FAU, 0x65CBU,
   0xDB6CU, 0xAE5DU, 0x310EU, 0x443FU, 0x7A99U, 0x0FA8U, 0x90FBU, 0xE5CAU,
   0xEDB7U, 0x9886U, 0x07D5U, 0x72E4U, 0x4C42U, 0x3973U, 0xA620U, 0xD311U,
   0xB6DAU, 0xC3EBU, 0x5CB8U, 0x2989U, 0x172FU, 0x621EU, 0xFD4DU, 0x887CU,
   0x8001U, 0xF530U, 0x6A63U, 0x1F52U, 0x21F4U, 0x54C5U, 0xCB96U, 0xBEA7U,
   0xC3E9U, 0xB6D8U, 0x298BU, 0x5CBAU, 0x621CU, 0x172DU, 0x887EU, 0xFD4FU,
   0xF532U, 0x8003U, 0x1F50U, 0x6A61U, 0x54C7U, 0x21F6U, 0xBEA5U, 0xCB94U,
   0xAE5FU, 0xDB6EU, 0x443DU, 0x310CU, 0x0FAAU, 0x7A9BU, 0xE5C8U, 0x90F9U,
   0x9884U, 0xEDB5U, 0x72E6U, 0x07D7U, 0x3971U, 0x4C40U, 0xD313U, 0xA622U,
   0x1885U, 0x6DB4U, 0xF2E7U, 0x87D6U, 0xB970U, 0xCC41U, 0x5312U, 0x2623U,
   0x2E5EU, 0x5B6FU, 0xC43CU, 0xB10DU, 0x8FABU, 0xFA9AU, 0x65C9U, 0x10F8U,
   0x7533U, 0x0002U, 0x9F51U, 0xEA60U, 0xD4C6U, 0xA1F7U, 0x3EA4U, 0x4B95U,
   0x43E8U, 0x36D9U, 0xA98AU, 0xDCBBU, 0xE21DU, 0x972CU, 0x087FU, 0x7D4EU,
   0xF2E3U, 0x87D2U, 0x1881U, 0x6DB0U, 0x5316U, 0x2627U, 0xB974U, 0xCC45U,
   0xC438U, 0xB109U, 0x2E5AU, 0x5B6BU, 0x65CDU, 0x10FCU, 0x8FAFU, 0xFA9EU,
   0x9F55U, 0xEA64U, 0x7537U, 0x0006U, 0x3EA0U, 0x4B91U, 0xD4C2U, 0xA1F3U,
   0xA98EU, 0xDCBFU, 0x43ECU, 0x36DDU, 0x087BU, 0x7D4AU, 0xE219U, 0x9728U,
   0x298FU, 0x5CBEU, 0xC3EDU, 0xB6DCU, 0x887AU, 0xFD4BU, 0x6218U, 0x1729U,
   0x1F54U, 0x6A65U, 0xF536U, 0x8007U, 0xBEA1U, 0xCB90U, 0x54C3U, 0x21F2U,
   0x4439U, 0x3108U, 0xAE5BU, 0xDB6AU, 0xE5CCU, 0x90FDU, 0x0FAEU, 0x7A9FU,
   0x72E2U, 0x07D3U, 0x9880U, 0xEDB1U, 0xD317U, 0xA626U, 0x3975U, 0x4C44U,
   0x310AU, 0x443BU, 0xDB68U, 0xAE59U, 0x90FFU, 0xE5CEU, 0x7A9DU, 0x0FACU,
   0x07D1U, 0x72E0U, 0xEDB3U, 0x9882U, 0xA624U, 0xD315U, 0x4C46U, 0x3977U,
   0x5CBCU, 0x298DU, 0xB6DEU, 0xC3EFU, 0xFD49U, 0x8878U, 0x172BU, 0x621AU,
   0x6A67U, 0x1F56U, 0x8005U, 0xF534U, 0xCB92U, 0xBEA3U, 0x21F0U, 0x54C1U,
   0xEA66U, 0x9F57U, 0x0004U, 0x7535U, 0x4B93U, 0x3EA2U, 0xA1F1U, 0xD4C0U,
   0xDCBDU, 0xA98CU, 0x36DFU, 0x43EEU, 0x7D48U, 0x0879U, 0x972AU, 0xE21BU,
   0x87D0U, 0xF2E1U, 0x6DB2U, 0x1883U, 0x2625U, 0x5314U, 0xCC47U, 0xB976U,
   0xB10BU, 0xC43AU, 0x5B69U, 0x2E58U, 0x10FEU, 0x65CFU, 0xFA9CU, 0x8FADU,
};

/**
 * This routine returns the CRC-16 of Cp_ptr.
 * This CRC routine is reversed to match the routine used in the EPO.
 *
 * @param  const uint8 *Cp_ptr - data pointer to the bit string to be calculated over
 * @param  const uint32 Len    - the number of bytes pointed to by Cp_ptr
 * @return the calculated 16-bit CRC
 * @ureq
 *    -# epoCRC16() shall calculate a 16-bit CRC on a block of data starting at an address of a specified length.
 *    -# epoCRC16() shall ASSERT if the passed in pointer is NULL.
 */
uint16 epoCRC16( const uint8 *Cp_ptr, const uint32 Len )
{
   ASSERT( Cp_ptr != NULL );

   // local variable
   uint16 crc = 0x0;

   // calculate the crc
   for ( uint16 i = Len; i > 0; i-- )
   {
      const uint8 TableIndex = ( *Cp_ptr ^ static_cast<uint8>( crc >> 8 ) );
      crc = static_cast<uint16>( crc << 8 );
      crc = ( crc ^ s_EPOCRC16Table[TableIndex] );
      Cp_ptr++;
   }

   return crc;
}

Answer 1

您首先需要CRC多项式,您可以在此类表的第二个条目或第128个条目中找到它,具体取决于CRC的位顺序.或者在这种情况下,您可以在评论中找到它0x17531.在1之后的x有多项式,它简单地定义了CRC的长度的第一项.它在计算中被删除.您可以0x7531在表格的第二个条目中找到.

此例程(在C中)仅使用多项式生成上表:

void make_crc_table(unsigned poly, unsigned *table)
{
    unsigned n, k, crc;

    for (n = 0; n < 256; n++) {
        crc = n << 8;
        for (k = 0; k < 8; k++)
            crc = crc & 0x8000 ? (crc << 1) ^ poly : crc << 1;
        table[n] = crc & 0xffff;
    }
}

这将被称为:make_crc_table(0x7531, table)where table是256个无符号整数的数组.

逐位CRC例程看起来大致相同:

unsigned epobit(unsigned char *buf, unsigned len, unsigned poly)
{
    unsigned n, k, crc = 0;

    for (n = len; n > 0; n--) {
        crc ^= (*buf++ << 8);
        for (k = 0; k < 8; k++)
            crc = crc & 0x8000 ? (crc << 1) ^ poly : crc << 1;
    }
    return crc & 0xffff;
}