crc16 algorithm from C++ to bash

crc-16 program in c
crc remainder calculator
crc-16-ccitt c code
crc16 table
c calculate crc16
crc calculator excel
crc-8 c code
network crc calculator

I am trying to implement a CRC16 checksum in bash. I'm porting from an existing piece of C++ code. I'm almost there, but I am getting different answers.

I don't quite see why the checksums between the C++ code and the bash script are different.

Another set of eyes would be a big help.

Here is the C++ code:

uint16_t Encoder::checksum(std::string thestring)
    uint8_t d, e, f;
    uint16_t c, r, crccalc;
    c = 0xffff;

    for (unsigned int i = 0; i < thestring.length(); i++)
        d = thestring[i];
        e = c ^ d;
        f = e ^ (e << 4);
        r = (c >> 8) ^ (f << 8) ^ (f << 3) ^ (f >> 4);
        c = r;
    c ^= 0xffff;
    crccalc = c;
    return crccalc;

And here is my bash code:

function calc_crc16()
    while read -d "" -n 1 ; do astring+=( "$reply" ) ; done <<< "$string"


    for ((x=0;x<$cnt;x++)); do
        e=$(($c ^ $char))
        s=$(($e << 4))
        f=$(($e ^ $s))
        t1=$(($c >> 8))
        t2=$(($f << 8))
        t3=$(($f << 3))
        t4=$(($f >> 4))
        r1=$(($t1 ^ $t2 ^ $t3 ^ $t4))
    c=$c ^ 0xffff
    echo "checksum = $c"

Is it going to have something to do with the size of the ints? I'm guessing there's not much I can do about that in bash.

I'm getting an actual number, but it doesn't match the C++, which I know works correctly. does anyone see anything where I may be screwing things up?

First problem is near the top

while read -d "" -n 1 ; do astring+=( "$reply" ) ; done <<< "$string"

$reply is wrong, since you didn't specify a variable name for read the name is $REPLY.

Next error is at the end

c=$c ^ 0xffff

This should be

c=$(($c ^ 0xffff))

At least this way it will run without errors, correctness and appropriateness is something else.

Correctness issues: What if the input string has a space? This will break horribly. Always quote variable exapnsions





Strangely this rule is different inside $(()) constructs. Your bit operations should reference variables without any $ in these cases

e=$(( c ^ char ))
s=$(( e << 4 ))
f=$(( e ^ s ))
t1=$(( c >> 8 ))
t2=$(( f << 8 ))
t3=$(( f << 3 ))
t4=$(( f >> 4 ))
r1=$(( t1 ^ t2 ^ t3 ^ t4))

And later

c=$(( c ^ 0xffff ))

This will cause variables to be expanded and whitespace not to blow things up.

In general you should also pass -r to read, see help read for what it does.

Why make an extra copy of $1 before processing it into an array? Using

while read -d "" -n 1 ; do astring+=( "$REPLY" ) ; done <<< "$1"

is sufficient.

It's probably not necessary to turn your input into an array before processing. Instead you can slice characters out of the string in your loop, which is closer to what the C++ version is doing. Replace




This is operating directly on the function paramter; since we're no longer making a copy of that we also need to get $cnt a different way


But you really have even bigger problems than this, like the fact that a character isn't an integer in bash. To convert you must use the following syntax:

printf '%d' \'a

where a is the character to convert. Inserting this into the context of the script it would be

char=$(printf '%d' \'"${1:$x:1}")

Now we're getting somewhere, but I really must ask you to consider whether all of this is really worth it. Even if you can make it work, what do you gain?

crc-16.c, #define POLY 0x8408 /* // 16 12 5 // this is the CCITT CRC 16 polynomial X + X + X + 1. // This works out to be 0x1021, but the way the algorithm works // lets us� Hello, I am unable to recreate a CRC vi to calculate the checksum. The procedure mentions calculation as follows:- The polynomial is 0x8005. The initial register value should be 0x0000. After the last bit of theCount and packet has been transmitted, the internal CRC Register should have a value that matches that in theblock. The first Checksum byte transmitted (N-2) is the most-significant

just for future reference, here is the awk script that i came up with.

this works just as fast as the C++ code i have, which is basically instantaneous. the bash takes about 10 seconds to run for the same string. the awk is much faster.

function awk_calc_crc16()
    output=$(echo $1 | awk 'function ord(c){return chmap[c];}
    BEGIN{c=65535; for (i=0; i < 256; i++){ chmap[sprintf("%c", i)] = i;}}
        split($0, chars, "");
        for(i = 1; i <= length(chars); i++)
            e=and(xor(c, ord(chars[i])), 0x00FF);
            s=and(lshift(e, 4), 0x00FF);
            f=and(xor(e, s), 0x00FF);
            r=xor(xor(xor(rshift(c, 8), lshift(f, 8)), lshift(f, 3)), rshift(f, 4));
    END{c=xor(c, 0xFFFF); printf("%hu", c);}')
    echo $output;

CRC-16 Calculation, This code is the function that calculates a CRC-16 for different purposes. crc16( unsigned int crcValue, unsigned char newByte); /***** crc16.c� A cyclic redundancy check (CRC) is an error-detecting code commonly used in digital networks and storage devices to detect accidental changes to raw data. Blocks of data entering these systems get a short check value attached, based on the remainder of a polynomial division of their contents.

ok. with Sorpigal's help, i've got a working version.

i suspect this could all be done within an awk script, which may run a lot faster. i may try that next.

thank you all for the help. i don't mean to steal the solution here, but i worked on it, and figure it is worth putting up.

anyway, here is a working version:

function calc_crc16()
    while read -r -d "" -n 1 ; do astring+=( "$REPLY" ) ; done <<< "$1"


    for ((x=0;x<$cnt;x++)); do
        char=$(printf '%d' \'"${1:$x:1}")
        e=$(((c ^ char) & 0x00FF))
        s=$(((e << 4) & 0x00FF))
        f=$(((e ^ s) & 0x00FF))
        r1=$(((c >> 8) ^ (f << 8) ^ (f << 3) ^ (f >> 4)))
    c=$((c ^ 0xffff))
    echo "checksum = $c"

src/crc16.c, Knowing that all CRC algorithms are simply long division algorithms in The divisor is a c+1-bit number known as the generator polynomial. #define CRC16_T10_DIF 0x8BB7 // SCSI DIF #define CRC16_DECT 0x0589 // Cordeless Telephones #define CRC16_ARINC 0xA02B // ACARS Aplications #define POLYNOM CRC16_XXX // Define the used polynom from one of the aboves // It calculates the new crc16 with the newByte. Variable crcValue is the actual or initial value (0).

shorter (and faster) version changes: - the while loop at the beginning is not needed - r1 is not needed - cnt is not needed - use capital letters for bash variables - one less "\" (backslash) in char=$(printf … - removed leading 0s for 0xFF

function calc_crc16() {

  for ((X=0; X<${#1}; X++)); do
    CHAR=$(printf '%d' "'${1:$X:1}")
    E=$(((CRC ^ CHAR) & 0xFF))
    S=$(((E << 4)     & 0xFF))
    F=$(((E ^ S)      & 0xFF))
    CRC=$(((CRC >> 8) ^ (F << 8) ^ (F << 3) ^ (F >> 4)))

  let CRC^=0xFFFF
  printf "0x%X\n" $CRC

CRC Series, Part 3: CRC Implementation Code in C/C++, Simple Example. The following example program shows how to compute a CRC byte in the C language. The outer loop processes each byte, and the inner loop� $ gcc -O2 -Wall -Wextra -ansi -pedantic MODBUS_CRC16.c -o MODBUS_CRC16 Algorithm 1: static uint16_t MODBUS_CRC16_v1 ( const unsigned char *buf, unsigned int len )

6.7.6. CRC Computation in C, LibCRC is an MIT licensed library written in C containing various checksum algorithms. These include the most common CRC implementations but also other � Implementation of CRC16-CCITT using polynomial 0x1021 = x^16 + x^12 + x^5 + 1. This site uses cookies to store information on your computer. By continuing to use our site, you consent to our cookies.

LibCRC – Open Source CRC Library in C – Checksum calculation , A C-language program is included which produces CRC values which conform to of “A Painless Guide to CRC Error Detection Algorithms” by Ross Williams. The byte order of the result of the calculation now matches the byte order used by the System.BitConverter class. Added CrcTest class to verify the correctness of algorithm calculations. The Crc class IsRight method now does not throw an exception if the calculated value is different from the test value.

CRC-CCITT -- 16-bit, Solved: Hello, I am unable to recreate a CRC vi to calculate the checksum. The procedure mentions calculation as follows:- The polynomial is� The purpose of these two c-bit constants is similar to the final bit inversion step added to the sum-of-bytes checksum algorithm. Each of these parameters helps eliminate one very special, though perhaps not uncommon, class of ordinarily undetectable difference. In effect, they bulletproof an already strong checksum algorithm.

  • write a quick test to see what max int is in C++ vs bash. Good luck.
  • I think the biggest question I have here is why
  • thank you. i fixed this. i am getting an error with the "e=$(($c ^ $char))" line, because $char is the '$' character (from the string i am passing to it). is there a way to keep it from throwing this error ? bash seems to be interpreting the string in an odd way.
  • @jasonmclose: See my edit for a few words on this and some other things.
  • i know the premise seems odd. but i am working on some code that is sending data over a serial line, and i need to checksum the data using a specific CRC. i don't control the receiving end code, so my checksum has to match. if i can keep this in a script, we can better manage things in the field (dunno if we will even have gcc on these boxes). i have a crc executable that works, and can fall back on it if need be. for the most part, this is a good bash lesson for me.