Hello,

I've only used assembler a few times. Therefore, I don't understand the structure of the can.s file and what different parts of the code do.
Could someone enlighten me?

Thanks

I am puzzled that the CAN stack uses a soft implementation of the CRC, rather than the CRC instruction - which should save a considerable amount of time. I was going to try this in a forked repo, but the current CAN implementation does not have any self-testing harness so I have held off for a bit.

Just to capture my thoughts:

The polynomial and bitstrings have to be reversed, after that it is plain sailing (some code attached that computes a crc over a 26 bit packet that includes 15-bit CRC and shows that a soft and hard CRC end up with the same '0000' indicating that the packet is correct)

On a normal packet:

• You use the first 9 bits of the ID to initialise the CRC (reverse order).
• You then fold in the 8 bits comprising two bits of the ID, the IDE, RTR, and r0 bits, and three bits of the DLC
• You then fold in every time you have received 7 bits of a byte.
• You then fold in every time you have received 7 bits of the CRC.
• You then fold in every time you have received the final 8 bits of the CRC, and bob is your uncle.

On an extended packet:

• You use the ID, SSR, IDE, and the first 16 bits of identifier B to initialise the CRC (reverse order).
• You then fold in the 8 bits comprising two bits of the ID, RTR, r0, and r1 bits, and three bits of the DLC
  (at this point you are in sync with the normal packet)

Code::

include <stdio.h>

include <xs1.h>

include <xclib.h>

int crc15(int nxtBit, unsigned int crc_rg) {
#pragma xta label "excludeCrc"
int crc_nxt = (nxtBit ^ (crc_rg >> 14)) & 0x1;
crc_rg = crc_rg << 1;
if (crc_nxt) {
crc_rg = crc_rg ^ 0x4599;
}
return crc_rg;
}

char bitstring[] = {
1,0,0, 1,1,0,1,0,0,1,0,
1,1,0,1,1,0,0,1,1,0,1,1,0,0,1
};

int main(void) {
unsigned int cc = 0;
int bits;
int poly = 0x4cd1;
for(int i = 0; i < 26; i++) {
cc = crc15(bitstring[i], cc);
}
printf("%04x\n", cc&0x7fff);
cc = 0;
bits = 0;
for(int i = 0; i < 2; i++) {
bits = bits << 1 | bitstring[i];
}
cc = bitrev(bits);
for(int j = 0; j < 3; j++) {
bits = 0;
for(int i = 2; i < 10; i++) {
bits = bits << 1 | bitstring[i+j*8];
}
crc8shr(cc, bitrev(bits)>>24, poly);
}
crc32(cc, 0, poly);
printf("%04x\n", bitrev(cc)>>17&0x7fff);
}