Optimized CRC implementation - (32-bit ANSI polynomial)

[deepikabhavnani]

Description

Adopted the optimized version of 32-bit ANSI CRC table (0x04c11db7). Saves 960 bytes of flash.

Pull request type

[ ] Fix
[X] Refactor
[ ] Target update
[ ] Feature
[ ] Breaking change

CC @geky

[0xc0170]

This functional change - the main drive is saving ROM, or is there any other benefit we are getting?

[deepikabhavnani]

@0xc0170 - Initial intent was to replace Littlefs util crc implementation with this, but that will need thread safety changes as well. We can mark this as fix. Will have another PR for thread safety and integration.

[geky]

Looks good to me 👍

[geky]

Hmm, so is HARDWARE not defined unless DEVICE_CRC is defined?

[deepikabhavnani]

HARDWARE is set as zero in enum as is available. But it should not be set if 'DEVICE_CRC` is defined.

Other option will be to return error and not 0 as per API.
* @return 0 on success or a negative error code on failure

[geky]

Oh that's a good idea. Though if HARDWARE is not set, trying to use it will be a compile error, which IMO is a slightly better.

[geky]

Looking around a bit, I found this good resource on half-byte CRCs:
https://create.stephan-brumme.com/crc32

They are ~4x slower, but 16x smaller (64B instead of 1KB)

It looks like there is a generator we could use for all CRC tables if we wanted to adopt this approach for all polynomials?

// from https://create.stephan-brumme.com/crc32
for (unsigned int i = 0; i < 16; i++) {
    unsigned int crc = i * 16;
    for (unsigned int j = 0; j < 8; j++)
        crc = (crc >> 1) ^ (-int(crc & 1) & Polynomial);
    lut[i] = crc;
}

Will have another PR for thread safety and integration.

👍

[cmonr]

/morph build

[mbed-ci]

Build : SUCCESS

Build number : 2782
Build artifacts/logs : http://mbed-os.s3-website-eu-west-1.amazonaws.com/?prefix=builds/7735/

Triggering tests

/morph test
/morph uvisor-test
/morph export-build
/morph mbed2-build

[mbed-ci]

Test : SUCCESS

Build number : 2519
Test logs :http://mbed-os-logs.s3-website-us-west-1.amazonaws.com/?prefix=logs/7735/2519

[mbed-ci]

Exporter Build : SUCCESS

Build number : 2413
Build artifacts/logs : http://mbed-os.s3-website-eu-west-1.amazonaws.com/?prefix=builds/exporter/7735/

[adbridge]

@deepikabhavnani What tests have been carried out to ensure that this optimised version has identical results to the original version ?

[cmonr]

@adbridge I would image that's what the CRC tests are for?

[deepikabhavnani]

@adbridge - https://github.com/ARMmbed/mbed-os/blob/master/TESTS/mbed_drivers/crc/main.cpp#L52 is the test case of the polynomial which was optimized

[deepikabhavnani]

Hello.. CRC PR's #7735, #7781, #7793 and #7824 are related and will be better to have them all in one patch. Can we change the tag of merged PR's to next patch