Feature Request: Comprehensive Espressif Math Tests

[gojimmypi]

Version

latest master

Description

As noted in #6380, there were some unexpected and incorrect Espressif math results as related to fp_int data provided from HW acceleration and the use (or lack thereof) fully initialized fp_int data structures.

There's also the handling of the used size value for values such as fp_int = 1 in fp_cmp. (e.g. should values of used=1 and used=8 be considered equal? currently they are not)

As noted in the comment, there may be peripheral consequences to the changes I've proposed in #6382.

I will be looking into any undesired side effects. Indeed if other functions have a similar, incorrect handling of the used scope, I could imagine a situation where a test might fail. The test itself may need to be addressed.

In any case, this issue is a Feature Request to add more comprehensive math testing to wolfcrypt/test that would have caught the fp_add and fp_cmp problems for Espressif and any other Hardware Acceleration that may be inserted into known, good math libraries.

[gojimmypi]

I have some additional hardware-specific tests in my test.c that I'll add at some later date.

Closing this lingering stale issue for now.