Refactor wc_rng_new to use wc_rng_new_ex, and to use WC_USE_DEVID as the devId if set at compile time #9312
Conversation
bigbrett
left a comment
There was a problem hiding this comment.
@night1rider this fixes the RNG test for me but now is revealing a few other related problems that make sense to fix in this PR
test.c:47429
WOLFSSL_TEST_SUBROUTINE wc_test_ret_t dilithium_test(void)
{
wc_test_ret_t ret;
WC_RNG rng;
#ifndef HAVE_FIPS
ret = wc_InitRng_ex(&rng, HEAP_HINT, INVALID_DEVID); // <--------- Needs global devId
#else
ret = wc_InitRng(&rng);
#endif
if (ret != 0) {
ret = WC_TEST_RET_ENC_EC(ret);
return ret;
}
....
|
@bigbrett Updated the 4 tests that where initializing RNG without the global devId. I see a bunch of other tests that use I do see this failing, unsure if its due to be running on an embedded target do you see a similar failure? |
|
@night1rider yes I'm on an embedded target. The tests wouldn't fail unless you were building with |
|
@night1rider brings up a good point...
@dgarske larger question - do you think it is ever appropriate for our unit tests to manually specify IMO if user sets |
61b7216 to
065899a
Compare
…the devId if set at compile time
…um_test, xmss_test, lms_test
065899a to
bd4099d
Compare
|
Rebased to use fixed kernel test in github actions |
Found that when doing our RNG test it would fail to use my CryptoCallback. Noticed the test would use
wc_rng_newand notwc_rng_new_exand not supply my default devId which is set withWC_USE_DEVID.Change is to assume that if
WC_USE_DEVIDis set that the user intends this to be the default device to use for crypto calls, so this means that ifwc_rng_newis called it should be assumed the user intends it to useWC_USE_DEVIDand notINVALID_DEVIDwhich is our default solution.INVALID_DEVIDcan still be request if used withwc_rng_new_exby the user