Added crypto callback for SHA3.

[aidangarske]

Description

Added crypto callback for SHA3 and extended the test.c tests for it in cryptocb_test.

Testing

./configure --enable-cryptocb --enable-sha3 CFLAGS="-DDEBUG_CRYPTOCB" --enable-debug
make 
./wolfcrypt/test/testwolfcrypt

[dgarske]

Retest this please

[billphipps]

Looks ok, but I'd like to see the existing pattern of using the p (block count) and len to determine the type instead of modifying and adding a member to track it. Also, there's a bug in InitSha3 that clears the devId, so the end of Final should either invoke the wc_InitSha3(..., sha3->devId) or InitSha3 should be updated to remember the devId.

[billphipps]

In the other cases (async, sw), the p parameter passed into SHA3_Update and SHA3_Final controls which type (256, 384, 512) of SHA3 is being used. Consider letting the the processing functions determine the info->hash->type based on that and remove the type parameter/member here. Comments below.

[billphipps]

Consider:
int type = 0;
switch(len) {
case WC_SHA3_224_DIGEST_SIZE: type = WC_HASH_TYPE_SHA3_224; break;
case WC_SHA3_256_DIGEST_SIZE: type = WC_HASH_TYPE_SHA3_256; break;
case WC_SHA3_384_DIGEST_SIZE: type = WC_HASH_TYPE_SHA3_384; break;
case WC_SHA3_512_DIGEST_SIZE: type = WC_HASH_TYPE_SHA3_512; break;
default: return WC_ERROR_BAD_ARG;
}
ret = wc_CryptoCb_Sha3Hash(sha, type, NULL, 0, hash);

[billphipps]

Consider:
int type = 0;
switch(p) {
case WC_SHA3_224_COUNT: type = WC_HASH_TYPE_SHA3_224; break;
case WC_SHA3_256_COUNT: type = WC_HASH_TYPE_SHA3_256; break;
case WC_SHA3_384_COUNT: type = WC_HASH_TYPE_SHA3_384; break;
case WC_SHA3_512_COUNT: type = WC_HASH_TYPE_SHA3_512; break;
default: return WC_ERROR_BAD_ARG;
}
ret = wc_CryptoCb_Sha3Hash(sha, type, NULL, 0, hash);