I've decided to pick it up again, but this time I need help from the community. One problem I am facing is the connection between Java and JNI. You see, the way I display the private and public key in the UI is by having two global variables in my main class/activity on the Java side, when a key is requested JNI takes over and generates a keypair and sets the Java global variables(privkey and pubkey;both are of type String but with no access modifiers) but nothing guarantees they are set(could happen if something went wrong with JNI). Then on the Java side is where I do the regular expression matching. Here lies the problem, in order to actually match them, I must return from JNI to Java, thereby exiting the generating function. Samr7 discovered that it's possible to batch up EC points and do batched modular inversion which is why he achieved such great speeds.
Code:
EC_POINTs_make_affine(pgroup, nbatch, ppnt, vxc_bnctx);
for (i = 0; i < nbatch; i++, vxc_delta++) {
// Hash the public key
len = EC_POINT_point2oct(pgroup, ppnt[i], POINT_CONVERSION_UNCOMPRESSED, eckey_buf, sizeof(eckey_buf), vxc_bnctx);
SHA256(eckey_buf, len, hash1);
RIPEMD160(hash1, sizeof(hash1), &vxc_binres[1]);
// Debug
char address[64];
Base58Encode(vxc_binres,21, address); //Funny, but I do not remember why I skip the 4 byte check code. Duh, for debug purposes :P
__android_log_print(ANDROID_LOG_INFO, "aVanityGen", "%s=%s", bin2hex(vxc_binres,21), address);
npoints = 0;
rekey_at = 0;
i = nbatch;
break; //No point in continuing further.
}
for (i = 0; i < nbatch; i++, vxc_delta++) {
// Hash the public key
len = EC_POINT_point2oct(pgroup, ppnt[i], POINT_CONVERSION_UNCOMPRESSED, eckey_buf, sizeof(eckey_buf), vxc_bnctx);
SHA256(eckey_buf, len, hash1);
RIPEMD160(hash1, sizeof(hash1), &vxc_binres[1]);
// Debug
char address[64];
Base58Encode(vxc_binres,21, address); //Funny, but I do not remember why I skip the 4 byte check code. Duh, for debug purposes :P
__android_log_print(ANDROID_LOG_INFO, "aVanityGen", "%s=%s", bin2hex(vxc_binres,21), address);
npoints = 0;
rekey_at = 0;
i = nbatch;
break; //No point in continuing further.
}
This is the relevant C code. I am still using it in my code despite the fact that it does NOTHING at this point. The problem is, I have no way to return to Java on every loop and check it before the next loop begins. Doing the actual checking in there sounds like a plan, but how? Using PCRE sounded like a nice idea, but it would add a lot of size to the program.
I offer an 100PPcoin bounty for real ideas on how to overcome this. I know 100 ppcoins is not nearly enough, but hey, I didn't see 100 litecoins being a lot at the time as well.
