danicoin/src/ringct/rctSigs.cpp

930 lines
38 KiB
C++
Raw Normal View History

// Copyright (c) 2016, Monero Research Labs
//
// Author: Shen Noether <shen.noether@gmx.com>
//
// All rights reserved.
//
// Redistribution and use in source and binary forms, with or without modification, are
// permitted provided that the following conditions are met:
//
// 1. Redistributions of source code must retain the above copyright notice, this list of
// conditions and the following disclaimer.
//
// 2. Redistributions in binary form must reproduce the above copyright notice, this list
// of conditions and the following disclaimer in the documentation and/or other
// materials provided with the distribution.
//
// 3. Neither the name of the copyright holder nor the names of its contributors may be
// used to endorse or promote products derived from this software without specific
// prior written permission.
//
// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND ANY
// EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
// MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL
// THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
// PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
// INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT,
// STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF
// THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
#include "misc_log_ex.h"
#include "common/perf_timer.h"
#include "common/task_region.h"
#include "common/thread_group.h"
2016-10-15 12:31:16 +00:00
#include "common/util.h"
#include "rctSigs.h"
#include "cryptonote_basic/cryptonote_format_utils.h"
using namespace crypto;
using namespace std;
Change logging to easylogging++ This replaces the epee and data_loggers logging systems with a single one, and also adds filename:line and explicit severity levels. Categories may be defined, and logging severity set by category (or set of categories). epee style 0-4 log level maps to a sensible severity configuration. Log files now also rotate when reaching 100 MB. To select which logs to output, use the MONERO_LOGS environment variable, with a comma separated list of categories (globs are supported), with their requested severity level after a colon. If a log matches more than one such setting, the last one in the configuration string applies. A few examples: This one is (mostly) silent, only outputting fatal errors: MONERO_LOGS=*:FATAL This one is very verbose: MONERO_LOGS=*:TRACE This one is totally silent (logwise): MONERO_LOGS="" This one outputs all errors and warnings, except for the "verify" category, which prints just fatal errors (the verify category is used for logs about incoming transactions and blocks, and it is expected that some/many will fail to verify, hence we don't want the spam): MONERO_LOGS=*:WARNING,verify:FATAL Log levels are, in decreasing order of priority: FATAL, ERROR, WARNING, INFO, DEBUG, TRACE Subcategories may be added using prefixes and globs. This example will output net.p2p logs at the TRACE level, but all other net* logs only at INFO: MONERO_LOGS=*:ERROR,net*:INFO,net.p2p:TRACE Logs which are intended for the user (which Monero was using a lot through epee, but really isn't a nice way to go things) should use the "global" category. There are a few helper macros for using this category, eg: MGINFO("this shows up by default") or MGINFO_RED("this is red"), to try to keep a similar look and feel for now. Existing epee log macros still exist, and map to the new log levels, but since they're used as a "user facing" UI element as much as a logging system, they often don't map well to log severities (ie, a log level 0 log may be an error, or may be something we want the user to see, such as an important info). In those cases, I tried to use the new macros. In other cases, I left the existing macros in. When modifying logs, it is probably best to switch to the new macros with explicit levels. The --log-level options and set_log commands now also accept category settings, in addition to the epee style log levels.
2017-01-01 16:34:23 +00:00
#undef MONERO_DEFAULT_LOG_CATEGORY
#define MONERO_DEFAULT_LOG_CATEGORY "ringct"
namespace rct {
2016-11-17 23:17:21 +00:00
//Borromean (c.f. gmax/andytoshi's paper)
boroSig genBorromean(const key64 x, const key64 P1, const key64 P2, const bits indices) {
key64 L[2], alpha;
key c;
2016-11-17 23:17:21 +00:00
int naught = 0, prime = 0, ii = 0, jj=0;
boroSig bb;
2016-11-17 23:17:21 +00:00
for (ii = 0 ; ii < 64 ; ii++) {
naught = indices[ii]; prime = (indices[ii] + 1) % 2;
skGen(alpha[ii]);
scalarmultBase(L[naught][ii], alpha[ii]);
if (naught == 0) {
skGen(bb.s1[ii]);
c = hash_to_scalar(L[naught][ii]);
addKeys2(L[prime][ii], bb.s1[ii], c, P2[ii]);
}
}
bb.ee = hash_to_scalar(L[1]); //or L[1]..
2016-11-17 23:17:21 +00:00
key LL, cc;
for (jj = 0 ; jj < 64 ; jj++) {
if (!indices[jj]) {
sc_mulsub(bb.s0[jj].bytes, x[jj].bytes, bb.ee.bytes, alpha[jj].bytes);
} else {
skGen(bb.s0[jj]);
addKeys2(LL, bb.s0[jj], bb.ee, P1[jj]); //different L0
2016-11-17 23:17:21 +00:00
cc = hash_to_scalar(LL);
sc_mulsub(bb.s1[jj].bytes, x[jj].bytes, cc.bytes, alpha[jj].bytes);
}
}
2016-11-17 23:17:21 +00:00
return bb;
}
2016-11-17 23:17:21 +00:00
//see above.
bool verifyBorromean(const boroSig &bb, const key64 P1, const key64 P2) {
key64 Lv1; key chash, LL;
2016-11-17 23:17:21 +00:00
int ii = 0;
for (ii = 0 ; ii < 64 ; ii++) {
addKeys2(LL, bb.s0[ii], bb.ee, P1[ii]);
chash = hash_to_scalar(LL);
addKeys2(Lv1[ii], bb.s1[ii], chash, P2[ii]);
2016-11-17 23:17:21 +00:00
}
key eeComputed = hash_to_scalar(Lv1); //hash function fine
2016-11-17 23:17:21 +00:00
return equalKeys(eeComputed, bb.ee);
}
//Multilayered Spontaneous Anonymous Group Signatures (MLSAG signatures)
//These are aka MG signatutes in earlier drafts of the ring ct paper
// c.f. http://eprint.iacr.org/2015/1098 section 2.
// keyImageV just does I[i] = xx[i] * Hash(xx[i] * G) for each i
// Gen creates a signature which proves that for some column in the keymatrix "pk"
// the signer knows a secret key for each row in that column
// Ver verifies that the MG sig was created correctly
keyV keyImageV(const keyV &xx) {
keyV II(xx.size());
size_t i = 0;
for (i = 0; i < xx.size(); i++) {
II[i] = scalarmultKey(hashToPoint(scalarmultBase(xx[i])), xx[i]);
}
return II;
}
//Multilayered Spontaneous Anonymous Group Signatures (MLSAG signatures)
//This is a just slghtly more efficient version than the ones described below
//(will be explained in more detail in Ring Multisig paper
//These are aka MG signatutes in earlier drafts of the ring ct paper
// c.f. http://eprint.iacr.org/2015/1098 section 2.
// keyImageV just does I[i] = xx[i] * Hash(xx[i] * G) for each i
// Gen creates a signature which proves that for some column in the keymatrix "pk"
// the signer knows a secret key for each row in that column
// Ver verifies that the MG sig was created correctly
mgSig MLSAG_Gen(const key &message, const keyM & pk, const keyV & xx, const unsigned int index, size_t dsRows) {
mgSig rv;
size_t cols = pk.size();
CHECK_AND_ASSERT_THROW_MES(cols >= 2, "Error! What is c if cols = 1!");
CHECK_AND_ASSERT_THROW_MES(index < cols, "Index out of range");
size_t rows = pk[0].size();
CHECK_AND_ASSERT_THROW_MES(rows >= 1, "Empty pk");
for (size_t i = 1; i < cols; ++i) {
CHECK_AND_ASSERT_THROW_MES(pk[i].size() == rows, "pk is not rectangular");
}
CHECK_AND_ASSERT_THROW_MES(xx.size() == rows, "Bad xx size");
CHECK_AND_ASSERT_THROW_MES(dsRows <= rows, "Bad dsRows size");
size_t i = 0, j = 0, ii = 0;
key c, c_old, L, R, Hi;
sc_0(c_old.bytes);
vector<geDsmp> Ip(dsRows);
rv.II = keyV(dsRows);
keyV alpha(rows);
keyV aG(rows);
rv.ss = keyM(cols, aG);
keyV aHP(dsRows);
keyV toHash(1 + 3 * dsRows + 2 * (rows - dsRows));
toHash[0] = message;
DP("here1");
for (i = 0; i < dsRows; i++) {
skpkGen(alpha[i], aG[i]); //need to save alphas for later..
Hi = hashToPoint(pk[index][i]);
aHP[i] = scalarmultKey(Hi, alpha[i]);
toHash[3 * i + 1] = pk[index][i];
toHash[3 * i + 2] = aG[i];
toHash[3 * i + 3] = aHP[i];
rv.II[i] = scalarmultKey(Hi, xx[i]);
precomp(Ip[i].k, rv.II[i]);
}
size_t ndsRows = 3 * dsRows; //non Double Spendable Rows (see identity chains paper)
for (i = dsRows, ii = 0 ; i < rows ; i++, ii++) {
skpkGen(alpha[i], aG[i]); //need to save alphas for later..
toHash[ndsRows + 2 * ii + 1] = pk[index][i];
toHash[ndsRows + 2 * ii + 2] = aG[i];
}
c_old = hash_to_scalar(toHash);
i = (index + 1) % cols;
if (i == 0) {
copy(rv.cc, c_old);
}
while (i != index) {
rv.ss[i] = skvGen(rows);
sc_0(c.bytes);
for (j = 0; j < dsRows; j++) {
addKeys2(L, rv.ss[i][j], c_old, pk[i][j]);
hashToPoint(Hi, pk[i][j]);
addKeys3(R, rv.ss[i][j], Hi, c_old, Ip[j].k);
toHash[3 * j + 1] = pk[i][j];
toHash[3 * j + 2] = L;
toHash[3 * j + 3] = R;
}
for (j = dsRows, ii = 0; j < rows; j++, ii++) {
addKeys2(L, rv.ss[i][j], c_old, pk[i][j]);
toHash[ndsRows + 2 * ii + 1] = pk[i][j];
toHash[ndsRows + 2 * ii + 2] = L;
}
c = hash_to_scalar(toHash);
copy(c_old, c);
i = (i + 1) % cols;
if (i == 0) {
copy(rv.cc, c_old);
}
}
for (j = 0; j < rows; j++) {
sc_mulsub(rv.ss[index][j].bytes, c.bytes, xx[j].bytes, alpha[j].bytes);
}
return rv;
}
//Multilayered Spontaneous Anonymous Group Signatures (MLSAG signatures)
//This is a just slghtly more efficient version than the ones described below
//(will be explained in more detail in Ring Multisig paper
//These are aka MG signatutes in earlier drafts of the ring ct paper
// c.f. http://eprint.iacr.org/2015/1098 section 2.
// keyImageV just does I[i] = xx[i] * Hash(xx[i] * G) for each i
// Gen creates a signature which proves that for some column in the keymatrix "pk"
// the signer knows a secret key for each row in that column
// Ver verifies that the MG sig was created correctly
bool MLSAG_Ver(const key &message, const keyM & pk, const mgSig & rv, size_t dsRows) {
size_t cols = pk.size();
CHECK_AND_ASSERT_MES(cols >= 2, false, "Error! What is c if cols = 1!");
size_t rows = pk[0].size();
CHECK_AND_ASSERT_MES(rows >= 1, false, "Empty pk");
for (size_t i = 1; i < cols; ++i) {
CHECK_AND_ASSERT_MES(pk[i].size() == rows, false, "pk is not rectangular");
}
CHECK_AND_ASSERT_MES(rv.II.size() == dsRows, false, "Bad II size");
CHECK_AND_ASSERT_MES(rv.ss.size() == cols, false, "Bad rv.ss size");
for (size_t i = 0; i < cols; ++i) {
CHECK_AND_ASSERT_MES(rv.ss[i].size() == rows, false, "rv.ss is not rectangular");
}
CHECK_AND_ASSERT_MES(dsRows <= rows, false, "Bad dsRows value");
for (size_t i = 0; i < rv.ss.size(); ++i)
for (size_t j = 0; j < rv.ss[i].size(); ++j)
CHECK_AND_ASSERT_MES(sc_check(rv.ss[i][j].bytes) == 0, false, "Bad ss slot");
CHECK_AND_ASSERT_MES(sc_check(rv.cc.bytes) == 0, false, "Bad cc");
size_t i = 0, j = 0, ii = 0;
key c, L, R, Hi;
key c_old = copy(rv.cc);
vector<geDsmp> Ip(dsRows);
for (i = 0 ; i < dsRows ; i++) {
precomp(Ip[i].k, rv.II[i]);
}
size_t ndsRows = 3 * dsRows; //non Double Spendable Rows (see identity chains paper
keyV toHash(1 + 3 * dsRows + 2 * (rows - dsRows));
toHash[0] = message;
i = 0;
while (i < cols) {
sc_0(c.bytes);
for (j = 0; j < dsRows; j++) {
addKeys2(L, rv.ss[i][j], c_old, pk[i][j]);
hashToPoint(Hi, pk[i][j]);
addKeys3(R, rv.ss[i][j], Hi, c_old, Ip[j].k);
toHash[3 * j + 1] = pk[i][j];
toHash[3 * j + 2] = L;
toHash[3 * j + 3] = R;
}
for (j = dsRows, ii = 0 ; j < rows ; j++, ii++) {
addKeys2(L, rv.ss[i][j], c_old, pk[i][j]);
toHash[ndsRows + 2 * ii + 1] = pk[i][j];
toHash[ndsRows + 2 * ii + 2] = L;
}
c = hash_to_scalar(toHash);
copy(c_old, c);
i = (i + 1);
}
sc_sub(c.bytes, c_old.bytes, rv.cc.bytes);
return sc_isnonzero(c.bytes) == 0;
}
//proveRange and verRange
//proveRange gives C, and mask such that \sumCi = C
// c.f. http://eprint.iacr.org/2015/1098 section 5.1
// and Ci is a commitment to either 0 or 2^i, i=0,...,63
// thus this proves that "amount" is in [0, 2^64]
// mask is a such that C = aG + bH, and b = amount
//verRange verifies that \sum Ci = C and that each Ci is a commitment to 0 or 2^i
rangeSig proveRange(key & C, key & mask, const xmr_amount & amount) {
sc_0(mask.bytes);
identity(C);
bits b;
d2b(b, amount);
rangeSig sig;
key64 ai;
key64 CiH;
int i = 0;
for (i = 0; i < ATOMS; i++) {
2016-05-30 00:34:43 +00:00
skGen(ai[i]);
if (b[i] == 0) {
scalarmultBase(sig.Ci[i], ai[i]);
}
if (b[i] == 1) {
addKeys1(sig.Ci[i], ai[i], H2[i]);
}
subKeys(CiH[i], sig.Ci[i], H2[i]);
sc_add(mask.bytes, mask.bytes, ai[i].bytes);
addKeys(C, C, sig.Ci[i]);
}
2016-11-17 23:17:21 +00:00
sig.asig = genBorromean(ai, sig.Ci, CiH, b);
return sig;
}
//proveRange and verRange
//proveRange gives C, and mask such that \sumCi = C
// c.f. http://eprint.iacr.org/2015/1098 section 5.1
// and Ci is a commitment to either 0 or 2^i, i=0,...,63
// thus this proves that "amount" is in [0, 2^64]
// mask is a such that C = aG + bH, and b = amount
//verRange verifies that \sum Ci = C and that each Ci is a commitment to 0 or 2^i
bool verRange(const key & C, const rangeSig & as) {
try
{
PERF_TIMER(verRange);
key64 CiH;
int i = 0;
key Ctmp = identity();
for (i = 0; i < 64; i++) {
subKeys(CiH[i], as.Ci[i], H2[i]);
addKeys(Ctmp, Ctmp, as.Ci[i]);
}
2016-08-27 09:10:19 +00:00
if (!equalKeys(C, Ctmp))
return false;
2016-11-17 23:17:21 +00:00
if (!verifyBorromean(as.asig, as.Ci, CiH))
2016-08-27 09:10:19 +00:00
return false;
return true;
}
// we can get deep throws from ge_frombytes_vartime if input isn't valid
catch (...) { return false; }
}
key get_pre_mlsag_hash(const rctSig &rv)
{
keyV hashes;
hashes.reserve(3);
hashes.push_back(rv.message);
crypto::hash h;
std::stringstream ss;
binary_archive<true> ba(ss);
const size_t inputs = rv.pseudoOuts.size();
const size_t outputs = rv.ecdhInfo.size();
CHECK_AND_ASSERT_THROW_MES(const_cast<rctSig&>(rv).serialize_rctsig_base(ba, inputs, outputs),
"Failed to serialize rctSigBase");
cryptonote::get_blob_hash(ss.str(), h);
hashes.push_back(hash2rct(h));
keyV kv;
kv.reserve((64*3+1) * rv.p.rangeSigs.size());
for (auto r: rv.p.rangeSigs)
{
for (size_t n = 0; n < 64; ++n)
2016-11-17 23:17:21 +00:00
kv.push_back(r.asig.s0[n]);
for (size_t n = 0; n < 64; ++n)
2016-11-17 23:17:21 +00:00
kv.push_back(r.asig.s1[n]);
kv.push_back(r.asig.ee);
for (size_t n = 0; n < 64; ++n)
kv.push_back(r.Ci[n]);
}
hashes.push_back(cn_fast_hash(kv));
return cn_fast_hash(hashes);
}
//Ring-ct MG sigs
//Prove:
// c.f. http://eprint.iacr.org/2015/1098 section 4. definition 10.
// This does the MG sig on the "dest" part of the given key matrix, and
// the last row is the sum of input commitments from that column - sum output commitments
// this shows that sum inputs = sum outputs
//Ver:
// verifies the above sig is created corretly
2016-07-10 11:57:22 +00:00
mgSig proveRctMG(const key &message, const ctkeyM & pubs, const ctkeyV & inSk, const ctkeyV &outSk, const ctkeyV & outPk, unsigned int index, key txnFeeKey) {
mgSig mg;
//setup vars
size_t cols = pubs.size();
CHECK_AND_ASSERT_THROW_MES(cols >= 1, "Empty pubs");
size_t rows = pubs[0].size();
CHECK_AND_ASSERT_THROW_MES(rows >= 1, "Empty pubs");
for (size_t i = 1; i < cols; ++i) {
CHECK_AND_ASSERT_THROW_MES(pubs[i].size() == rows, "pubs is not rectangular");
}
CHECK_AND_ASSERT_THROW_MES(inSk.size() == rows, "Bad inSk size");
CHECK_AND_ASSERT_THROW_MES(outSk.size() == outPk.size(), "Bad outSk/outPk size");
keyV sk(rows + 1);
keyV tmp(rows + 1);
size_t i = 0, j = 0;
for (i = 0; i < rows + 1; i++) {
sc_0(sk[i].bytes);
identity(tmp[i]);
}
keyM M(cols, tmp);
//create the matrix to mg sig
for (i = 0; i < cols; i++) {
M[i][rows] = identity();
for (j = 0; j < rows; j++) {
M[i][j] = pubs[i][j].dest;
addKeys(M[i][rows], M[i][rows], pubs[i][j].mask); //add input commitments in last row
}
}
sc_0(sk[rows].bytes);
for (j = 0; j < rows; j++) {
sk[j] = copy(inSk[j].dest);
sc_add(sk[rows].bytes, sk[rows].bytes, inSk[j].mask.bytes); //add masks in last row
}
for (i = 0; i < cols; i++) {
for (size_t j = 0; j < outPk.size(); j++) {
subKeys(M[i][rows], M[i][rows], outPk[j].mask); //subtract output Ci's in last row
}
2016-06-12 20:53:01 +00:00
//subtract txn fee output in last row
subKeys(M[i][rows], M[i][rows], txnFeeKey);
}
for (size_t j = 0; j < outPk.size(); j++) {
sc_sub(sk[rows].bytes, sk[rows].bytes, outSk[j].mask.bytes); //subtract output masks in last row..
}
return MLSAG_Gen(message, M, sk, index, rows);
}
//Ring-ct MG sigs Simple
// Simple version for when we assume only
// post rct inputs
// here pubs is a vector of (P, C) length mixin
// inSk is x, a_in corresponding to signing index
// a_out, Cout is for the output commitment
// index is the signing index..
mgSig proveRctMGSimple(const key &message, const ctkeyV & pubs, const ctkey & inSk, const key &a , const key &Cout, unsigned int index) {
mgSig mg;
//setup vars
size_t rows = 1;
size_t cols = pubs.size();
CHECK_AND_ASSERT_THROW_MES(cols >= 1, "Empty pubs");
keyV tmp(rows + 1);
keyV sk(rows + 1);
size_t i;
keyM M(cols, tmp);
for (i = 0; i < cols; i++) {
M[i][0] = pubs[i].dest;
subKeys(M[i][1], pubs[i].mask, Cout);
sk[0] = copy(inSk.dest);
sc_sub(sk[1].bytes, inSk.mask.bytes, a.bytes);
}
return MLSAG_Gen(message, M, sk, index, rows);
}
//Ring-ct MG sigs
//Prove:
// c.f. http://eprint.iacr.org/2015/1098 section 4. definition 10.
// This does the MG sig on the "dest" part of the given key matrix, and
// the last row is the sum of input commitments from that column - sum output commitments
// this shows that sum inputs = sum outputs
//Ver:
// verifies the above sig is created corretly
bool verRctMG(const mgSig &mg, const ctkeyM & pubs, const ctkeyV & outPk, key txnFeeKey, const key &message) {
PERF_TIMER(verRctMG);
//setup vars
size_t cols = pubs.size();
CHECK_AND_ASSERT_MES(cols >= 1, false, "Empty pubs");
size_t rows = pubs[0].size();
CHECK_AND_ASSERT_MES(rows >= 1, false, "Empty pubs");
for (size_t i = 1; i < cols; ++i) {
CHECK_AND_ASSERT_MES(pubs[i].size() == rows, false, "pubs is not rectangular");
}
keyV tmp(rows + 1);
size_t i = 0, j = 0;
for (i = 0; i < rows + 1; i++) {
identity(tmp[i]);
}
keyM M(cols, tmp);
//create the matrix to mg sig
for (j = 0; j < rows; j++) {
for (i = 0; i < cols; i++) {
M[i][j] = pubs[i][j].dest;
2016-06-12 20:53:01 +00:00
addKeys(M[i][rows], M[i][rows], pubs[i][j].mask); //add Ci in last row
}
}
2016-06-12 20:53:01 +00:00
for (i = 0; i < cols; i++) {
for (j = 0; j < outPk.size(); j++) {
subKeys(M[i][rows], M[i][rows], outPk[j].mask); //subtract output Ci's in last row
}
2016-06-12 20:53:01 +00:00
//subtract txn fee output in last row
subKeys(M[i][rows], M[i][rows], txnFeeKey);
}
return MLSAG_Ver(message, M, mg, rows);
}
//Ring-ct Simple MG sigs
//Ver:
//This does a simplified version, assuming only post Rct
//inputs
bool verRctMGSimple(const key &message, const mgSig &mg, const ctkeyV & pubs, const key & C) {
try
{
PERF_TIMER(verRctMGSimple);
//setup vars
size_t rows = 1;
size_t cols = pubs.size();
CHECK_AND_ASSERT_MES(cols >= 1, false, "Empty pubs");
keyV tmp(rows + 1);
size_t i;
keyM M(cols, tmp);
//create the matrix to mg sig
for (i = 0; i < cols; i++) {
M[i][0] = pubs[i].dest;
subKeys(M[i][1], pubs[i].mask, C);
}
//DP(C);
return MLSAG_Ver(message, M, mg, rows);
}
catch (...) { return false; }
}
2016-10-15 12:31:16 +00:00
//These functions get keys from blockchain
//replace these when connecting blockchain
//getKeyFromBlockchain grabs a key from the blockchain at "reference_index" to mix with
//populateFromBlockchain creates a keymatrix with "mixin" columns and one of the columns is inPk
// the return value are the key matrix, and the index where inPk was put (random).
void getKeyFromBlockchain(ctkey & a, size_t reference_index) {
a.mask = pkGen();
a.dest = pkGen();
}
//These functions get keys from blockchain
//replace these when connecting blockchain
//getKeyFromBlockchain grabs a key from the blockchain at "reference_index" to mix with
2016-06-04 08:37:38 +00:00
//populateFromBlockchain creates a keymatrix with "mixin" + 1 columns and one of the columns is inPk
// the return value are the key matrix, and the index where inPk was put (random).
tuple<ctkeyM, xmr_amount> populateFromBlockchain(ctkeyV inPk, int mixin) {
int rows = inPk.size();
2016-06-04 08:37:38 +00:00
ctkeyM rv(mixin + 1, inPk);
int index = randXmrAmount(mixin);
int i = 0, j = 0;
2016-06-04 08:37:38 +00:00
for (i = 0; i <= mixin; i++) {
if (i != index) {
for (j = 0; j < rows; j++) {
getKeyFromBlockchain(rv[i][j], (size_t)randXmrAmount);
}
}
}
return make_tuple(rv, index);
}
//These functions get keys from blockchain
//replace these when connecting blockchain
//getKeyFromBlockchain grabs a key from the blockchain at "reference_index" to mix with
//populateFromBlockchain creates a keymatrix with "mixin" columns and one of the columns is inPk
// the return value are the key matrix, and the index where inPk was put (random).
xmr_amount populateFromBlockchainSimple(ctkeyV & mixRing, const ctkey & inPk, int mixin) {
int index = randXmrAmount(mixin);
int i = 0;
for (i = 0; i <= mixin; i++) {
if (i != index) {
getKeyFromBlockchain(mixRing[i], (size_t)randXmrAmount(1000));
} else {
mixRing[i] = inPk;
}
}
return index;
}
//RingCT protocol
//genRct:
// creates an rctSig with all data necessary to verify the rangeProofs and that the signer owns one of the
// columns that are claimed as inputs, and that the sum of inputs = sum of outputs.
// Also contains masked "amount" and "mask" so the receiver can see how much they received
//verRct:
// verifies that all signatures (rangeProogs, MG sig, sum inputs = outputs) are correct
//decodeRct: (c.f. http://eprint.iacr.org/2015/1098 section 5.1.1)
// uses the attached ecdh info to find the amounts represented by each output commitment
// must know the destination private key to find the correct amount, else will return a random number
2016-06-12 20:53:01 +00:00
// Note: For txn fees, the last index in the amounts vector should contain that
// Thus the amounts vector will be "one" longer than the destinations vectort
rctSig genRct(const key &message, const ctkeyV & inSk, const keyV & destinations, const vector<xmr_amount> & amounts, const ctkeyM &mixRing, const keyV &amount_keys, unsigned int index, ctkeyV &outSk) {
2016-06-12 21:41:40 +00:00
CHECK_AND_ASSERT_THROW_MES(amounts.size() == destinations.size() || amounts.size() == destinations.size() + 1, "Different number of amounts/destinations");
CHECK_AND_ASSERT_THROW_MES(amount_keys.size() == destinations.size(), "Different number of amount_keys/destinations");
CHECK_AND_ASSERT_THROW_MES(index < mixRing.size(), "Bad index into mixRing");
for (size_t n = 0; n < mixRing.size(); ++n) {
CHECK_AND_ASSERT_THROW_MES(mixRing[n].size() == inSk.size(), "Bad mixRing size");
}
rctSig rv;
rv.type = RCTTypeFull;
rv.message = message;
rv.outPk.resize(destinations.size());
rv.p.rangeSigs.resize(destinations.size());
rv.ecdhInfo.resize(destinations.size());
size_t i = 0;
keyV masks(destinations.size()); //sk mask..
2016-07-11 22:14:58 +00:00
outSk.resize(destinations.size());
for (i = 0; i < destinations.size(); i++) {
//add destination to sig
rv.outPk[i].dest = copy(destinations[i]);
//compute range proof
rv.p.rangeSigs[i] = proveRange(rv.outPk[i].mask, outSk[i].mask, amounts[i]);
#ifdef DBG
CHECK_AND_ASSERT_THROW_MES(verRange(rv.outPk[i].mask, rv.p.rangeSigs[i]), "verRange failed on newly created proof");
#endif
//mask amount and mask
rv.ecdhInfo[i].mask = copy(outSk[i].mask);
rv.ecdhInfo[i].amount = d2h(amounts[i]);
ecdhEncode(rv.ecdhInfo[i], amount_keys[i]);
}
2016-06-12 20:53:01 +00:00
//set txn fee
2016-06-12 21:41:40 +00:00
if (amounts.size() > destinations.size())
{
rv.txnFee = amounts[destinations.size()];
}
else
{
rv.txnFee = 0;
}
2016-06-12 20:53:01 +00:00
key txnFeeKey = scalarmultH(d2h(rv.txnFee));
rv.mixRing = mixRing;
rv.p.MGs.push_back(proveRctMG(get_pre_mlsag_hash(rv), rv.mixRing, inSk, outSk, rv.outPk, index, txnFeeKey));
return rv;
}
rctSig genRct(const key &message, const ctkeyV & inSk, const ctkeyV & inPk, const keyV & destinations, const vector<xmr_amount> & amounts, const keyV &amount_keys, const int mixin) {
unsigned int index;
ctkeyM mixRing;
2016-07-11 22:14:58 +00:00
ctkeyV outSk;
tie(mixRing, index) = populateFromBlockchain(inPk, mixin);
return genRct(message, inSk, destinations, amounts, mixRing, amount_keys, index, outSk);
}
//RCT simple
//for post-rct only
rctSig genRctSimple(const key &message, const ctkeyV & inSk, const keyV & destinations, const vector<xmr_amount> &inamounts, const vector<xmr_amount> &outamounts, xmr_amount txnFee, const ctkeyM & mixRing, const keyV &amount_keys, const std::vector<unsigned int> & index, ctkeyV &outSk) {
CHECK_AND_ASSERT_THROW_MES(inamounts.size() > 0, "Empty inamounts");
CHECK_AND_ASSERT_THROW_MES(inamounts.size() == inSk.size(), "Different number of inamounts/inSk");
CHECK_AND_ASSERT_THROW_MES(outamounts.size() == destinations.size(), "Different number of amounts/destinations");
CHECK_AND_ASSERT_THROW_MES(amount_keys.size() == destinations.size(), "Different number of amount_keys/destinations");
CHECK_AND_ASSERT_THROW_MES(index.size() == inSk.size(), "Different number of index/inSk");
CHECK_AND_ASSERT_THROW_MES(mixRing.size() == inSk.size(), "Different number of mixRing/inSk");
for (size_t n = 0; n < mixRing.size(); ++n) {
CHECK_AND_ASSERT_THROW_MES(index[n] < mixRing[n].size(), "Bad index into mixRing");
}
2016-07-10 11:57:22 +00:00
rctSig rv;
rv.type = RCTTypeSimple;
rv.message = message;
rv.outPk.resize(destinations.size());
rv.p.rangeSigs.resize(destinations.size());
rv.ecdhInfo.resize(destinations.size());
size_t i;
keyV masks(destinations.size()); //sk mask..
2016-07-11 22:14:58 +00:00
outSk.resize(destinations.size());
key sumout = zero();
for (i = 0; i < destinations.size(); i++) {
//add destination to sig
rv.outPk[i].dest = copy(destinations[i]);
//compute range proof
rv.p.rangeSigs[i] = proveRange(rv.outPk[i].mask, outSk[i].mask, outamounts[i]);
#ifdef DBG
verRange(rv.outPk[i].mask, rv.p.rangeSigs[i]);
#endif
sc_add(sumout.bytes, outSk[i].mask.bytes, sumout.bytes);
//mask amount and mask
rv.ecdhInfo[i].mask = copy(outSk[i].mask);
rv.ecdhInfo[i].amount = d2h(outamounts[i]);
ecdhEncode(rv.ecdhInfo[i], amount_keys[i]);
}
//set txn fee
rv.txnFee = txnFee;
// TODO: unused ??
// key txnFeeKey = scalarmultH(d2h(rv.txnFee));
rv.mixRing = mixRing;
rv.pseudoOuts.resize(inamounts.size());
rv.p.MGs.resize(inamounts.size());
key sumpouts = zero(); //sum pseudoOut masks
keyV a(inamounts.size());
for (i = 0 ; i < inamounts.size() - 1; i++) {
skGen(a[i]);
sc_add(sumpouts.bytes, a[i].bytes, sumpouts.bytes);
genC(rv.pseudoOuts[i], a[i], inamounts[i]);
}
rv.mixRing = mixRing;
sc_sub(a[i].bytes, sumout.bytes, sumpouts.bytes);
genC(rv.pseudoOuts[i], a[i], inamounts[i]);
DP(rv.pseudoOuts[i]);
key full_message = get_pre_mlsag_hash(rv);
for (i = 0 ; i < inamounts.size(); i++) {
rv.p.MGs[i] = proveRctMGSimple(full_message, rv.mixRing[i], inSk[i], a[i], rv.pseudoOuts[i], index[i]);
}
return rv;
}
rctSig genRctSimple(const key &message, const ctkeyV & inSk, const ctkeyV & inPk, const keyV & destinations, const vector<xmr_amount> &inamounts, const vector<xmr_amount> &outamounts, const keyV &amount_keys, xmr_amount txnFee, unsigned int mixin) {
std::vector<unsigned int> index;
index.resize(inPk.size());
ctkeyM mixRing;
2016-07-11 22:14:58 +00:00
ctkeyV outSk;
mixRing.resize(inPk.size());
for (size_t i = 0; i < inPk.size(); ++i) {
mixRing[i].resize(mixin+1);
index[i] = populateFromBlockchainSimple(mixRing[i], inPk[i], mixin);
}
return genRctSimple(message, inSk, destinations, inamounts, outamounts, txnFee, mixRing, amount_keys, index, outSk);
}
//RingCT protocol
//genRct:
// creates an rctSig with all data necessary to verify the rangeProofs and that the signer owns one of the
// columns that are claimed as inputs, and that the sum of inputs = sum of outputs.
// Also contains masked "amount" and "mask" so the receiver can see how much they received
//verRct:
// verifies that all signatures (rangeProogs, MG sig, sum inputs = outputs) are correct
//decodeRct: (c.f. http://eprint.iacr.org/2015/1098 section 5.1.1)
// uses the attached ecdh info to find the amounts represented by each output commitment
// must know the destination private key to find the correct amount, else will return a random number
bool verRct(const rctSig & rv, bool semantics) {
PERF_TIMER(verRct);
CHECK_AND_ASSERT_MES(rv.type == RCTTypeFull, false, "verRct called on non-full rctSig");
if (semantics)
{
CHECK_AND_ASSERT_MES(rv.outPk.size() == rv.p.rangeSigs.size(), false, "Mismatched sizes of outPk and rv.p.rangeSigs");
CHECK_AND_ASSERT_MES(rv.outPk.size() == rv.ecdhInfo.size(), false, "Mismatched sizes of outPk and rv.ecdhInfo");
CHECK_AND_ASSERT_MES(rv.p.MGs.size() == 1, false, "full rctSig has not one MG");
}
else
{
// semantics check is early, we don't have the MGs resolved yet
}
// some rct ops can throw
try
{
if (semantics) {
std::deque<bool> results(rv.outPk.size(), false);
tools::thread_group threadpool(tools::thread_group::optimal_with_max(rv.outPk.size()));
tools::task_region(threadpool, [&] (tools::task_region_handle& region) {
DP("range proofs verified?");
for (size_t i = 0; i < rv.outPk.size(); i++) {
region.run([&, i] {
results[i] = verRange(rv.outPk[i].mask, rv.p.rangeSigs[i]);
});
}
});
for (size_t i = 0; i < rv.outPk.size(); ++i) {
if (!results[i]) {
LOG_PRINT_L1("Range proof verified failed for output " << i);
return false;
}
}
}
2016-10-15 12:31:16 +00:00
if (!semantics) {
//compute txn fee
key txnFeeKey = scalarmultH(d2h(rv.txnFee));
bool mgVerd = verRctMG(rv.p.MGs[0], rv.mixRing, rv.outPk, txnFeeKey, get_pre_mlsag_hash(rv));
DP("mg sig verified?");
DP(mgVerd);
if (!mgVerd) {
LOG_PRINT_L1("MG signature verification failed");
return false;
2016-10-15 12:31:16 +00:00
}
}
2016-10-15 12:31:16 +00:00
return true;
}
catch(...)
{
return false;
}
}
//ver RingCT simple
//assumes only post-rct style inputs (at least for max anonymity)
bool verRctSimple(const rctSig & rv, bool semantics) {
try
{
PERF_TIMER(verRctSimple);
CHECK_AND_ASSERT_MES(rv.type == RCTTypeSimple, false, "verRctSimple called on non simple rctSig");
if (semantics)
{
CHECK_AND_ASSERT_MES(rv.outPk.size() == rv.p.rangeSigs.size(), false, "Mismatched sizes of outPk and rv.p.rangeSigs");
CHECK_AND_ASSERT_MES(rv.outPk.size() == rv.ecdhInfo.size(), false, "Mismatched sizes of outPk and rv.ecdhInfo");
CHECK_AND_ASSERT_MES(rv.pseudoOuts.size() == rv.p.MGs.size(), false, "Mismatched sizes of rv.pseudoOuts and rv.p.MGs");
}
else
{
// semantics check is early, and mixRing/MGs aren't resolved yet
CHECK_AND_ASSERT_MES(rv.pseudoOuts.size() == rv.mixRing.size(), false, "Mismatched sizes of rv.pseudoOuts and mixRing");
}
const size_t threads = std::max(rv.outPk.size(), rv.mixRing.size());
std::deque<bool> results(threads);
tools::thread_group threadpool(tools::thread_group::optimal_with_max(threads));
if (semantics) {
key sumOutpks = identity();
for (size_t i = 0; i < rv.outPk.size(); i++) {
addKeys(sumOutpks, sumOutpks, rv.outPk[i].mask);
}
DP(sumOutpks);
key txnFeeKey = scalarmultH(d2h(rv.txnFee));
addKeys(sumOutpks, txnFeeKey, sumOutpks);
2016-10-15 12:31:16 +00:00
key sumPseudoOuts = identity();
for (size_t i = 0 ; i < rv.pseudoOuts.size() ; i++) {
addKeys(sumPseudoOuts, sumPseudoOuts, rv.pseudoOuts[i]);
}
DP(sumPseudoOuts);
//check pseudoOuts vs Outs..
if (!equalKeys(sumPseudoOuts, sumOutpks)) {
LOG_PRINT_L1("Sum check failed");
return false;
2016-10-15 12:31:16 +00:00
}
results.clear();
results.resize(rv.outPk.size());
tools::task_region(threadpool, [&] (tools::task_region_handle& region) {
for (size_t i = 0; i < rv.outPk.size(); i++) {
region.run([&, i] {
results[i] = verRange(rv.outPk[i].mask, rv.p.rangeSigs[i]);
});
}
});
for (size_t i = 0; i < results.size(); ++i) {
if (!results[i]) {
LOG_PRINT_L1("Range proof verified failed for output " << i);
return false;
}
}
2016-10-15 12:31:16 +00:00
}
else {
const key message = get_pre_mlsag_hash(rv);
2016-10-15 12:31:16 +00:00
results.clear();
results.resize(rv.mixRing.size());
tools::task_region(threadpool, [&] (tools::task_region_handle& region) {
for (size_t i = 0 ; i < rv.mixRing.size() ; i++) {
region.run([&, i] {
results[i] = verRctMGSimple(message, rv.p.MGs[i], rv.mixRing[i], rv.pseudoOuts[i]);
});
}
});
for (size_t i = 0; i < results.size(); ++i) {
if (!results[i]) {
LOG_PRINT_L1("verRctMGSimple failed for input " << i);
return false;
}
}
}
return true;
}
// we can get deep throws from ge_frombytes_vartime if input isn't valid
catch (...) { return false; }
}
//RingCT protocol
//genRct:
// creates an rctSig with all data necessary to verify the rangeProofs and that the signer owns one of the
// columns that are claimed as inputs, and that the sum of inputs = sum of outputs.
// Also contains masked "amount" and "mask" so the receiver can see how much they received
//verRct:
// verifies that all signatures (rangeProogs, MG sig, sum inputs = outputs) are correct
//decodeRct: (c.f. http://eprint.iacr.org/2015/1098 section 5.1.1)
// uses the attached ecdh info to find the amounts represented by each output commitment
// must know the destination private key to find the correct amount, else will return a random number
xmr_amount decodeRct(const rctSig & rv, const key & sk, unsigned int i, key & mask) {
CHECK_AND_ASSERT_MES(rv.type == RCTTypeFull, false, "decodeRct called on non-full rctSig");
CHECK_AND_ASSERT_THROW_MES(rv.outPk.size() == rv.ecdhInfo.size(), "Mismatched sizes of rv.outPk and rv.ecdhInfo");
CHECK_AND_ASSERT_THROW_MES(i < rv.ecdhInfo.size(), "Bad index");
//mask amount and mask
ecdhTuple ecdh_info = rv.ecdhInfo[i];
ecdhDecode(ecdh_info, sk);
mask = ecdh_info.mask;
key amount = ecdh_info.amount;
key C = rv.outPk[i].mask;
DP("C");
DP(C);
key Ctmp;
addKeys2(Ctmp, mask, amount, H);
DP("Ctmp");
DP(Ctmp);
if (equalKeys(C, Ctmp) == false) {
CHECK_AND_ASSERT_THROW_MES(false, "warning, amount decoded incorrectly, will be unable to spend");
}
return h2d(amount);
}
xmr_amount decodeRct(const rctSig & rv, const key & sk, unsigned int i) {
key mask;
return decodeRct(rv, sk, i, mask);
}
xmr_amount decodeRctSimple(const rctSig & rv, const key & sk, unsigned int i, key &mask) {
CHECK_AND_ASSERT_MES(rv.type == RCTTypeSimple, false, "decodeRct called on non simple rctSig");
CHECK_AND_ASSERT_THROW_MES(rv.outPk.size() == rv.ecdhInfo.size(), "Mismatched sizes of rv.outPk and rv.ecdhInfo");
CHECK_AND_ASSERT_THROW_MES(i < rv.ecdhInfo.size(), "Bad index");
//mask amount and mask
ecdhTuple ecdh_info = rv.ecdhInfo[i];
ecdhDecode(ecdh_info, sk);
2016-07-10 11:57:22 +00:00
mask = ecdh_info.mask;
key amount = ecdh_info.amount;
key C = rv.outPk[i].mask;
DP("C");
DP(C);
key Ctmp;
addKeys2(Ctmp, mask, amount, H);
DP("Ctmp");
DP(Ctmp);
if (equalKeys(C, Ctmp) == false) {
CHECK_AND_ASSERT_THROW_MES(false, "warning, amount decoded incorrectly, will be unable to spend");
}
return h2d(amount);
}
2016-07-10 11:57:22 +00:00
xmr_amount decodeRctSimple(const rctSig & rv, const key & sk, unsigned int i) {
key mask;
return decodeRctSimple(rv, sk, i, mask);
}
}