bitcoin/src/test/fuzz/package_eval.cpp
Sjors Provoost d511adb664
[miner] omit dummy extraNonce via IPC
Previously the coinbase transaction generated by our miner code was
not used downstream, because the getblocktemplate RPC excludes it.

Since the Mining IPC interface was introduced in #30200 we do expose
this dummy coinbase transaction. In Stratum v2 several parts of it
are communicated downstream, including the scriptSig.

This commit removes the dummy extraNonce from the coinbase scriptSig
in block templates requested via IPC. This limits the scriptSig
to what is essential for consensus (BIP34) and removes the need for
external mining software to remove the dummy, or even ignore
the scriptSig we provide and generate it some other way. This
could cause problems if a future soft fork requires additional
data to be committed here.

A test is added to verify the new IPC behavior.

It achieves this by introducing an include_dummy_extranonce
option which defaults to false with all test code updated to
set it to true. Because this option is not exposed via IPC,
callers will no longer see it.

The caller needs to ensure that for blocks 1 through 16
they pad the scriptSig in order to avoid bad-cb-length.

Co-authored-by: Anthony Towns <aj@erisian.com.au>
2026-01-27 14:45:35 +01:00

542 lines
24 KiB
C++

// Copyright (c) 2023-present The Bitcoin Core developers
// Distributed under the MIT software license, see the accompanying
// file COPYING or http://www.opensource.org/licenses/mit-license.php.
#include <consensus/validation.h>
#include <node/context.h>
#include <node/mempool_args.h>
#include <node/miner.h>
#include <policy/truc_policy.h>
#include <test/fuzz/FuzzedDataProvider.h>
#include <test/fuzz/fuzz.h>
#include <test/fuzz/util.h>
#include <test/fuzz/util/mempool.h>
#include <test/util/mining.h>
#include <test/util/script.h>
#include <test/util/setup_common.h>
#include <test/util/txmempool.h>
#include <util/check.h>
#include <util/rbf.h>
#include <util/translation.h>
#include <validation.h>
#include <validationinterface.h>
using node::BlockAssembler;
using node::NodeContext;
namespace {
const TestingSetup* g_setup;
std::vector<COutPoint> g_outpoints_coinbase_init_mature;
struct MockedTxPool : public CTxMemPool {
void RollingFeeUpdate() EXCLUSIVE_LOCKS_REQUIRED(!cs)
{
LOCK(cs);
lastRollingFeeUpdate = GetTime();
blockSinceLastRollingFeeBump = true;
}
};
void initialize_tx_pool()
{
static const auto testing_setup = MakeNoLogFileContext<const TestingSetup>();
g_setup = testing_setup.get();
SetMockTime(WITH_LOCK(g_setup->m_node.chainman->GetMutex(), return g_setup->m_node.chainman->ActiveTip()->Time()));
BlockAssembler::Options options;
options.coinbase_output_script = P2WSH_EMPTY;
options.include_dummy_extranonce = true;
for (int i = 0; i < 2 * COINBASE_MATURITY; ++i) {
COutPoint prevout{MineBlock(g_setup->m_node, options)};
if (i < COINBASE_MATURITY) {
// Remember the txids to avoid expensive disk access later on
g_outpoints_coinbase_init_mature.push_back(prevout);
}
}
g_setup->m_node.validation_signals->SyncWithValidationInterfaceQueue();
}
struct OutpointsUpdater final : public CValidationInterface {
std::set<COutPoint>& m_mempool_outpoints;
explicit OutpointsUpdater(std::set<COutPoint>& r)
: m_mempool_outpoints{r} {}
void TransactionAddedToMempool(const NewMempoolTransactionInfo& tx, uint64_t /* mempool_sequence */) override
{
// for coins spent we always want to be able to rbf so they're not removed
// outputs from this tx can now be spent
for (uint32_t index{0}; index < tx.info.m_tx->vout.size(); ++index) {
m_mempool_outpoints.insert(COutPoint{tx.info.m_tx->GetHash(), index});
}
}
void TransactionRemovedFromMempool(const CTransactionRef& tx, MemPoolRemovalReason reason, uint64_t /* mempool_sequence */) override
{
// outpoints spent by this tx are now available
for (const auto& input : tx->vin) {
// Could already exist if this was a replacement
m_mempool_outpoints.insert(input.prevout);
}
// outpoints created by this tx no longer exist
for (uint32_t index{0}; index < tx->vout.size(); ++index) {
m_mempool_outpoints.erase(COutPoint{tx->GetHash(), index});
}
}
};
struct TransactionsDelta final : public CValidationInterface {
std::set<CTransactionRef>& m_added;
explicit TransactionsDelta(std::set<CTransactionRef>& a)
: m_added{a} {}
void TransactionAddedToMempool(const NewMempoolTransactionInfo& tx, uint64_t /* mempool_sequence */) override
{
// Transactions may be entered and booted any number of times
m_added.insert(tx.info.m_tx);
}
void TransactionRemovedFromMempool(const CTransactionRef& tx, MemPoolRemovalReason reason, uint64_t /* mempool_sequence */) override
{
// Transactions may be entered and booted any number of times
m_added.erase(tx);
}
};
void MockTime(FuzzedDataProvider& fuzzed_data_provider, const Chainstate& chainstate)
{
const auto time = ConsumeTime(fuzzed_data_provider,
chainstate.m_chain.Tip()->GetMedianTimePast() + 1,
std::numeric_limits<decltype(chainstate.m_chain.Tip()->nTime)>::max());
SetMockTime(time);
}
std::unique_ptr<CTxMemPool> MakeMempool(FuzzedDataProvider& fuzzed_data_provider, const NodeContext& node)
{
// Take the default options for tests...
CTxMemPool::Options mempool_opts{MemPoolOptionsForTest(node)};
// ...override specific options for this specific fuzz suite
mempool_opts.limits.ancestor_count = fuzzed_data_provider.ConsumeIntegralInRange<unsigned>(0, 50);
mempool_opts.limits.descendant_count = fuzzed_data_provider.ConsumeIntegralInRange<unsigned>(0, 50);
mempool_opts.max_size_bytes = fuzzed_data_provider.ConsumeIntegralInRange<unsigned>(0, 200) * 1'000'000;
mempool_opts.expiry = std::chrono::hours{fuzzed_data_provider.ConsumeIntegralInRange<unsigned>(0, 999)};
// Only interested in 2 cases: sigop cost 0 or when single legacy sigop cost is >> 1KvB
nBytesPerSigOp = fuzzed_data_provider.ConsumeIntegralInRange<unsigned>(0, 1) * 10'000;
mempool_opts.check_ratio = 1;
mempool_opts.require_standard = fuzzed_data_provider.ConsumeBool();
bilingual_str error;
// ...and construct a CTxMemPool from it
auto mempool{std::make_unique<CTxMemPool>(std::move(mempool_opts), error)};
// ... ignore the error since it might be beneficial to fuzz even when the
// mempool size is unreasonably small
Assert(error.empty() || error.original.starts_with("-maxmempool must be at least "));
return mempool;
}
std::unique_ptr<CTxMemPool> MakeEphemeralMempool(const NodeContext& node)
{
// Take the default options for tests...
CTxMemPool::Options mempool_opts{MemPoolOptionsForTest(node)};
mempool_opts.check_ratio = 1;
// Require standardness rules otherwise ephemeral dust is no-op
mempool_opts.require_standard = true;
// And set minrelay to 0 to allow ephemeral parent tx even with non-TRUC
mempool_opts.min_relay_feerate = CFeeRate(0);
bilingual_str error;
// ...and construct a CTxMemPool from it
auto mempool{std::make_unique<CTxMemPool>(std::move(mempool_opts), error)};
Assert(error.empty());
return mempool;
}
// Scan mempool for a tx that has spent dust and return a
// prevout of the child that isn't the dusty parent itself.
// This is used to double-spend the child out of the mempool,
// leaving the parent childless.
// This assumes CheckMempoolEphemeralInvariants has passed for tx_pool.
std::optional<COutPoint> GetChildEvictingPrevout(const CTxMemPool& tx_pool)
{
LOCK(tx_pool.cs);
for (const auto& tx_info : tx_pool.infoAll()) {
const auto& entry = *Assert(tx_pool.GetEntry(tx_info.tx->GetHash()));
std::vector<uint32_t> dust_indexes{GetDust(*tx_info.tx, tx_pool.m_opts.dust_relay_feerate)};
if (!dust_indexes.empty()) {
const auto& children = tx_pool.GetChildren(entry);
if (!children.empty()) {
Assert(children.size() == 1);
// Find an input that doesn't spend from parent's txid
const auto& only_child = children.begin()->get().GetTx();
for (const auto& tx_input : only_child.vin) {
if (tx_input.prevout.hash != tx_info.tx->GetHash()) {
return tx_input.prevout;
}
}
}
}
}
return std::nullopt;
}
FUZZ_TARGET(ephemeral_package_eval, .init = initialize_tx_pool)
{
SeedRandomStateForTest(SeedRand::ZEROS);
FuzzedDataProvider fuzzed_data_provider(buffer.data(), buffer.size());
const auto& node = g_setup->m_node;
auto& chainstate{static_cast<DummyChainState&>(node.chainman->ActiveChainstate())};
MockTime(fuzzed_data_provider, chainstate);
// All RBF-spendable outpoints outside of the unsubmitted package
std::set<COutPoint> mempool_outpoints;
std::unordered_map<COutPoint, CAmount, SaltedOutpointHasher> outpoints_value;
for (const auto& outpoint : g_outpoints_coinbase_init_mature) {
Assert(mempool_outpoints.insert(outpoint).second);
outpoints_value[outpoint] = 50 * COIN;
}
auto outpoints_updater = std::make_shared<OutpointsUpdater>(mempool_outpoints);
node.validation_signals->RegisterSharedValidationInterface(outpoints_updater);
auto tx_pool_{MakeEphemeralMempool(node)};
MockedTxPool& tx_pool = *static_cast<MockedTxPool*>(tx_pool_.get());
chainstate.SetMempool(&tx_pool);
LIMITED_WHILE(fuzzed_data_provider.remaining_bytes() > 0, 300)
{
Assert(!mempool_outpoints.empty());
std::vector<CTransactionRef> txs;
// Find something we may want to double-spend with two input single tx
std::optional<COutPoint> outpoint_to_rbf{fuzzed_data_provider.ConsumeBool() ? GetChildEvictingPrevout(tx_pool) : std::nullopt};
// Make small packages
const auto num_txs = outpoint_to_rbf ? 1 : fuzzed_data_provider.ConsumeIntegralInRange<size_t>(1, 4);
std::set<COutPoint> package_outpoints;
while (txs.size() < num_txs) {
// Create transaction to add to the mempool
txs.emplace_back([&] {
CMutableTransaction tx_mut;
tx_mut.version = CTransaction::CURRENT_VERSION;
tx_mut.nLockTime = 0;
// Last transaction in a package needs to be a child of parents to get further in validation
// so the last transaction to be generated(in a >1 package) must spend all package-made outputs
// Note that this test currently only spends package outputs in last transaction.
bool last_tx = num_txs > 1 && txs.size() == num_txs - 1;
const auto num_in = outpoint_to_rbf ? 2 :
last_tx ? fuzzed_data_provider.ConsumeIntegralInRange<int>(package_outpoints.size()/2 + 1, package_outpoints.size()) :
fuzzed_data_provider.ConsumeIntegralInRange<int>(1, 4);
const auto num_out = outpoint_to_rbf ? 1 : fuzzed_data_provider.ConsumeIntegralInRange<int>(1, 4);
auto& outpoints = last_tx ? package_outpoints : mempool_outpoints;
Assert((int)outpoints.size() >= num_in && num_in > 0);
CAmount amount_in{0};
for (int i = 0; i < num_in; ++i) {
// Pop random outpoint. We erase them to avoid double-spending
// while in this loop, but later add them back (unless last_tx).
auto pop = outpoints.begin();
std::advance(pop, fuzzed_data_provider.ConsumeIntegralInRange<size_t>(0, outpoints.size() - 1));
auto outpoint = *pop;
if (i == 0 && outpoint_to_rbf) {
outpoint = *outpoint_to_rbf;
outpoints.erase(outpoint);
} else {
outpoints.erase(pop);
}
// no need to update or erase from outpoints_value
amount_in += outpoints_value.at(outpoint);
// Create input
CTxIn in;
in.prevout = outpoint;
in.scriptWitness.stack = P2WSH_EMPTY_TRUE_STACK;
tx_mut.vin.push_back(in);
}
const auto amount_fee = fuzzed_data_provider.ConsumeIntegralInRange<CAmount>(0, amount_in);
const auto amount_out = (amount_in - amount_fee) / num_out;
for (int i = 0; i < num_out; ++i) {
tx_mut.vout.emplace_back(amount_out, P2WSH_EMPTY);
}
// Note output amounts can naturally drop to dust on their own.
if (!outpoint_to_rbf && fuzzed_data_provider.ConsumeBool()) {
uint32_t dust_index = fuzzed_data_provider.ConsumeIntegralInRange<uint32_t>(0, num_out);
tx_mut.vout.insert(tx_mut.vout.begin() + dust_index, CTxOut(0, P2WSH_EMPTY));
}
auto tx = MakeTransactionRef(tx_mut);
// Restore previously removed outpoints, except in-package outpoints (to allow RBF)
if (!last_tx) {
for (const auto& in : tx->vin) {
Assert(outpoints.insert(in.prevout).second);
}
// Cache the in-package outpoints being made
for (size_t i = 0; i < tx->vout.size(); ++i) {
package_outpoints.emplace(tx->GetHash(), i);
}
}
// We need newly-created values for the duration of this run
for (size_t i = 0; i < tx->vout.size(); ++i) {
outpoints_value[COutPoint(tx->GetHash(), i)] = tx->vout[i].nValue;
}
return tx;
}());
}
if (fuzzed_data_provider.ConsumeBool()) {
const auto& txid = fuzzed_data_provider.ConsumeBool() ?
txs.back()->GetHash() :
PickValue(fuzzed_data_provider, mempool_outpoints).hash;
const auto delta = fuzzed_data_provider.ConsumeIntegralInRange<CAmount>(-50 * COIN, +50 * COIN);
// We only prioritise out of mempool transactions since PrioritiseTransaction doesn't
// filter for ephemeral dust
if (tx_pool.exists(txid)) {
const auto tx_info{tx_pool.info(txid)};
if (GetDust(*tx_info.tx, tx_pool.m_opts.dust_relay_feerate).empty()) {
tx_pool.PrioritiseTransaction(txid, delta);
}
}
}
auto single_submit = txs.size() == 1;
const auto result_package = WITH_LOCK(::cs_main,
return ProcessNewPackage(chainstate, tx_pool, txs, /*test_accept=*/single_submit, /*client_maxfeerate=*/{}));
const auto res = WITH_LOCK(::cs_main, return AcceptToMemoryPool(chainstate, txs.back(), GetTime(),
/*bypass_limits=*/false, /*test_accept=*/!single_submit));
if (!single_submit && result_package.m_state.GetResult() != PackageValidationResult::PCKG_POLICY) {
// We don't know anything about the validity since transactions were randomly generated, so
// just use result_package.m_state here. This makes the expect_valid check meaningless, but
// we can still verify that the contents of m_tx_results are consistent with m_state.
const bool expect_valid{result_package.m_state.IsValid()};
Assert(!CheckPackageMempoolAcceptResult(txs, result_package, expect_valid, &tx_pool));
}
node.validation_signals->SyncWithValidationInterfaceQueue();
CheckMempoolEphemeralInvariants(tx_pool);
}
node.validation_signals->UnregisterSharedValidationInterface(outpoints_updater);
WITH_LOCK(::cs_main, tx_pool.check(chainstate.CoinsTip(), chainstate.m_chain.Height() + 1));
}
FUZZ_TARGET(tx_package_eval, .init = initialize_tx_pool)
{
SeedRandomStateForTest(SeedRand::ZEROS);
FuzzedDataProvider fuzzed_data_provider(buffer.data(), buffer.size());
const auto& node = g_setup->m_node;
auto& chainstate{static_cast<DummyChainState&>(node.chainman->ActiveChainstate())};
MockTime(fuzzed_data_provider, chainstate);
// All RBF-spendable outpoints outside of the unsubmitted package
std::set<COutPoint> mempool_outpoints;
std::unordered_map<COutPoint, CAmount, SaltedOutpointHasher> outpoints_value;
for (const auto& outpoint : g_outpoints_coinbase_init_mature) {
Assert(mempool_outpoints.insert(outpoint).second);
outpoints_value[outpoint] = 50 * COIN;
}
auto outpoints_updater = std::make_shared<OutpointsUpdater>(mempool_outpoints);
node.validation_signals->RegisterSharedValidationInterface(outpoints_updater);
auto tx_pool_{MakeMempool(fuzzed_data_provider, node)};
MockedTxPool& tx_pool = *static_cast<MockedTxPool*>(tx_pool_.get());
chainstate.SetMempool(&tx_pool);
LIMITED_WHILE(fuzzed_data_provider.remaining_bytes() > 0, 300)
{
Assert(!mempool_outpoints.empty());
std::vector<CTransactionRef> txs;
// Make packages of 1-to-26 transactions
const auto num_txs = fuzzed_data_provider.ConsumeIntegralInRange<size_t>(1, 26);
std::set<COutPoint> package_outpoints;
while (txs.size() < num_txs) {
// Create transaction to add to the mempool
txs.emplace_back([&] {
CMutableTransaction tx_mut;
tx_mut.version = fuzzed_data_provider.ConsumeBool() ? TRUC_VERSION : CTransaction::CURRENT_VERSION;
tx_mut.nLockTime = fuzzed_data_provider.ConsumeBool() ? 0 : fuzzed_data_provider.ConsumeIntegral<uint32_t>();
// Last transaction in a package needs to be a child of parents to get further in validation
// so the last transaction to be generated(in a >1 package) must spend all package-made outputs
// Note that this test currently only spends package outputs in last transaction.
bool last_tx = num_txs > 1 && txs.size() == num_txs - 1;
const auto num_in = last_tx ? package_outpoints.size() : fuzzed_data_provider.ConsumeIntegralInRange<int>(1, mempool_outpoints.size());
auto num_out = fuzzed_data_provider.ConsumeIntegralInRange<int>(1, mempool_outpoints.size() * 2);
auto& outpoints = last_tx ? package_outpoints : mempool_outpoints;
Assert(!outpoints.empty());
CAmount amount_in{0};
for (size_t i = 0; i < num_in; ++i) {
// Pop random outpoint. We erase them to avoid double-spending
// while in this loop, but later add them back (unless last_tx).
auto pop = outpoints.begin();
std::advance(pop, fuzzed_data_provider.ConsumeIntegralInRange<size_t>(0, outpoints.size() - 1));
const auto outpoint = *pop;
outpoints.erase(pop);
// no need to update or erase from outpoints_value
amount_in += outpoints_value.at(outpoint);
// Create input
const auto sequence = ConsumeSequence(fuzzed_data_provider);
const auto script_sig = CScript{};
const auto script_wit_stack = fuzzed_data_provider.ConsumeBool() ? P2WSH_EMPTY_TRUE_STACK : P2WSH_EMPTY_TWO_STACK;
CTxIn in;
in.prevout = outpoint;
in.nSequence = sequence;
in.scriptSig = script_sig;
in.scriptWitness.stack = script_wit_stack;
tx_mut.vin.push_back(in);
}
// Duplicate an input
bool dup_input = fuzzed_data_provider.ConsumeBool();
if (dup_input) {
tx_mut.vin.push_back(tx_mut.vin.back());
}
// Refer to a non-existent input
if (fuzzed_data_provider.ConsumeBool()) {
tx_mut.vin.emplace_back();
}
// Make a p2pk output to make sigops adjusted vsize to violate TRUC rules, potentially, which is never spent
if (last_tx && amount_in > 1000 && fuzzed_data_provider.ConsumeBool()) {
tx_mut.vout.emplace_back(1000, CScript() << std::vector<unsigned char>(33, 0x02) << OP_CHECKSIG);
// Don't add any other outputs.
num_out = 1;
amount_in -= 1000;
}
const auto amount_fee = fuzzed_data_provider.ConsumeIntegralInRange<CAmount>(0, amount_in);
const auto amount_out = (amount_in - amount_fee) / num_out;
for (int i = 0; i < num_out; ++i) {
tx_mut.vout.emplace_back(amount_out, P2WSH_EMPTY);
}
auto tx = MakeTransactionRef(tx_mut);
// Restore previously removed outpoints, except in-package outpoints
if (!last_tx) {
for (const auto& in : tx->vin) {
// It's a fake input, or a new input, or a duplicate
Assert(in == CTxIn() || outpoints.insert(in.prevout).second || dup_input);
}
// Cache the in-package outpoints being made
for (size_t i = 0; i < tx->vout.size(); ++i) {
package_outpoints.emplace(tx->GetHash(), i);
}
}
// We need newly-created values for the duration of this run
for (size_t i = 0; i < tx->vout.size(); ++i) {
outpoints_value[COutPoint(tx->GetHash(), i)] = tx->vout[i].nValue;
}
return tx;
}());
}
if (fuzzed_data_provider.ConsumeBool()) {
MockTime(fuzzed_data_provider, chainstate);
}
if (fuzzed_data_provider.ConsumeBool()) {
tx_pool.RollingFeeUpdate();
}
if (fuzzed_data_provider.ConsumeBool()) {
const auto& txid = fuzzed_data_provider.ConsumeBool() ?
txs.back()->GetHash() :
PickValue(fuzzed_data_provider, mempool_outpoints).hash;
const auto delta = fuzzed_data_provider.ConsumeIntegralInRange<CAmount>(-50 * COIN, +50 * COIN);
tx_pool.PrioritiseTransaction(txid, delta);
}
// Remember all added transactions
std::set<CTransactionRef> added;
auto txr = std::make_shared<TransactionsDelta>(added);
node.validation_signals->RegisterSharedValidationInterface(txr);
// When there are multiple transactions in the package, we call ProcessNewPackage(txs, test_accept=false)
// and AcceptToMemoryPool(txs.back(), test_accept=true). When there is only 1 transaction, we might flip it
// (the package is a test accept and ATMP is a submission).
auto single_submit = txs.size() == 1 && fuzzed_data_provider.ConsumeBool();
// Exercise client_maxfeerate logic
std::optional<CFeeRate> client_maxfeerate{};
if (fuzzed_data_provider.ConsumeBool()) {
client_maxfeerate = CFeeRate(fuzzed_data_provider.ConsumeIntegralInRange<CAmount>(-1, 50 * COIN), 100);
}
const auto result_package = WITH_LOCK(::cs_main,
return ProcessNewPackage(chainstate, tx_pool, txs, /*test_accept=*/single_submit, client_maxfeerate));
// Always set bypass_limits to false because it is not supported in ProcessNewPackage and
// can be a source of divergence.
const auto res = WITH_LOCK(::cs_main, return AcceptToMemoryPool(chainstate, txs.back(), GetTime(),
/*bypass_limits=*/false, /*test_accept=*/!single_submit));
const bool passed = res.m_result_type == MempoolAcceptResult::ResultType::VALID;
node.validation_signals->SyncWithValidationInterfaceQueue();
node.validation_signals->UnregisterSharedValidationInterface(txr);
// There is only 1 transaction in the package. We did a test-package-accept and a ATMP
if (single_submit) {
Assert(passed != added.empty());
Assert(passed == res.m_state.IsValid());
if (passed) {
Assert(added.size() == 1);
Assert(txs.back() == *added.begin());
}
} else if (result_package.m_state.GetResult() != PackageValidationResult::PCKG_POLICY) {
// We don't know anything about the validity since transactions were randomly generated, so
// just use result_package.m_state here. This makes the expect_valid check meaningless, but
// we can still verify that the contents of m_tx_results are consistent with m_state.
const bool expect_valid{result_package.m_state.IsValid()};
Assert(!CheckPackageMempoolAcceptResult(txs, result_package, expect_valid, &tx_pool));
} else {
// This is empty if it fails early checks, or "full" if transactions are looked at deeper
Assert(result_package.m_tx_results.size() == txs.size() || result_package.m_tx_results.empty());
}
CheckMempoolTRUCInvariants(tx_pool);
// Dust checks only make sense when dust is enforced
if (tx_pool.m_opts.require_standard) {
CheckMempoolEphemeralInvariants(tx_pool);
}
}
node.validation_signals->UnregisterSharedValidationInterface(outpoints_updater);
WITH_LOCK(::cs_main, tx_pool.check(chainstate.CoinsTip(), chainstate.m_chain.Height() + 1));
}
} // namespace