Files
bitcoin/src/test/rbf_tests.cpp
Suhas Daftuar a3c31dfd71 scripted-diff: rename AddToMempool -> TryAddToMempool
-BEGIN VERIFY SCRIPT-
find src/test -type f -exec sed -i 's/AddToMempool/TryAddToMempool/g' {} +
find src/bench -type f -exec sed -i 's/AddToMempool/TryAddToMempool/g' {} +
-END VERIFY SCRIPT-
2025-11-30 10:57:48 -05:00

562 lines
28 KiB
C++

// Copyright (c) 2021-2022 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 <common/system.h>
#include <policy/rbf.h>
#include <random.h>
#include <test/util/txmempool.h>
#include <txmempool.h>
#include <util/time.h>
#include <test/util/setup_common.h>
#include <boost/test/unit_test.hpp>
#include <optional>
#include <vector>
BOOST_FIXTURE_TEST_SUITE(rbf_tests, TestingSetup)
static inline CTransactionRef make_tx(const std::vector<CTransactionRef>& inputs,
const std::vector<CAmount>& output_values)
{
CMutableTransaction tx = CMutableTransaction();
tx.vin.resize(inputs.size());
tx.vout.resize(output_values.size());
for (size_t i = 0; i < inputs.size(); ++i) {
tx.vin[i].prevout.hash = inputs[i]->GetHash();
tx.vin[i].prevout.n = 0;
// Add a witness so wtxid != txid
CScriptWitness witness;
witness.stack.emplace_back(i + 10);
tx.vin[i].scriptWitness = witness;
}
for (size_t i = 0; i < output_values.size(); ++i) {
tx.vout[i].scriptPubKey = CScript() << OP_11 << OP_EQUAL;
tx.vout[i].nValue = output_values[i];
}
return MakeTransactionRef(tx);
}
static CTransactionRef add_descendants(const CTransactionRef& tx, int32_t num_descendants, CTxMemPool& pool)
EXCLUSIVE_LOCKS_REQUIRED(::cs_main, pool.cs)
{
AssertLockHeld(::cs_main);
AssertLockHeld(pool.cs);
TestMemPoolEntryHelper entry;
// Assumes this isn't already spent in mempool
auto tx_to_spend = tx;
for (int32_t i{0}; i < num_descendants; ++i) {
auto next_tx = make_tx(/*inputs=*/{tx_to_spend}, /*output_values=*/{(50 - i) * CENT});
TryAddToMempool(pool, entry.FromTx(next_tx));
BOOST_CHECK(pool.GetIter(next_tx->GetHash()).has_value());
tx_to_spend = next_tx;
}
// Return last created tx
return tx_to_spend;
}
BOOST_FIXTURE_TEST_CASE(rbf_helper_functions, TestChain100Setup)
{
CTxMemPool& pool = *Assert(m_node.mempool);
LOCK2(::cs_main, pool.cs);
TestMemPoolEntryHelper entry;
const CAmount low_fee{CENT/100};
const CAmount normal_fee{CENT/10};
const CAmount high_fee{CENT};
// Create a parent tx1 and child tx2 with normal fees:
const auto tx1 = make_tx(/*inputs=*/ {m_coinbase_txns[0]}, /*output_values=*/ {10 * COIN});
TryAddToMempool(pool, entry.Fee(normal_fee).FromTx(tx1));
const auto tx2 = make_tx(/*inputs=*/ {tx1}, /*output_values=*/ {995 * CENT});
TryAddToMempool(pool, entry.Fee(normal_fee).FromTx(tx2));
// Create a low-feerate parent tx3 and high-feerate child tx4 (cpfp)
const auto tx3 = make_tx(/*inputs=*/ {m_coinbase_txns[1]}, /*output_values=*/ {1099 * CENT});
TryAddToMempool(pool, entry.Fee(low_fee).FromTx(tx3));
const auto tx4 = make_tx(/*inputs=*/ {tx3}, /*output_values=*/ {999 * CENT});
TryAddToMempool(pool, entry.Fee(high_fee).FromTx(tx4));
// Create a parent tx5 and child tx6 where both have very low fees
const auto tx5 = make_tx(/*inputs=*/ {m_coinbase_txns[2]}, /*output_values=*/ {1099 * CENT});
TryAddToMempool(pool, entry.Fee(low_fee).FromTx(tx5));
const auto tx6 = make_tx(/*inputs=*/ {tx5}, /*output_values=*/ {1098 * CENT});
TryAddToMempool(pool, entry.Fee(low_fee).FromTx(tx6));
// Make tx6's modified fee much higher than its base fee. This should cause it to pass
// the fee-related checks despite being low-feerate.
pool.PrioritiseTransaction(tx6->GetHash(), 1 * COIN);
// Two independent high-feerate transactions, tx7 and tx8
const auto tx7 = make_tx(/*inputs=*/ {m_coinbase_txns[3]}, /*output_values=*/ {999 * CENT});
TryAddToMempool(pool, entry.Fee(high_fee).FromTx(tx7));
const auto tx8 = make_tx(/*inputs=*/ {m_coinbase_txns[4]}, /*output_values=*/ {999 * CENT});
TryAddToMempool(pool, entry.Fee(high_fee).FromTx(tx8));
// Will make these two parents of single child
const auto tx11 = make_tx(/*inputs=*/ {m_coinbase_txns[7]}, /*output_values=*/ {995 * CENT});
TryAddToMempool(pool, entry.Fee(normal_fee).FromTx(tx11));
const auto tx12 = make_tx(/*inputs=*/ {m_coinbase_txns[8]}, /*output_values=*/ {995 * CENT});
TryAddToMempool(pool, entry.Fee(normal_fee).FromTx(tx12));
// Will make two children of this single parent
const auto tx13 = make_tx(/*inputs=*/ {m_coinbase_txns[9]}, /*output_values=*/ {995 * CENT, 995 * CENT});
TryAddToMempool(pool, entry.Fee(normal_fee).FromTx(tx13));
const auto entry1_normal = pool.GetIter(tx1->GetHash()).value();
const auto entry2_normal = pool.GetIter(tx2->GetHash()).value();
const auto entry3_low = pool.GetIter(tx3->GetHash()).value();
const auto entry4_high = pool.GetIter(tx4->GetHash()).value();
const auto entry5_low = pool.GetIter(tx5->GetHash()).value();
const auto entry6_low_prioritised = pool.GetIter(tx6->GetHash()).value();
const auto entry7_high = pool.GetIter(tx7->GetHash()).value();
const auto entry8_high = pool.GetIter(tx8->GetHash()).value();
BOOST_CHECK_EQUAL(entry1_normal->GetFee(), normal_fee);
BOOST_CHECK_EQUAL(entry2_normal->GetFee(), normal_fee);
BOOST_CHECK_EQUAL(entry3_low->GetFee(), low_fee);
BOOST_CHECK_EQUAL(entry4_high->GetFee(), high_fee);
BOOST_CHECK_EQUAL(entry5_low->GetFee(), low_fee);
BOOST_CHECK_EQUAL(entry6_low_prioritised->GetFee(), low_fee);
BOOST_CHECK_EQUAL(entry7_high->GetFee(), high_fee);
BOOST_CHECK_EQUAL(entry8_high->GetFee(), high_fee);
CTxMemPool::setEntries set_12_normal{entry1_normal, entry2_normal};
CTxMemPool::setEntries set_34_cpfp{entry3_low, entry4_high};
CTxMemPool::setEntries set_56_low{entry5_low, entry6_low_prioritised};
CTxMemPool::setEntries set_78_high{entry7_high, entry8_high};
CTxMemPool::setEntries all_entries{entry1_normal, entry2_normal, entry3_low, entry4_high,
entry5_low, entry6_low_prioritised, entry7_high, entry8_high};
CTxMemPool::setEntries empty_set;
const auto unused_txid = Txid::FromUint256(GetRandHash());
// Tests for EntriesAndTxidsDisjoint
BOOST_CHECK(EntriesAndTxidsDisjoint(empty_set, {tx1->GetHash()}, unused_txid) == std::nullopt);
BOOST_CHECK(EntriesAndTxidsDisjoint(set_12_normal, {tx3->GetHash()}, unused_txid) == std::nullopt);
BOOST_CHECK(EntriesAndTxidsDisjoint({entry2_normal}, {tx2->GetHash()}, unused_txid).has_value());
BOOST_CHECK(EntriesAndTxidsDisjoint(set_12_normal, {tx1->GetHash()}, unused_txid).has_value());
BOOST_CHECK(EntriesAndTxidsDisjoint(set_12_normal, {tx2->GetHash()}, unused_txid).has_value());
// EntriesAndTxidsDisjoint does not calculate descendants of iters_conflicting; it uses whatever
// the caller passed in. As such, no error is returned even though entry2_normal is a descendant of tx1.
BOOST_CHECK(EntriesAndTxidsDisjoint({entry2_normal}, {tx1->GetHash()}, unused_txid) == std::nullopt);
// Tests for PaysForRBF
const CFeeRate incremental_relay_feerate{DEFAULT_INCREMENTAL_RELAY_FEE};
const CFeeRate higher_relay_feerate{2 * DEFAULT_INCREMENTAL_RELAY_FEE};
// Must pay at least as much as the original.
BOOST_CHECK(PaysForRBF(/*original_fees=*/high_fee,
/*replacement_fees=*/high_fee,
/*replacement_vsize=*/1,
/*relay_fee=*/CFeeRate(0),
/*txid=*/unused_txid)
== std::nullopt);
BOOST_CHECK(PaysForRBF(high_fee, high_fee - 1, 1, CFeeRate(0), unused_txid).has_value());
BOOST_CHECK(PaysForRBF(high_fee + 1, high_fee, 1, CFeeRate(0), unused_txid).has_value());
// Additional fees must cover the replacement's vsize at incremental relay fee
BOOST_CHECK(PaysForRBF(high_fee, high_fee + 1, 11, incremental_relay_feerate, unused_txid).has_value());
BOOST_CHECK(PaysForRBF(high_fee, high_fee + 1, 10, incremental_relay_feerate, unused_txid) == std::nullopt);
BOOST_CHECK(PaysForRBF(high_fee, high_fee + 2, 11, higher_relay_feerate, unused_txid).has_value());
BOOST_CHECK(PaysForRBF(high_fee, high_fee + 4, 20, higher_relay_feerate, unused_txid) == std::nullopt);
BOOST_CHECK(PaysForRBF(low_fee, high_fee, 99999999, incremental_relay_feerate, unused_txid).has_value());
BOOST_CHECK(PaysForRBF(low_fee, high_fee + 99999999, 99999999, incremental_relay_feerate, unused_txid) == std::nullopt);
}
BOOST_FIXTURE_TEST_CASE(rbf_conflicts_calculator, TestChain100Setup)
{
CTxMemPool& pool = *Assert(m_node.mempool);
LOCK2(::cs_main, pool.cs);
TestMemPoolEntryHelper entry;
const CAmount normal_fee{CENT/10};
// Create two parent transactions with 51 outputs each
const int NUM_OUTPUTS = 51;
std::vector<CAmount> output_values;
output_values.reserve(NUM_OUTPUTS);
for (int i = 0; i < NUM_OUTPUTS; ++i) {
output_values.push_back(1 * COIN);
}
const auto parent_tx_1 = make_tx(/*inputs=*/ {m_coinbase_txns[0]}, /*output_values=*/ output_values);
const auto parent_tx_2 = make_tx(/*inputs=*/ {m_coinbase_txns[1]}, /*output_values=*/ output_values);
TryAddToMempool(pool, entry.Fee(normal_fee).FromTx(parent_tx_1));
TryAddToMempool(pool, entry.Fee(normal_fee).FromTx(parent_tx_2));
std::vector<CTransactionRef> direct_children;
// Create individual spends of these outputs
for (const auto& parent_tx : {parent_tx_1, parent_tx_2}) {
for (auto i = 0; i < NUM_OUTPUTS; ++i) {
auto pretx = make_tx(/*inputs=*/ {parent_tx}, /*output_values=*/ {995 * CENT});
CMutableTransaction tx(*pretx);
tx.vin[0].prevout.n = i;
TryAddToMempool(pool, entry.Fee(normal_fee).FromTx(tx));
BOOST_CHECK(pool.GetIter(tx.GetHash()).has_value());
direct_children.push_back(MakeTransactionRef(tx));
}
}
// At this point, we should have 2 clusters in the mempool, each with 52
// transactions.
// parent_tx and all children are in one cluster, so we can have as many
// conflicts within this cluster as we want without violating the RBF conflicts
// limit.
const auto parent_entry_1 = pool.GetIter(parent_tx_1->GetHash()).value();
const auto parent_entry_2 = pool.GetIter(parent_tx_2->GetHash()).value();
const auto conflicting_transaction = make_tx({parent_tx_1, parent_tx_2}, {50 * CENT});
CTxMemPool::setEntries all_conflicts, dummy;
BOOST_CHECK(GetEntriesForConflicts(/*tx=*/ *conflicting_transaction.get(),
/*pool=*/ pool,
/*iters_conflicting=*/ {parent_entry_1, parent_entry_2},
/*all_conflicts=*/ all_conflicts) == std::nullopt);
dummy.clear();
// Conflicting directly with all those conflicts doesn't change anything.
BOOST_CHECK(GetEntriesForConflicts(/*tx=*/ *conflicting_transaction.get(),
/*pool=*/ pool,
/*iters_conflicting=*/ all_conflicts,
/*all_conflicts=*/ dummy) == std::nullopt);
BOOST_CHECK_EQUAL(all_conflicts.size(), dummy.size());
dummy.clear();
// If we mine the parent_tx's, then the clusters split (102 clusters).
pool.removeForBlock({parent_tx_1, parent_tx_2}, /* dummy */ 1);
// Add some descendants now to each of the direct children (we can do this now that the clusters have split).
for (const auto& child : direct_children) {
add_descendants(child, 10, pool);
}
// We can conflict with 100 different clusters, even if they have lots of transactions.
CTxMemPool::setEntries conflicts;
for (auto i = 0; i < 100; ++i) {
conflicts.insert(pool.GetIter(direct_children[i]->GetHash()).value());
}
BOOST_CHECK(GetEntriesForConflicts(/*tx=*/ *conflicting_transaction.get(),
/*pool=*/ pool,
/*iters_conflicting=*/ conflicts,
/*all_conflicts=*/ dummy) == std::nullopt);
// Conflicting with 1 more distinct cluster causes failure, however.
conflicts.insert(pool.GetIter(direct_children[100]->GetHash()).value());
BOOST_CHECK(GetEntriesForConflicts(/*tx=*/ *conflicting_transaction.get(),
/*pool=*/ pool,
/*iters_conflicting=*/ conflicts,
/*all_conflicts=*/ dummy).has_value());
}
BOOST_FIXTURE_TEST_CASE(improves_feerate, TestChain100Setup)
{
CTxMemPool& pool = *Assert(m_node.mempool);
LOCK2(::cs_main, pool.cs);
TestMemPoolEntryHelper entry;
const CAmount low_fee{CENT/100};
const CAmount normal_fee{CENT/10};
// low feerate parent with normal feerate child
const auto tx1 = make_tx(/*inputs=*/ {m_coinbase_txns[0], m_coinbase_txns[1]}, /*output_values=*/ {10 * COIN});
TryAddToMempool(pool, entry.Fee(low_fee).FromTx(tx1));
const auto tx2 = make_tx(/*inputs=*/ {tx1}, /*output_values=*/ {995 * CENT});
TryAddToMempool(pool, entry.Fee(normal_fee).FromTx(tx2));
const auto entry1 = pool.GetIter(tx1->GetHash()).value();
const auto tx1_fee = entry1->GetModifiedFee();
const auto entry2 = pool.GetIter(tx2->GetHash()).value();
const auto tx2_fee = entry2->GetModifiedFee();
// conflicting transactions
const auto tx1_conflict = make_tx(/*inputs=*/ {m_coinbase_txns[0], m_coinbase_txns[2]}, /*output_values=*/ {10 * COIN});
const auto tx3 = make_tx(/*inputs=*/ {tx1_conflict}, /*output_values=*/ {995 * CENT});
auto entry3 = entry.FromTx(tx3);
// Now test ImprovesFeerateDiagram with various levels of "package rbf" feerates
// It doesn't improve itself
auto changeset = pool.GetChangeSet();
changeset->StageRemoval(entry1);
changeset->StageRemoval(entry2);
changeset->StageAddition(tx1_conflict, tx1_fee, 0, 1, 0, false, 4, LockPoints());
changeset->StageAddition(tx3, tx2_fee, 0, 1, 0, false, 4, LockPoints());
const auto res1 = ImprovesFeerateDiagram(*changeset);
BOOST_CHECK(res1.has_value());
BOOST_CHECK(res1.value().first == DiagramCheckError::FAILURE);
BOOST_CHECK(res1.value().second == "insufficient feerate: does not improve feerate diagram");
// With one more satoshi it does
changeset.reset();
changeset = pool.GetChangeSet();
changeset->StageRemoval(entry1);
changeset->StageRemoval(entry2);
changeset->StageAddition(tx1_conflict, tx1_fee+1, 0, 1, 0, false, 4, LockPoints());
changeset->StageAddition(tx3, tx2_fee, 0, 1, 0, false, 4, LockPoints());
BOOST_CHECK(ImprovesFeerateDiagram(*changeset) == std::nullopt);
changeset.reset();
// With prioritisation of in-mempool conflicts, it affects the results of the comparison using the same args as just above
pool.PrioritiseTransaction(entry1->GetSharedTx()->GetHash(), /*nFeeDelta=*/1);
changeset = pool.GetChangeSet();
changeset->StageRemoval(entry1);
changeset->StageRemoval(entry2);
changeset->StageAddition(tx1_conflict, tx1_fee+1, 0, 1, 0, false, 4, LockPoints());
changeset->StageAddition(tx3, tx2_fee, 0, 1, 0, false, 4, LockPoints());
const auto res2 = ImprovesFeerateDiagram(*changeset);
BOOST_CHECK(res2.has_value());
BOOST_CHECK(res2.value().first == DiagramCheckError::FAILURE);
BOOST_CHECK(res2.value().second == "insufficient feerate: does not improve feerate diagram");
changeset.reset();
pool.PrioritiseTransaction(entry1->GetSharedTx()->GetHash(), /*nFeeDelta=*/-1);
// With fewer vbytes it does
CMutableTransaction tx4{entry3.GetTx()};
tx4.vin[0].scriptWitness = CScriptWitness(); // Clear out the witness, to reduce size
auto entry4 = entry.FromTx(MakeTransactionRef(tx4));
changeset = pool.GetChangeSet();
changeset->StageRemoval(entry1);
changeset->StageRemoval(entry2);
changeset->StageAddition(tx1_conflict, tx1_fee, 0, 1, 0, false, 4, LockPoints());
changeset->StageAddition(entry4.GetSharedTx(), tx2_fee, 0, 1, 0, false, 4, LockPoints());
BOOST_CHECK(ImprovesFeerateDiagram(*changeset) == std::nullopt);
changeset.reset();
// Adding a grandchild makes the cluster size 3, which is also calculable
const auto tx5 = make_tx(/*inputs=*/ {tx2}, /*output_values=*/ {995 * CENT});
TryAddToMempool(pool, entry.Fee(normal_fee).FromTx(tx5));
const auto entry5 = pool.GetIter(tx5->GetHash()).value();
changeset = pool.GetChangeSet();
changeset->StageRemoval(entry1);
changeset->StageRemoval(entry2);
changeset->StageRemoval(entry5);
changeset->StageAddition(tx1_conflict, tx1_fee, 0, 1, 0, false, 4, LockPoints());
changeset->StageAddition(entry4.GetSharedTx(), tx2_fee + entry5->GetModifiedFee() + 1, 0, 1, 0, false, 4, LockPoints());
const auto res3 = ImprovesFeerateDiagram(*changeset);
BOOST_CHECK(res3 == std::nullopt);
}
BOOST_FIXTURE_TEST_CASE(calc_feerate_diagram_rbf, TestChain100Setup)
{
CTxMemPool& pool = *Assert(m_node.mempool);
LOCK2(::cs_main, pool.cs);
TestMemPoolEntryHelper entry;
const CAmount low_fee{CENT/100};
const CAmount high_fee{CENT};
// low -> high -> medium fee transactions that would result in two chunks together since they
// are all same size
const auto low_tx = make_tx(/*inputs=*/ {m_coinbase_txns[0]}, /*output_values=*/ {10 * COIN});
TryAddToMempool(pool, entry.Fee(low_fee).FromTx(low_tx));
const auto entry_low = pool.GetIter(low_tx->GetHash()).value();
const auto low_size = entry_low->GetAdjustedWeight();
const auto replacement_tx = make_tx(/*inputs=*/ {m_coinbase_txns[0]}, /*output_values=*/ {9 * COIN});
auto entry_replacement = entry.FromTx(replacement_tx);
// Replacement of size 1
{
auto changeset = pool.GetChangeSet();
changeset->StageRemoval(entry_low);
changeset->StageAddition(replacement_tx, 0, 0, 1, 0, false, 4, LockPoints());
const auto replace_one{changeset->CalculateChunksForRBF()};
BOOST_CHECK(replace_one.has_value());
std::vector<FeeFrac> expected_old_chunks{{low_fee, low_size}};
BOOST_CHECK(replace_one->first == expected_old_chunks);
std::vector<FeeFrac> expected_new_chunks{{0, entry_replacement.GetAdjustedWeight()}};
BOOST_CHECK(replace_one->second == expected_new_chunks);
}
// Non-zero replacement fee/size
{
auto changeset = pool.GetChangeSet();
changeset->StageRemoval(entry_low);
changeset->StageAddition(replacement_tx, high_fee, 0, 1, 0, false, 4, LockPoints());
const auto replace_one_fee{changeset->CalculateChunksForRBF()};
BOOST_CHECK(replace_one_fee.has_value());
std::vector<FeeFrac> expected_old_diagram{{low_fee, low_size}};
BOOST_CHECK(replace_one_fee->first == expected_old_diagram);
std::vector<FeeFrac> expected_new_diagram{{high_fee, entry_replacement.GetAdjustedWeight()}};
BOOST_CHECK(replace_one_fee->second == expected_new_diagram);
}
// Add a second transaction to the cluster that will make a single chunk, to be evicted in the RBF
const auto high_tx = make_tx(/*inputs=*/ {low_tx}, /*output_values=*/ {995 * CENT});
TryAddToMempool(pool, entry.Fee(high_fee).FromTx(high_tx));
const auto entry_high = pool.GetIter(high_tx->GetHash()).value();
const auto high_size = entry_high->GetAdjustedWeight();
{
auto changeset = pool.GetChangeSet();
changeset->StageRemoval(entry_low);
changeset->StageRemoval(entry_high);
changeset->StageAddition(replacement_tx, high_fee, 0, 1, 0, false, 4, LockPoints());
const auto replace_single_chunk{changeset->CalculateChunksForRBF()};
BOOST_CHECK(replace_single_chunk.has_value());
std::vector<FeeFrac> expected_old_chunks{{low_fee + high_fee, low_size + high_size}};
BOOST_CHECK(replace_single_chunk->first == expected_old_chunks);
std::vector<FeeFrac> expected_new_chunks{{high_fee, entry_replacement.GetAdjustedWeight()}};
BOOST_CHECK(replace_single_chunk->second == expected_new_chunks);
}
// Conflict with the 2nd tx, resulting in new diagram with three entries
{
auto changeset = pool.GetChangeSet();
changeset->StageRemoval(entry_high);
changeset->StageAddition(replacement_tx, high_fee, 0, 1, 0, false, 4, LockPoints());
const auto replace_cpfp_child{changeset->CalculateChunksForRBF()};
BOOST_CHECK(replace_cpfp_child.has_value());
std::vector<FeeFrac> expected_old_chunks{{low_fee + high_fee, low_size + high_size}};
BOOST_CHECK(replace_cpfp_child->first == expected_old_chunks);
std::vector<FeeFrac> expected_new_chunks{{high_fee, entry_replacement.GetAdjustedWeight()}, {low_fee, low_size}};
BOOST_CHECK(replace_cpfp_child->second == expected_new_chunks);
}
// Make a size 2 cluster that is itself two chunks; evict both txns
const auto high_tx_2 = make_tx(/*inputs=*/ {m_coinbase_txns[1]}, /*output_values=*/ {10 * COIN});
TryAddToMempool(pool, entry.Fee(high_fee).FromTx(high_tx_2));
const auto entry_high_2 = pool.GetIter(high_tx_2->GetHash()).value();
const auto high_size_2 = entry_high_2->GetAdjustedWeight();
const auto low_tx_2 = make_tx(/*inputs=*/ {high_tx_2}, /*output_values=*/ {9 * COIN});
TryAddToMempool(pool, entry.Fee(low_fee).FromTx(low_tx_2));
const auto entry_low_2 = pool.GetIter(low_tx_2->GetHash()).value();
const auto low_size_2 = entry_low_2->GetAdjustedWeight();
{
auto changeset = pool.GetChangeSet();
changeset->StageRemoval(entry_high_2);
changeset->StageRemoval(entry_low_2);
changeset->StageAddition(replacement_tx, high_fee, 0, 1, 0, false, 4, LockPoints());
const auto replace_two_chunks_single_cluster{changeset->CalculateChunksForRBF()};
BOOST_CHECK(replace_two_chunks_single_cluster.has_value());
std::vector<FeeFrac> expected_old_chunks{{high_fee, high_size_2}, {low_fee, low_size_2}};
BOOST_CHECK(replace_two_chunks_single_cluster->first == expected_old_chunks);
std::vector<FeeFrac> expected_new_chunks{{high_fee, low_size_2}};
BOOST_CHECK(replace_two_chunks_single_cluster->second == expected_new_chunks);
}
// You can have more than two direct conflicts
const auto conflict_1 = make_tx(/*inputs=*/ {m_coinbase_txns[2]}, /*output_values=*/ {10 * COIN});
TryAddToMempool(pool, entry.Fee(low_fee).FromTx(conflict_1));
const auto conflict_1_entry = pool.GetIter(conflict_1->GetHash()).value();
const auto conflict_2 = make_tx(/*inputs=*/ {m_coinbase_txns[3]}, /*output_values=*/ {10 * COIN});
TryAddToMempool(pool, entry.Fee(low_fee).FromTx(conflict_2));
const auto conflict_2_entry = pool.GetIter(conflict_2->GetHash()).value();
const auto conflict_3 = make_tx(/*inputs=*/ {m_coinbase_txns[4]}, /*output_values=*/ {10 * COIN});
TryAddToMempool(pool, entry.Fee(low_fee).FromTx(conflict_3));
const auto conflict_3_entry = pool.GetIter(conflict_3->GetHash()).value();
{
auto changeset = pool.GetChangeSet();
changeset->StageRemoval(conflict_1_entry);
changeset->StageRemoval(conflict_2_entry);
changeset->StageRemoval(conflict_3_entry);
changeset->StageAddition(replacement_tx, high_fee, 0, 1, 0, false, 4, LockPoints());
const auto replace_multiple_clusters{changeset->CalculateChunksForRBF()};
BOOST_CHECK(replace_multiple_clusters.has_value());
BOOST_CHECK(replace_multiple_clusters->first.size() == 3);
BOOST_CHECK(replace_multiple_clusters->second.size() == 1);
}
// Add a child transaction to conflict_1 and make it cluster size 2, two chunks due to same feerate
const auto conflict_1_child = make_tx(/*inputs=*/{conflict_1}, /*output_values=*/ {995 * CENT});
TryAddToMempool(pool, entry.Fee(low_fee).FromTx(conflict_1_child));
const auto conflict_1_child_entry = pool.GetIter(conflict_1_child->GetHash()).value();
{
auto changeset = pool.GetChangeSet();
changeset->StageRemoval(conflict_1_entry);
changeset->StageRemoval(conflict_2_entry);
changeset->StageRemoval(conflict_3_entry);
changeset->StageRemoval(conflict_1_child_entry);
changeset->StageAddition(replacement_tx, high_fee, 0, 1, 0, false, 4, LockPoints());
const auto replace_multiple_clusters_2{changeset->CalculateChunksForRBF()};
BOOST_CHECK(replace_multiple_clusters_2.has_value());
BOOST_CHECK(replace_multiple_clusters_2->first.size() == 4);
BOOST_CHECK(replace_multiple_clusters_2->second.size() == 1);
}
}
BOOST_AUTO_TEST_CASE(feerate_chunks_utilities)
{
// Sanity check the correctness of the feerate chunks comparison.
// A strictly better case.
std::vector<FeeFrac> old_chunks{{{950, 300}, {100, 100}}};
std::vector<FeeFrac> new_chunks{{{1000, 300}, {50, 100}}};
BOOST_CHECK(std::is_lt(CompareChunks(old_chunks, new_chunks)));
BOOST_CHECK(std::is_gt(CompareChunks(new_chunks, old_chunks)));
// Incomparable diagrams
old_chunks = {{950, 300}, {100, 100}};
new_chunks = {{1000, 300}, {0, 100}};
BOOST_CHECK(CompareChunks(old_chunks, new_chunks) == std::partial_ordering::unordered);
BOOST_CHECK(CompareChunks(new_chunks, old_chunks) == std::partial_ordering::unordered);
// Strictly better but smaller size.
old_chunks = {{950, 300}, {100, 100}};
new_chunks = {{1100, 300}};
BOOST_CHECK(std::is_lt(CompareChunks(old_chunks, new_chunks)));
BOOST_CHECK(std::is_gt(CompareChunks(new_chunks, old_chunks)));
// New diagram is strictly better due to the first chunk, even though
// second chunk contributes no fees
old_chunks = {{950, 300}, {100, 100}};
new_chunks = {{1100, 100}, {0, 100}};
BOOST_CHECK(std::is_lt(CompareChunks(old_chunks, new_chunks)));
BOOST_CHECK(std::is_gt(CompareChunks(new_chunks, old_chunks)));
// Feerate of first new chunk is better with, but second chunk is worse
old_chunks = {{950, 300}, {100, 100}};
new_chunks = {{750, 100}, {249, 250}, {151, 650}};
BOOST_CHECK(CompareChunks(old_chunks, new_chunks) == std::partial_ordering::unordered);
BOOST_CHECK(CompareChunks(new_chunks, old_chunks) == std::partial_ordering::unordered);
// If we make the second chunk slightly better, the new diagram now wins.
old_chunks = {{950, 300}, {100, 100}};
new_chunks = {{750, 100}, {250, 250}, {150, 150}};
BOOST_CHECK(std::is_lt(CompareChunks(old_chunks, new_chunks)));
BOOST_CHECK(std::is_gt(CompareChunks(new_chunks, old_chunks)));
// Identical diagrams, cannot be strictly better
old_chunks = {{950, 300}, {100, 100}};
new_chunks = {{950, 300}, {100, 100}};
BOOST_CHECK(std::is_eq(CompareChunks(old_chunks, new_chunks)));
BOOST_CHECK(std::is_eq(CompareChunks(new_chunks, old_chunks)));
// Same aggregate fee, but different total size (trigger single tail fee check step)
old_chunks = {{950, 300}, {100, 99}};
new_chunks = {{950, 300}, {100, 100}};
// No change in evaluation when tail check needed.
BOOST_CHECK(std::is_gt(CompareChunks(old_chunks, new_chunks)));
BOOST_CHECK(std::is_lt(CompareChunks(new_chunks, old_chunks)));
// Trigger multiple tail fee check steps
old_chunks = {{950, 300}, {100, 99}};
new_chunks = {{950, 300}, {100, 100}, {0, 1}, {0, 1}};
BOOST_CHECK(std::is_gt(CompareChunks(old_chunks, new_chunks)));
BOOST_CHECK(std::is_lt(CompareChunks(new_chunks, old_chunks)));
// Multiple tail fee check steps, unordered result
new_chunks = {{950, 300}, {100, 100}, {0, 1}, {0, 1}, {1, 1}};
BOOST_CHECK(CompareChunks(old_chunks, new_chunks) == std::partial_ordering::unordered);
BOOST_CHECK(CompareChunks(new_chunks, old_chunks) == std::partial_ordering::unordered);
}
BOOST_AUTO_TEST_SUITE_END()