fuzz: Add fuzzer for block index

This fuzz target creates arbitrary tree-like structure of indices,
simulating the following events:
- Adding a header to the block tree db
- Receiving the full block (may be valid or not)
- Reorging to a new chain tip (possibly encountering invalid blocks on
  the way)
- pruning
The test skips all actual validation of header/ block / transaction data
by just simulating the outcome, and also doesn't interact with the data directory.

The main goal is to test the integrity of the block index tree in
all fuzzed constellations, by calling CheckBlockIndex()
at the end of each iteration.

src/test/fuzz/CMakeLists.txt

@@ -19,6 +19,7 @@ add_executable(fuzz
   block.cpp
   block_header.cpp
   block_index.cpp
+  block_index_tree.cpp
   blockfilter.cpp
   bloom_filter.cpp
   buffered_file.cpp

src/test/fuzz/block_index_tree.cpp

@@ -0,0 +1,209 @@
+// Copyright (c) 2020-2022 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 <chain.h>
+#include <chainparams.h>
+#include <flatfile.h>
+#include <primitives/block.h>
+#include <primitives/transaction.h>
+#include <test/fuzz/FuzzedDataProvider.h>
+#include <test/fuzz/fuzz.h>
+#include <test/fuzz/util.h>
+#include <test/util/setup_common.h>
+#include <test/util/validation.h>
+#include <validation.h>
+
+#include <ranges>
+#include <vector>
+
+const TestingSetup* g_setup;
+
+CBlockHeader ConsumeBlockHeader(FuzzedDataProvider& provider, uint256 prev_hash, int& nonce_counter)
+{
+    CBlockHeader header;
+    header.nVersion = provider.ConsumeIntegral<decltype(header.nVersion)>();
+    header.hashPrevBlock = prev_hash;
+    header.hashMerkleRoot = uint256{}; // never used
+    header.nTime = provider.ConsumeIntegral<decltype(header.nTime)>();
+    header.nBits = Params().GenesisBlock().nBits; // not fuzzed because not used (validation is mocked).
+    header.nNonce = nonce_counter++;              // prevent creating multiple block headers with the same hash
+    return header;
+}
+
+void initialize_block_index_tree()
+{
+    static const auto testing_setup = MakeNoLogFileContext<const TestingSetup>();
+    g_setup = testing_setup.get();
+}
+
+FUZZ_TARGET(block_index_tree, .init = initialize_block_index_tree)
+{
+    FuzzedDataProvider fuzzed_data_provider(buffer.data(), buffer.size());
+    SetMockTime(ConsumeTime(fuzzed_data_provider));
+    auto& chainman = static_cast<TestChainstateManager&>(*g_setup->m_node.chainman);
+    auto& blockman = static_cast<TestBlockManager&>(chainman.m_blockman);
+    CBlockIndex* genesis = chainman.ActiveChainstate().m_chain[0];
+    int nonce_counter = 0;
+    std::vector<CBlockIndex*> blocks;
+    blocks.push_back(genesis);
+    bool abort_run{false};
+
+    LIMITED_WHILE(fuzzed_data_provider.ConsumeBool(), 1000)
+    {
+        if (abort_run) break;
+        CallOneOf(
+            fuzzed_data_provider,
+            [&] {
+                // Receive a header building on an existing valid one. This assumes headers are valid, so PoW is not relevant here.
+                LOCK(cs_main);
+                CBlockIndex* prev_block = PickValue(fuzzed_data_provider, blocks);
+                if (!(prev_block->nStatus & BLOCK_FAILED_MASK)) {
+                    CBlockHeader header = ConsumeBlockHeader(fuzzed_data_provider, prev_block->GetBlockHash(), nonce_counter);
+                    CBlockIndex* index = blockman.AddToBlockIndex(header, chainman.m_best_header);
+                    assert(index->nStatus & BLOCK_VALID_TREE);
+                    assert(index->pprev == prev_block);
+                    blocks.push_back(index);
+                }
+            },
+            [&] {
+                // Receive a full block (valid or invalid) for an existing header, but don't attempt to connect it yet
+                LOCK(cs_main);
+                CBlockIndex* index = PickValue(fuzzed_data_provider, blocks);
+                // Must be new to us and not known to be invalid (e.g. because of an invalid ancestor).
+                if (index->nTx == 0 && !(index->nStatus & BLOCK_FAILED_MASK)) {
+                    if (fuzzed_data_provider.ConsumeBool()) { // Invalid
+                        BlockValidationState state;
+                        state.Invalid(BlockValidationResult::BLOCK_CONSENSUS, "consensus-invalid");
+                        chainman.InvalidBlockFound(index, state);
+                    } else {
+                        size_t nTx = fuzzed_data_provider.ConsumeIntegralInRange<size_t>(1, 1000);
+                        CBlock block; // Dummy block, so that ReceivedBlockTransactions can infer a nTx value.
+                        block.vtx = std::vector<CTransactionRef>(nTx);
+                        FlatFilePos pos(0, fuzzed_data_provider.ConsumeIntegralInRange<int>(1, 1000));
+                        chainman.ReceivedBlockTransactions(block, index, pos);
+                        assert(index->nStatus & BLOCK_VALID_TRANSACTIONS);
+                        assert(index->nStatus & BLOCK_HAVE_DATA);
+                    }
+                }
+            },
+            [&] {
+                // Simplified ActivateBestChain(): Try to move to a chain with more work - with the possibility of finding blocks to be invalid on the way
+                LOCK(cs_main);
+                auto& chain = chainman.ActiveChain();
+                CBlockIndex* old_tip = chain.Tip();
+                assert(old_tip);
+                do {
+                    CBlockIndex* best_tip = chainman.FindMostWorkChain();
+                    assert(best_tip);                   // Should at least return current tip
+                    if (best_tip == chain.Tip()) break; // Nothing to do
+                    // Rewind chain to forking point
+                    const CBlockIndex* fork = chain.FindFork(best_tip);
+                    // If we can't go back to the fork point due to pruned data, abort this run. In reality, a pruned node would also currently just crash in this scenario.
+                    // This is very unlikely to happen due to the minimum pruning threshold of 550MiB.
+                    CBlockIndex* it = chain.Tip();
+                    while (it && it->nHeight != fork->nHeight) {
+                        if (!(it->nStatus & BLOCK_HAVE_UNDO)) {
+                            assert(blockman.m_have_pruned);
+                            abort_run = true;
+                            return;
+                        }
+                        it = it->pprev;
+                    }
+                    chain.SetTip(*chain[fork->nHeight]);
+
+                    // Prepare new blocks to connect
+                    std::vector<CBlockIndex*> to_connect;
+                    it = best_tip;
+                    while (it && it->nHeight != fork->nHeight) {
+                        to_connect.push_back(it);
+                        it = it->pprev;
+                    }
+                    // Connect blocks, possibly fail
+                    for (CBlockIndex* block : to_connect | std::views::reverse) {
+                        assert(!(block->nStatus & BLOCK_FAILED_MASK));
+                        assert(block->nStatus & BLOCK_HAVE_DATA);
+                        if (!block->IsValid(BLOCK_VALID_SCRIPTS)) {
+                            if (fuzzed_data_provider.ConsumeBool()) { // Invalid
+                                BlockValidationState state;
+                                state.Invalid(BlockValidationResult::BLOCK_CONSENSUS, "consensus-invalid");
+                                chainman.InvalidBlockFound(block, state);
+                                // This results in duplicate calls to InvalidChainFound, but mirrors the behavior in validation
+                                chainman.InvalidChainFound(to_connect.front());
+                                break;
+                            } else {
+                                block->RaiseValidity(BLOCK_VALID_SCRIPTS);
+                                block->nStatus |= BLOCK_HAVE_UNDO;
+                            }
+                        }
+                        chain.SetTip(*block);
+                        chainman.ActiveChainstate().PruneBlockIndexCandidates();
+                        // ActivateBestChainStep may release cs_main / not connect all blocks in one go - but only if we have at least as much chain work as we had at the start.
+                        if (block->nChainWork > old_tip->nChainWork && fuzzed_data_provider.ConsumeBool()) {
+                            break;
+                        }
+                    }
+                } while (node::CBlockIndexWorkComparator()(chain.Tip(), old_tip));
+                assert(chain.Tip()->nChainWork >= old_tip->nChainWork);
+            },
+            [&] {
+                // Prune chain - dealing with block files is beyond the scope of this test, so just prune random blocks, making no assumptions
+                // about what blocks are pruned together because they are in the same block file.
+                // Also don't prune blocks outside of the chain for now - this would make the fuzzer crash because of the problem described in
+                // https://github.com/bitcoin/bitcoin/issues/31512
+                LOCK(cs_main);
+                auto& chain = chainman.ActiveChain();
+                int prune_height = fuzzed_data_provider.ConsumeIntegralInRange<int>(0, chain.Height());
+                CBlockIndex* prune_block{chain[prune_height]};
+                if (prune_block != chain.Tip() && (prune_block->nStatus & BLOCK_HAVE_DATA)) {
+                    blockman.m_have_pruned = true;
+                    prune_block->nStatus &= ~BLOCK_HAVE_DATA;
+                    prune_block->nStatus &= ~BLOCK_HAVE_UNDO;
+                    prune_block->nFile = 0;
+                    prune_block->nDataPos = 0;
+                    prune_block->nUndoPos = 0;
+                    auto range = blockman.m_blocks_unlinked.equal_range(prune_block->pprev);
+                    while (range.first != range.second) {
+                        std::multimap<CBlockIndex*, CBlockIndex*>::iterator _it = range.first;
+                        range.first++;
+                        if (_it->second == prune_block) {
+                            blockman.m_blocks_unlinked.erase(_it);
+                        }
+                    }
+                }
+            });
+    }
+    if (!abort_run) {
+        chainman.CheckBlockIndex();
+    }
+
+    // clean up global state changed by last iteration and prepare for next iteration
+    {
+        LOCK(cs_main);
+        genesis->nStatus |= BLOCK_HAVE_DATA;
+        genesis->nStatus |= BLOCK_HAVE_UNDO;
+        chainman.m_best_header = genesis;
+        chainman.ResetBestInvalid();
+        chainman.nBlockSequenceId = 2;
+        chainman.ActiveChain().SetTip(*genesis);
+        chainman.ActiveChainstate().setBlockIndexCandidates.clear();
+        chainman.m_cached_finished_ibd = false;
+        blockman.m_blocks_unlinked.clear();
+        blockman.m_have_pruned = false;
+        blockman.CleanupForFuzzing();
+        // Delete all blocks but Genesis from block index
+        uint256 genesis_hash = genesis->GetBlockHash();
+        for (auto it = blockman.m_block_index.begin(); it != blockman.m_block_index.end();) {
+            if (it->first != genesis_hash) {
+                it = blockman.m_block_index.erase(it);
+            } else {
+                ++it;
+            }
+        }
+        chainman.ActiveChainstate().TryAddBlockIndexCandidate(genesis);
+        assert(blockman.m_block_index.size() == 1);
+        assert(chainman.ActiveChainstate().setBlockIndexCandidates.size() == 1);
+        assert(chainman.ActiveChain().Height() == 0);
+    }
+}