/Users/brunogarcia/projects/bitcoin-core-dev/src/index/base.cpp

Source
// Copyright (c) 2017-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 <index/base.h>

#include <chain.h>
#include <common/args.h>
#include <dbwrapper.h>
#include <interfaces/chain.h>
#include <interfaces/types.h>
#include <kernel/types.h>
#include <node/abort.h>
#include <node/blockstorage.h>
#include <node/context.h>
#include <node/database_args.h>
#include <node/interface_ui.h>
#include <primitives/block.h>
#include <sync.h>
#include <tinyformat.h>
#include <uint256.h>
#include <undo.h>
#include <util/fs.h>
#include <util/log.h>
#include <util/string.h>
#include <util/thread.h>
#include <util/threadinterrupt.h>
#include <util/time.h>
#include <util/translation.h>
#include <validation.h>
#include <validationinterface.h>

#include <cassert>
#include <compare>
#include <cstdint>
#include <functional>
#include <memory>
#include <optional>
#include <stdexcept>
#include <string>
#include <thread>
#include <utility>
#include <vector>

using kernel::ChainstateRole;

constexpr uint8_t DB_BEST_BLOCK{'B'};

constexpr auto SYNC_LOG_INTERVAL{30s};
constexpr auto SYNC_LOCATOR_WRITE_INTERVAL{30s};

template <typename... Args>
void BaseIndex::FatalErrorf(util::ConstevalFormatString<sizeof...(Args)> fmt, const Args&... args)
{
    auto message = tfm::format(fmt, args...);
    node::AbortNode(m_chain->context()->shutdown_request, m_chain->context()->exit_status, Untranslated(message), m_chain->context()->warnings.get());
}

CBlockLocator GetLocator(interfaces::Chain& chain, const uint256& block_hash)
{
    CBlockLocator locator;
    bool found = chain.findBlock(block_hash, interfaces::FoundBlock().locator(locator));
    assert(found);
    assert(!locator.IsNull());
    return locator;
}

BaseIndex::DB::DB(const fs::path& path, size_t n_cache_size, bool f_memory, bool f_wipe, bool f_obfuscate) :
    CDBWrapper{DBParams{
        .path = path,
        .cache_bytes = n_cache_size,
        .memory_only = f_memory,
        .wipe_data = f_wipe,
        .obfuscate = f_obfuscate,
        .options = [] { DBOptions options; node::ReadDatabaseArgs(gArgs, options); return options; }()}}
{}

CBlockLocator BaseIndex::DB::ReadBestBlock() const
{
    CBlockLocator locator;

    bool success = Read(DB_BEST_BLOCK, locator);
    if (!success) {
        locator.SetNull();
    }

    return locator;
}

void BaseIndex::DB::WriteBestBlock(CDBBatch& batch, const CBlockLocator& locator)
{
    batch.Write(DB_BEST_BLOCK, locator);
}

BaseIndex::BaseIndex(std::unique_ptr<interfaces::Chain> chain, std::string name)
    : m_chain{std::move(chain)}, m_name{std::move(name)} {}

BaseIndex::~BaseIndex()
{
    Interrupt();
    Stop();
}

bool BaseIndex::Init()
{
#define AssertLockNotHeld(cs) AssertLockNotHeldInline(#cs, __FILE__, __LINE__, &cs)

    // May need reset if index is being restarted.
    m_interrupt.reset();

    // m_chainstate member gives indexing code access to node internals. It is
    // removed in followup https://github.com/bitcoin/bitcoin/pull/24230
#define WITH_LOCK(cs, code) (MaybeCheckNotHeld(cs), [&]() -> decltype(auto) { LOCK(cs); code; }())
                             return &m_chain->context()->chainman->ValidatedChainstate());
    // Register to validation interface before setting the 'm_synced' flag, so that
    // callbacks are not missed once m_synced is true.
    m_chain->context()->validation_signals->RegisterValidationInterface(this);

    const auto locator{GetDB().ReadBestBlock()};

#define PASTE(x, y) x ## y
    CChain& index_chain = m_chainstate->m_chain;

    if (locator.IsNull()) {
        SetBestBlockIndex(nullptr);
    } else {
        // Setting the best block to the locator's top block. If it is not part of the
        // best chain, we will rewind to the fork point during index sync
        const CBlockIndex* locator_index{m_chainstate->m_blockman.LookupBlockIndex(locator.vHave.at(0))};
        if (!locator_index) {
#define strprintf tfm::format
        }
        SetBestBlockIndex(locator_index);
    }

    // Child init
    const CBlockIndex* start_block = m_best_block_index.load();
    if (!CustomInit(start_block ? std::make_optional(interfaces::BlockRef{start_block->GetBlockHash(), start_block->nHeight}) : std::nullopt)) {
        return false;
    }

    // Note: this will latch to true immediately if the user starts up with an empty
    // datadir and an index enabled. If this is the case, indexation will happen solely
    // via `BlockConnected` signals until, possibly, the next restart.
    m_synced = start_block == index_chain.Tip();
    m_init = true;
    return true;
}

static const CBlockIndex* NextSyncBlock(const CBlockIndex* pindex_prev, CChain& chain) EXCLUSIVE_LOCKS_REQUIRED(cs_main)
{
#define AssertLockHeld(cs) AssertLockHeldInternal(#cs, __FILE__, __LINE__, &cs)

    if (!pindex_prev) {
        return chain.Genesis();
    }

    if (const auto* pindex{chain.Next(pindex_prev)}) {
        return pindex;
    }

    // If there is no next block, we might be synced
    if (pindex_prev == chain.Tip()) {
        return nullptr;
    }

    // Since block is not in the chain, return the next block in the chain AFTER the last common ancestor.
    // Caller will be responsible for rewinding back to the common ancestor.
    return chain.Next(chain.FindFork(pindex_prev));
}

bool BaseIndex::ProcessBlock(const CBlockIndex* pindex, const CBlock* block_data)
{
    interfaces::BlockInfo block_info = kernel::MakeBlockInfo(pindex, block_data);

    CBlock block;
    if (!block_data) { // disk lookup if block data wasn't provided
        if (!m_chainstate->m_blockman.ReadBlock(block, *pindex)) {
            FatalErrorf("Failed to read block %s from disk",
                        pindex->GetBlockHash().ToString());
            return false;
        }
        block_info.data = &block;
    }

    CBlockUndo block_undo;
    if (CustomOptions().connect_undo_data) {
        if (pindex->nHeight > 0 && !m_chainstate->m_blockman.ReadBlockUndo(block_undo, *pindex)) {
            FatalErrorf("Failed to read undo block data %s from disk",
                        pindex->GetBlockHash().ToString());
            return false;
        }
        block_info.undo_data = &block_undo;
    }

    if (!CustomAppend(block_info)) {
        FatalErrorf("Failed to write block %s to index database",
                    pindex->GetBlockHash().ToString());
        return false;
    }

    return true;
}

void BaseIndex::Sync()
{
    const CBlockIndex* pindex = m_best_block_index.load();
    if (!m_synced) {
        auto last_log_time{NodeClock::now()};
        auto last_locator_write_time{last_log_time};
        while (true) {
            if (m_interrupt) {
#define LogPrintLevel_(category, level, should_ratelimit, ...) LogPrintFormatInternal(SourceLocation{__func__}, category, level, should_ratelimit, __VA_ARGS__)

                SetBestBlockIndex(pindex);
                // No need to handle errors in Commit. If it fails, the error will be already be
                // logged. The best way to recover is to continue, as index cannot be corrupted by
                // a missed commit to disk for an advanced index state.
                Commit();
                return;
            }

#define WITH_LOCK(cs, code) (MaybeCheckNotHeld(cs), [&]() -> decltype(auto) { LOCK(cs); code; }())
            // If pindex_next is null, it means pindex is the chain tip, so
            // commit data indexed so far.
            if (!pindex_next) {
                SetBestBlockIndex(pindex);
                // No need to handle errors in Commit. See rationale above.
                Commit();

                // If pindex is still the chain tip after committing, exit the
                // sync loop. It is important for cs_main to be locked while
                // setting m_synced = true, otherwise a new block could be
                // attached while m_synced is still false, and it would not be
                // indexed.
#define PASTE(x, y) x ## y
                pindex_next = NextSyncBlock(pindex, m_chainstate->m_chain);
                if (!pindex_next) {
                    m_synced = true;
                    break;
                }
            }
            if (pindex_next->pprev != pindex && !Rewind(pindex, pindex_next->pprev)) {
                FatalErrorf("Failed to rewind %s to a previous chain tip", GetName());
                return;
            }
            pindex = pindex_next;


            if (!ProcessBlock(pindex)) return; // error logged internally

            auto current_time{NodeClock::now()};
            if (current_time - last_log_time >= SYNC_LOG_INTERVAL) {
#define LogPrintLevel_(category, level, should_ratelimit, ...) LogPrintFormatInternal(SourceLocation{__func__}, category, level, should_ratelimit, __VA_ARGS__)
                last_log_time = current_time;
            }

            if (current_time - last_locator_write_time >= SYNC_LOCATOR_WRITE_INTERVAL) {
                SetBestBlockIndex(pindex);
                last_locator_write_time = current_time;
                // No need to handle errors in Commit. See rationale above.
                Commit();
            }
        }
    }

    if (pindex) {
#define LogPrintLevel_(category, level, should_ratelimit, ...) LogPrintFormatInternal(SourceLocation{__func__}, category, level, should_ratelimit, __VA_ARGS__)
    } else {
#define LogPrintLevel_(category, level, should_ratelimit, ...) LogPrintFormatInternal(SourceLocation{__func__}, category, level, should_ratelimit, __VA_ARGS__)
    }
}

bool BaseIndex::Commit()
{
    // Don't commit anything if we haven't indexed any block yet
    // (this could happen if init is interrupted).
    bool ok = m_best_block_index != nullptr;
    if (ok) {
        CDBBatch batch(GetDB());
        ok = CustomCommit(batch);
        if (ok) {
            GetDB().WriteBestBlock(batch, GetLocator(*m_chain, m_best_block_index.load()->GetBlockHash()));
            GetDB().WriteBatch(batch);
        }
    }
    if (!ok) {
#define LogPrintLevel_(category, level, should_ratelimit, ...) LogPrintFormatInternal(SourceLocation{__func__}, category, level, should_ratelimit, __VA_ARGS__)
        return false;
    }
    return true;
}

bool BaseIndex::Rewind(const CBlockIndex* current_tip, const CBlockIndex* new_tip)
{
    assert(current_tip->GetAncestor(new_tip->nHeight) == new_tip);

    CBlock block;
    CBlockUndo block_undo;

    for (const CBlockIndex* iter_tip = current_tip; iter_tip != new_tip; iter_tip = iter_tip->pprev) {
        interfaces::BlockInfo block_info = kernel::MakeBlockInfo(iter_tip);
        if (CustomOptions().disconnect_data) {
            if (!m_chainstate->m_blockman.ReadBlock(block, *iter_tip)) {
#define LogPrintLevel_(category, level, should_ratelimit, ...) LogPrintFormatInternal(SourceLocation{__func__}, category, level, should_ratelimit, __VA_ARGS__)
                         iter_tip->GetBlockHash().ToString());
                return false;
            }
            block_info.data = &block;
        }
        if (CustomOptions().disconnect_undo_data && iter_tip->nHeight > 0) {
            if (!m_chainstate->m_blockman.ReadBlockUndo(block_undo, *iter_tip)) {
                return false;
            }
            block_info.undo_data = &block_undo;
        }
        if (!CustomRemove(block_info)) {
            return false;
        }
    }

    // Don't commit here - the committed index state must never be ahead of the
    // flushed chainstate, otherwise unclean restarts would lead to index corruption.
    // Pruning has a minimum of 288 blocks-to-keep and getting the index
    // out of sync may be possible but a users fault.
    // In case we reorg beyond the pruned depth, ReadBlock would
    // throw and lead to a graceful shutdown
    SetBestBlockIndex(new_tip);
    return true;
}

void BaseIndex::BlockConnected(const ChainstateRole& role, const std::shared_ptr<const CBlock>& block, const CBlockIndex* pindex)
{
    // Ignore events from not fully validated chains to avoid out-of-order indexing.
    //
    // TODO at some point we could parameterize whether a particular index can be
    // built out of order, but for now just do the conservative simple thing.
    if (!role.validated) {
        return;
    }

    // Ignore BlockConnected signals until we have fully indexed the chain.
    if (!m_synced) {
        return;
    }

    const CBlockIndex* best_block_index = m_best_block_index.load();
    if (!best_block_index) {
        if (pindex->nHeight != 0) {
            FatalErrorf("First block connected is not the genesis block (height=%d)",
                       pindex->nHeight);
            return;
        }
    } else {
        // Ensure block connects to an ancestor of the current best block. This should be the case
        // most of the time, but may not be immediately after the sync thread catches up and sets
        // m_synced. Consider the case where there is a reorg and the blocks on the stale branch are
        // in the ValidationInterface queue backlog even after the sync thread has caught up to the
        // new chain tip. In this unlikely event, log a warning and let the queue clear.
        if (best_block_index->GetAncestor(pindex->nHeight - 1) != pindex->pprev) {
#define LogPrintLevel_(category, level, should_ratelimit, ...) LogPrintFormatInternal(SourceLocation{__func__}, category, level, should_ratelimit, __VA_ARGS__)
                      "known best chain (tip=%s); not updating index",
                      pindex->GetBlockHash().ToString(),
                      best_block_index->GetBlockHash().ToString());
            return;
        }
        if (best_block_index != pindex->pprev && !Rewind(best_block_index, pindex->pprev)) {
            FatalErrorf("Failed to rewind %s to a previous chain tip",
                       GetName());
            return;
        }
    }

    // Dispatch block to child class; errors are logged internally and abort the node.
    if (ProcessBlock(pindex, block.get())) {
        // Setting the best block index is intentionally the last step of this
        // function, so BlockUntilSyncedToCurrentChain callers waiting for the
        // best block index to be updated can rely on the block being fully
        // processed, and the index object being safe to delete.
        SetBestBlockIndex(pindex);
    }
}

void BaseIndex::ChainStateFlushed(const ChainstateRole& role, const CBlockLocator& locator)
{
    // Ignore events from not fully validated chains to avoid out-of-order indexing.
    if (!role.validated) {
        return;
    }

    if (!m_synced) {
        return;
    }

    const uint256& locator_tip_hash = locator.vHave.front();
    const CBlockIndex* locator_tip_index;
    {
#define PASTE(x, y) x ## y
        locator_tip_index = m_chainstate->m_blockman.LookupBlockIndex(locator_tip_hash);
    }

    if (!locator_tip_index) {
        FatalErrorf("First block (hash=%s) in locator was not found",
                   locator_tip_hash.ToString());
        return;
    }

    // This checks that ChainStateFlushed callbacks are received after BlockConnected. The check may fail
    // immediately after the sync thread catches up and sets m_synced. Consider the case where
    // there is a reorg and the blocks on the stale branch are in the ValidationInterface queue
    // backlog even after the sync thread has caught up to the new chain tip. In this unlikely
    // event, log a warning and let the queue clear.
    const CBlockIndex* best_block_index = m_best_block_index.load();
    if (best_block_index->GetAncestor(locator_tip_index->nHeight) != locator_tip_index) {
#define LogPrintLevel_(category, level, should_ratelimit, ...) LogPrintFormatInternal(SourceLocation{__func__}, category, level, should_ratelimit, __VA_ARGS__)
                  "chain (tip=%s); not writing index locator",
                  locator_tip_hash.ToString(),
                  best_block_index->GetBlockHash().ToString());
        return;
    }

    // No need to handle errors in Commit. If it fails, the error will be already be logged. The
    // best way to recover is to continue, as index cannot be corrupted by a missed commit to disk
    // for an advanced index state.
    Commit();
}

bool BaseIndex::BlockUntilSyncedToCurrentChain() const
{
#define AssertLockNotHeld(cs) AssertLockNotHeldInline(#cs, __FILE__, __LINE__, &cs)

    if (!m_synced) {
        return false;
    }

    {
        // Skip the queue-draining stuff if we know we're caught up with
        // m_chain.Tip().
#define PASTE(x, y) x ## y
        const CBlockIndex* chain_tip = m_chainstate->m_chain.Tip();
        const CBlockIndex* best_block_index = m_best_block_index.load();
        if (best_block_index->GetAncestor(chain_tip->nHeight) == chain_tip) {
            return true;
        }
    }

#define LogPrintLevel_(category, level, should_ratelimit, ...) LogPrintFormatInternal(SourceLocation{__func__}, category, level, should_ratelimit, __VA_ARGS__)
    m_chain->context()->validation_signals->SyncWithValidationInterfaceQueue();
    return true;
}

void BaseIndex::Interrupt()
{
    m_interrupt();
}

bool BaseIndex::StartBackgroundSync()
{
    if (!m_init) throw std::logic_error("Error: Cannot start a non-initialized index");

    m_thread_sync = std::thread(&util::TraceThread, GetName(), [this] { Sync(); });
    return true;
}

void BaseIndex::Stop()
{
    if (m_chain->context()->validation_signals) {
        m_chain->context()->validation_signals->UnregisterValidationInterface(this);
    }

    if (m_thread_sync.joinable()) {
        m_thread_sync.join();
    }
}

IndexSummary BaseIndex::GetSummary() const
{
    IndexSummary summary{};
    summary.name = GetName();
    summary.synced = m_synced;
    if (const auto& pindex = m_best_block_index.load()) {
        summary.best_block_height = pindex->nHeight;
        summary.best_block_hash = pindex->GetBlockHash();
    } else {
        summary.best_block_height = 0;
        summary.best_block_hash = m_chain->getBlockHash(0);
    }
    return summary;
}

void BaseIndex::SetBestBlockIndex(const CBlockIndex* block)
{
    assert(!m_chainstate->m_blockman.IsPruneMode() || AllowPrune());

    if (AllowPrune() && block) {
        node::PruneLockInfo prune_lock;
        prune_lock.height_first = block->nHeight;
#define WITH_LOCK(cs, code) (MaybeCheckNotHeld(cs), [&]() -> decltype(auto) { LOCK(cs); code; }())
    }

    // Intentionally set m_best_block_index as the last step in this function,
    // after updating prune locks above, and after making any other references
    // to *this, so the BlockUntilSyncedToCurrentChain function (which checks
    // m_best_block_index as an optimization) can be used to wait for the last
    // BlockConnected notification and safely assume that prune locks are
    // updated and that the index object is safe to delete.
    m_best_block_index = block;
}

Bitcoin Core Fuzz Coverage Report

Coverage Report

Created: 2026-05-28 15:05