Merge #11824: Block ActivateBestChain to empty validationinterface queue

97d2b09c12 Add helper to wait for validation interface queue to catch up (Matt Corallo)
36137497f1 Block ActivateBestChain to empty validationinterface queue (Matt Corallo)
5a933cefcc Add an interface to get the queue depth out of CValidationInterface (Matt Corallo)
a99b76f269 Require no cs_main lock for ProcessNewBlock/ActivateBestChain (Matt Corallo)
a734896038 Avoid cs_main in net_processing ActivateBestChain calls (Matt Corallo)
66aa1d58a1 Refactor ProcessGetData in anticipation of avoiding cs_main for ABC (Matt Corallo)
818075adac Create new mutex for orphans, no cs_main in PLV::BlockConnected (Matt Corallo)

Pull request description:

  This should fix #11822.

  It ended up bigger than I hoped for, but its not too gnarly. Note that "
  Require no cs_main lock for ProcessNewBlock/ActivateBestChain" is mostly pure code-movement.

Tree-SHA512: 1127688545926f6099449dca6a4e6609eefc3abbd72f1c66e03d32bd8c7b31e82097d8307822cfd1dec0321703579cfdd82069cab6e17b1024e75eac694122cb

src/net_processing.cpp

@@ -51,12 +51,13 @@ struct COrphanTx {
     NodeId fromPeer;
     int64_t nTimeExpire;
 };
-std::map<uint256, COrphanTx> mapOrphanTransactions GUARDED_BY(cs_main);
-std::map<COutPoint, std::set<std::map<uint256, COrphanTx>::iterator, IteratorComparator>> mapOrphanTransactionsByPrev GUARDED_BY(cs_main);
-void EraseOrphansFor(NodeId peer) EXCLUSIVE_LOCKS_REQUIRED(cs_main);
+static CCriticalSection g_cs_orphans;
+std::map<uint256, COrphanTx> mapOrphanTransactions GUARDED_BY(g_cs_orphans);
+std::map<COutPoint, std::set<std::map<uint256, COrphanTx>::iterator, IteratorComparator>> mapOrphanTransactionsByPrev GUARDED_BY(g_cs_orphans);
+void EraseOrphansFor(NodeId peer);
 
-static size_t vExtraTxnForCompactIt = 0;
-static std::vector<std::pair<uint256, CTransactionRef>> vExtraTxnForCompact GUARDED_BY(cs_main);
+static size_t vExtraTxnForCompactIt GUARDED_BY(g_cs_orphans) = 0;
+static std::vector<std::pair<uint256, CTransactionRef>> vExtraTxnForCompact GUARDED_BY(g_cs_orphans);
 
 static const uint64_t RANDOMIZER_ID_ADDRESS_RELAY = 0x3cac0035b5866b90ULL; // SHA256("main address relay")[0:8]
 
@@ -127,7 +128,7 @@ namespace {
     int g_outbound_peers_with_protect_from_disconnect = 0;
 
     /** When our tip was last updated. */
-    int64_t g_last_tip_update = 0;
+    std::atomic<int64_t> g_last_tip_update(0);
 
     /** Relay map, protected by cs_main. */
     typedef std::map<uint256, CTransactionRef> MapRelay;
@@ -631,7 +632,7 @@ bool GetNodeStateStats(NodeId nodeid, CNodeStateStats &stats) {
 // mapOrphanTransactions
 //
 
-void AddToCompactExtraTransactions(const CTransactionRef& tx) EXCLUSIVE_LOCKS_REQUIRED(cs_main)
+void AddToCompactExtraTransactions(const CTransactionRef& tx) EXCLUSIVE_LOCKS_REQUIRED(g_cs_orphans)
 {
     size_t max_extra_txn = gArgs.GetArg("-blockreconstructionextratxn", DEFAULT_BLOCK_RECONSTRUCTION_EXTRA_TXN);
     if (max_extra_txn <= 0)
@@ -642,7 +643,7 @@ void AddToCompactExtraTransactions(const CTransactionRef& tx) EXCLUSIVE_LOCKS_RE
     vExtraTxnForCompactIt = (vExtraTxnForCompactIt + 1) % max_extra_txn;
 }
 
-bool AddOrphanTx(const CTransactionRef& tx, NodeId peer) EXCLUSIVE_LOCKS_REQUIRED(cs_main)
+bool AddOrphanTx(const CTransactionRef& tx, NodeId peer) EXCLUSIVE_LOCKS_REQUIRED(g_cs_orphans)
 {
     const uint256& hash = tx->GetHash();
     if (mapOrphanTransactions.count(hash))
@@ -675,7 +676,7 @@ bool AddOrphanTx(const CTransactionRef& tx, NodeId peer) EXCLUSIVE_LOCKS_REQUIRE
     return true;
 }
 
-int static EraseOrphanTx(uint256 hash) EXCLUSIVE_LOCKS_REQUIRED(cs_main)
+int static EraseOrphanTx(uint256 hash) EXCLUSIVE_LOCKS_REQUIRED(g_cs_orphans)
 {
     std::map<uint256, COrphanTx>::iterator it = mapOrphanTransactions.find(hash);
     if (it == mapOrphanTransactions.end())
@@ -695,6 +696,7 @@ int static EraseOrphanTx(uint256 hash) EXCLUSIVE_LOCKS_REQUIRED(cs_main)
 
 void EraseOrphansFor(NodeId peer)
 {
+    LOCK(g_cs_orphans);
     int nErased = 0;
     std::map<uint256, COrphanTx>::iterator iter = mapOrphanTransactions.begin();
     while (iter != mapOrphanTransactions.end())
@@ -709,8 +711,10 @@ void EraseOrphansFor(NodeId peer)
 }
 
 
-unsigned int LimitOrphanTxSize(unsigned int nMaxOrphans) EXCLUSIVE_LOCKS_REQUIRED(cs_main)
+unsigned int LimitOrphanTxSize(unsigned int nMaxOrphans)
 {
+    LOCK(g_cs_orphans);
+
     unsigned int nEvicted = 0;
     static int64_t nNextSweep;
     int64_t nNow = GetTime();
@@ -804,7 +808,7 @@ PeerLogicValidation::PeerLogicValidation(CConnman* connmanIn, CScheduler &schedu
 }
 
 void PeerLogicValidation::BlockConnected(const std::shared_ptr<const CBlock>& pblock, const CBlockIndex* pindex, const std::vector<CTransactionRef>& vtxConflicted) {
-    LOCK(cs_main);
+    LOCK(g_cs_orphans);
 
     std::vector<uint256> vOrphanErase;
 
@@ -971,9 +975,13 @@ bool static AlreadyHave(const CInv& inv) EXCLUSIVE_LOCKS_REQUIRED(cs_main)
                 recentRejects->reset();
             }
 
+            {
+                LOCK(g_cs_orphans);
+                if (mapOrphanTransactions.count(inv.hash)) return true;
+            }
+
             return recentRejects->contains(inv.hash) ||
                    mempool.exists(inv.hash) ||
-                   mapOrphanTransactions.count(inv.hash) ||
                    pcoinsTip->HaveCoinInCache(COutPoint(inv.hash, 0)) || // Best effort: only try output 0 and 1
                    pcoinsTip->HaveCoinInCache(COutPoint(inv.hash, 1));
         }
@@ -1030,180 +1038,198 @@ static void RelayAddress(const CAddress& addr, bool fReachable, CConnman* connma
     connman->ForEachNodeThen(std::move(sortfunc), std::move(pushfunc));
 }
 
-void static ProcessGetData(CNode* pfrom, const Consensus::Params& consensusParams, CConnman* connman, const std::atomic<bool>& interruptMsgProc)
+void static ProcessGetBlockData(CNode* pfrom, const Consensus::Params& consensusParams, const CInv& inv, CConnman* connman, const std::atomic<bool>& interruptMsgProc)
 {
-    std::deque<CInv>::iterator it = pfrom->vRecvGetData.begin();
-    std::vector<CInv> vNotFound;
-    const CNetMsgMaker msgMaker(pfrom->GetSendVersion());
-    LOCK(cs_main);
-
-    while (it != pfrom->vRecvGetData.end()) {
-        // Don't bother if send buffer is too full to respond anyway
-        if (pfrom->fPauseSend)
-            break;
+    bool send = false;
+    std::shared_ptr<const CBlock> a_recent_block;
+    std::shared_ptr<const CBlockHeaderAndShortTxIDs> a_recent_compact_block;
+    bool fWitnessesPresentInARecentCompactBlock;
+    {
+        LOCK(cs_most_recent_block);
+        a_recent_block = most_recent_block;
+        a_recent_compact_block = most_recent_compact_block;
+        fWitnessesPresentInARecentCompactBlock = fWitnessesPresentInMostRecentCompactBlock;
+    }
 
-        const CInv &inv = *it;
+    bool need_activate_chain = false;
+    {
+        LOCK(cs_main);
+        BlockMap::iterator mi = mapBlockIndex.find(inv.hash);
+        if (mi != mapBlockIndex.end())
         {
-            if (interruptMsgProc)
-                return;
+            if (mi->second->nChainTx && !mi->second->IsValid(BLOCK_VALID_SCRIPTS) &&
+                    mi->second->IsValid(BLOCK_VALID_TREE)) {
+                // If we have the block and all of its parents, but have not yet validated it,
+                // we might be in the middle of connecting it (ie in the unlock of cs_main
+                // before ActivateBestChain but after AcceptBlock).
+                // In this case, we need to run ActivateBestChain prior to checking the relay
+                // conditions below.
+                need_activate_chain = true;
+            }
+        }
+    } // release cs_main before calling ActivateBestChain
+    if (need_activate_chain) {
+        CValidationState dummy;
+        ActivateBestChain(dummy, Params(), a_recent_block);
+    }
 
-            it++;
+    LOCK(cs_main);
+    BlockMap::iterator mi = mapBlockIndex.find(inv.hash);
+    if (mi != mapBlockIndex.end()) {
+        send = BlockRequestAllowed(mi->second, consensusParams);
+        if (!send) {
+            LogPrint(BCLog::NET, "%s: ignoring request from peer=%i for old block that isn't in the main chain\n", __func__, pfrom->GetId());
+        }
+    }
+    const CNetMsgMaker msgMaker(pfrom->GetSendVersion());
+    // disconnect node in case we have reached the outbound limit for serving historical blocks
+    // never disconnect whitelisted nodes
+    if (send && connman->OutboundTargetReached(true) && ( ((pindexBestHeader != nullptr) && (pindexBestHeader->GetBlockTime() - mi->second->GetBlockTime() > HISTORICAL_BLOCK_AGE)) || inv.type == MSG_FILTERED_BLOCK) && !pfrom->fWhitelisted)
+    {
+        LogPrint(BCLog::NET, "historical block serving limit reached, disconnect peer=%d\n", pfrom->GetId());
+
+        //disconnect node
+        pfrom->fDisconnect = true;
+        send = false;
+    }
+    // Avoid leaking prune-height by never sending blocks below the NODE_NETWORK_LIMITED threshold
+    if (send && !pfrom->fWhitelisted && (
+            (((pfrom->GetLocalServices() & NODE_NETWORK_LIMITED) == NODE_NETWORK_LIMITED) && ((pfrom->GetLocalServices() & NODE_NETWORK) != NODE_NETWORK) && (chainActive.Tip()->nHeight - mi->second->nHeight > (int)NODE_NETWORK_LIMITED_MIN_BLOCKS + 2 /* add two blocks buffer extension for possible races */) )
+       )) {
+        LogPrint(BCLog::NET, "Ignore block request below NODE_NETWORK_LIMITED threshold from peer=%d\n", pfrom->GetId());
 
-            if (inv.type == MSG_BLOCK || inv.type == MSG_FILTERED_BLOCK || inv.type == MSG_CMPCT_BLOCK || inv.type == MSG_WITNESS_BLOCK)
+        //disconnect node and prevent it from stalling (would otherwise wait for the missing block)
+        pfrom->fDisconnect = true;
+        send = false;
+    }
+    // Pruned nodes may have deleted the block, so check whether
+    // it's available before trying to send.
+    if (send && (mi->second->nStatus & BLOCK_HAVE_DATA))
+    {
+        std::shared_ptr<const CBlock> pblock;
+        if (a_recent_block && a_recent_block->GetHash() == (*mi).second->GetBlockHash()) {
+            pblock = a_recent_block;
+        } else {
+            // Send block from disk
+            std::shared_ptr<CBlock> pblockRead = std::make_shared<CBlock>();
+            if (!ReadBlockFromDisk(*pblockRead, (*mi).second, consensusParams))
+                assert(!"cannot load block from disk");
+            pblock = pblockRead;
+        }
+        if (inv.type == MSG_BLOCK)
+            connman->PushMessage(pfrom, msgMaker.Make(SERIALIZE_TRANSACTION_NO_WITNESS, NetMsgType::BLOCK, *pblock));
+        else if (inv.type == MSG_WITNESS_BLOCK)
+            connman->PushMessage(pfrom, msgMaker.Make(NetMsgType::BLOCK, *pblock));
+        else if (inv.type == MSG_FILTERED_BLOCK)
+        {
+            bool sendMerkleBlock = false;
+            CMerkleBlock merkleBlock;
             {
-                bool send = false;
-                BlockMap::iterator mi = mapBlockIndex.find(inv.hash);
-                std::shared_ptr<const CBlock> a_recent_block;
-                std::shared_ptr<const CBlockHeaderAndShortTxIDs> a_recent_compact_block;
-                bool fWitnessesPresentInARecentCompactBlock;
-                {
-                    LOCK(cs_most_recent_block);
-                    a_recent_block = most_recent_block;
-                    a_recent_compact_block = most_recent_compact_block;
-                    fWitnessesPresentInARecentCompactBlock = fWitnessesPresentInMostRecentCompactBlock;
+                LOCK(pfrom->cs_filter);
+                if (pfrom->pfilter) {
+                    sendMerkleBlock = true;
+                    merkleBlock = CMerkleBlock(*pblock, *pfrom->pfilter);
                 }
-                if (mi != mapBlockIndex.end())
-                {
-                    if (mi->second->nChainTx && !mi->second->IsValid(BLOCK_VALID_SCRIPTS) &&
-                            mi->second->IsValid(BLOCK_VALID_TREE)) {
-                        // If we have the block and all of its parents, but have not yet validated it,
-                        // we might be in the middle of connecting it (ie in the unlock of cs_main
-                        // before ActivateBestChain but after AcceptBlock).
-                        // In this case, we need to run ActivateBestChain prior to checking the relay
-                        // conditions below.
-                        CValidationState dummy;
-                        ActivateBestChain(dummy, Params(), a_recent_block);
-                    }
-                    send = BlockRequestAllowed(mi->second, consensusParams);
-                    if (!send) {
-                        LogPrint(BCLog::NET, "%s: ignoring request from peer=%i for old block that isn't in the main chain\n", __func__, pfrom->GetId());
-                    }
+            }
+            if (sendMerkleBlock) {
+                connman->PushMessage(pfrom, msgMaker.Make(NetMsgType::MERKLEBLOCK, merkleBlock));
+                // CMerkleBlock just contains hashes, so also push any transactions in the block the client did not see
+                // This avoids hurting performance by pointlessly requiring a round-trip
+                // Note that there is currently no way for a node to request any single transactions we didn't send here -
+                // they must either disconnect and retry or request the full block.
+                // Thus, the protocol spec specified allows for us to provide duplicate txn here,
+                // however we MUST always provide at least what the remote peer needs
+                typedef std::pair<unsigned int, uint256> PairType;
+                for (PairType& pair : merkleBlock.vMatchedTxn)
+                    connman->PushMessage(pfrom, msgMaker.Make(SERIALIZE_TRANSACTION_NO_WITNESS, NetMsgType::TX, *pblock->vtx[pair.first]));
+            }
+            // else
+                // no response
+        }
+        else if (inv.type == MSG_CMPCT_BLOCK)
+        {
+            // If a peer is asking for old blocks, we're almost guaranteed
+            // they won't have a useful mempool to match against a compact block,
+            // and we don't feel like constructing the object for them, so
+            // instead we respond with the full, non-compact block.
+            bool fPeerWantsWitness = State(pfrom->GetId())->fWantsCmpctWitness;
+            int nSendFlags = fPeerWantsWitness ? 0 : SERIALIZE_TRANSACTION_NO_WITNESS;
+            if (CanDirectFetch(consensusParams) && mi->second->nHeight >= chainActive.Height() - MAX_CMPCTBLOCK_DEPTH) {
+                if ((fPeerWantsWitness || !fWitnessesPresentInARecentCompactBlock) && a_recent_compact_block && a_recent_compact_block->header.GetHash() == mi->second->GetBlockHash()) {
+                    connman->PushMessage(pfrom, msgMaker.Make(nSendFlags, NetMsgType::CMPCTBLOCK, *a_recent_compact_block));
+                } else {
+                    CBlockHeaderAndShortTxIDs cmpctblock(*pblock, fPeerWantsWitness);
+                    connman->PushMessage(pfrom, msgMaker.Make(nSendFlags, NetMsgType::CMPCTBLOCK, cmpctblock));
                 }
-                // disconnect node in case we have reached the outbound limit for serving historical blocks
-                // never disconnect whitelisted nodes
-                if (send && connman->OutboundTargetReached(true) && ( ((pindexBestHeader != nullptr) && (pindexBestHeader->GetBlockTime() - mi->second->GetBlockTime() > HISTORICAL_BLOCK_AGE)) || inv.type == MSG_FILTERED_BLOCK) && !pfrom->fWhitelisted)
-                {
-                    LogPrint(BCLog::NET, "historical block serving limit reached, disconnect peer=%d\n", pfrom->GetId());
+            } else {
+                connman->PushMessage(pfrom, msgMaker.Make(nSendFlags, NetMsgType::BLOCK, *pblock));
+            }
+        }
 
-                    //disconnect node
-                    pfrom->fDisconnect = true;
-                    send = false;
-                }
-                // Avoid leaking prune-height by never sending blocks below the NODE_NETWORK_LIMITED threshold
-                if (send && !pfrom->fWhitelisted && (
-                        (((pfrom->GetLocalServices() & NODE_NETWORK_LIMITED) == NODE_NETWORK_LIMITED) && ((pfrom->GetLocalServices() & NODE_NETWORK) != NODE_NETWORK) && (chainActive.Tip()->nHeight - mi->second->nHeight > (int)NODE_NETWORK_LIMITED_MIN_BLOCKS + 2 /* add two blocks buffer extension for possible races */) )
-                   )) {
-                    LogPrint(BCLog::NET, "Ignore block request below NODE_NETWORK_LIMITED threshold from peer=%d\n", pfrom->GetId());
+        // Trigger the peer node to send a getblocks request for the next batch of inventory
+        if (inv.hash == pfrom->hashContinue)
+        {
+            // Bypass PushInventory, this must send even if redundant,
+            // and we want it right after the last block so they don't
+            // wait for other stuff first.
+            std::vector<CInv> vInv;
+            vInv.push_back(CInv(MSG_BLOCK, chainActive.Tip()->GetBlockHash()));
+            connman->PushMessage(pfrom, msgMaker.Make(NetMsgType::INV, vInv));
+            pfrom->hashContinue.SetNull();
+        }
+    }
+}
 
-                    //disconnect node and prevent it from stalling (would otherwise wait for the missing block)
-                    pfrom->fDisconnect = true;
-                    send = false;
-                }
-                // Pruned nodes may have deleted the block, so check whether
-                // it's available before trying to send.
-                if (send && (mi->second->nStatus & BLOCK_HAVE_DATA))
-                {
-                    std::shared_ptr<const CBlock> pblock;
-                    if (a_recent_block && a_recent_block->GetHash() == (*mi).second->GetBlockHash()) {
-                        pblock = a_recent_block;
-                    } else {
-                        // Send block from disk
-                        std::shared_ptr<CBlock> pblockRead = std::make_shared<CBlock>();
-                        if (!ReadBlockFromDisk(*pblockRead, (*mi).second, consensusParams))
-                            assert(!"cannot load block from disk");
-                        pblock = pblockRead;
-                    }
-                    if (inv.type == MSG_BLOCK)
-                        connman->PushMessage(pfrom, msgMaker.Make(SERIALIZE_TRANSACTION_NO_WITNESS, NetMsgType::BLOCK, *pblock));
-                    else if (inv.type == MSG_WITNESS_BLOCK)
-                        connman->PushMessage(pfrom, msgMaker.Make(NetMsgType::BLOCK, *pblock));
-                    else if (inv.type == MSG_FILTERED_BLOCK)
-                    {
-                        bool sendMerkleBlock = false;
-                        CMerkleBlock merkleBlock;
-                        {
-                            LOCK(pfrom->cs_filter);
-                            if (pfrom->pfilter) {
-                                sendMerkleBlock = true;
-                                merkleBlock = CMerkleBlock(*pblock, *pfrom->pfilter);
-                            }
-                        }
-                        if (sendMerkleBlock) {
-                            connman->PushMessage(pfrom, msgMaker.Make(NetMsgType::MERKLEBLOCK, merkleBlock));
-                            // CMerkleBlock just contains hashes, so also push any transactions in the block the client did not see
-                            // This avoids hurting performance by pointlessly requiring a round-trip
-                            // Note that there is currently no way for a node to request any single transactions we didn't send here -
-                            // they must either disconnect and retry or request the full block.
-                            // Thus, the protocol spec specified allows for us to provide duplicate txn here,
-                            // however we MUST always provide at least what the remote peer needs
-                            typedef std::pair<unsigned int, uint256> PairType;
-                            for (PairType& pair : merkleBlock.vMatchedTxn)
-                                connman->PushMessage(pfrom, msgMaker.Make(SERIALIZE_TRANSACTION_NO_WITNESS, NetMsgType::TX, *pblock->vtx[pair.first]));
-                        }
-                        // else
-                            // no response
-                    }
-                    else if (inv.type == MSG_CMPCT_BLOCK)
-                    {
-                        // If a peer is asking for old blocks, we're almost guaranteed
-                        // they won't have a useful mempool to match against a compact block,
-                        // and we don't feel like constructing the object for them, so
-                        // instead we respond with the full, non-compact block.
-                        bool fPeerWantsWitness = State(pfrom->GetId())->fWantsCmpctWitness;
-                        int nSendFlags = fPeerWantsWitness ? 0 : SERIALIZE_TRANSACTION_NO_WITNESS;
-                        if (CanDirectFetch(consensusParams) && mi->second->nHeight >= chainActive.Height() - MAX_CMPCTBLOCK_DEPTH) {
-                            if ((fPeerWantsWitness || !fWitnessesPresentInARecentCompactBlock) && a_recent_compact_block && a_recent_compact_block->header.GetHash() == mi->second->GetBlockHash()) {
-                                connman->PushMessage(pfrom, msgMaker.Make(nSendFlags, NetMsgType::CMPCTBLOCK, *a_recent_compact_block));
-                            } else {
-                                CBlockHeaderAndShortTxIDs cmpctblock(*pblock, fPeerWantsWitness);
-                                connman->PushMessage(pfrom, msgMaker.Make(nSendFlags, NetMsgType::CMPCTBLOCK, cmpctblock));
-                            }
-                        } else {
-                            connman->PushMessage(pfrom, msgMaker.Make(nSendFlags, NetMsgType::BLOCK, *pblock));
-                        }
-                    }
+void static ProcessGetData(CNode* pfrom, const Consensus::Params& consensusParams, CConnman* connman, const std::atomic<bool>& interruptMsgProc)
+{
+    AssertLockNotHeld(cs_main);
 
-                    // Trigger the peer node to send a getblocks request for the next batch of inventory
-                    if (inv.hash == pfrom->hashContinue)
-                    {
-                        // Bypass PushInventory, this must send even if redundant,
-                        // and we want it right after the last block so they don't
-                        // wait for other stuff first.
-                        std::vector<CInv> vInv;
-                        vInv.push_back(CInv(MSG_BLOCK, chainActive.Tip()->GetBlockHash()));
-                        connman->PushMessage(pfrom, msgMaker.Make(NetMsgType::INV, vInv));
-                        pfrom->hashContinue.SetNull();
-                    }
-                }
-            }
-            else if (inv.type == MSG_TX || inv.type == MSG_WITNESS_TX)
-            {
-                // Send stream from relay memory
-                bool push = false;
-                auto mi = mapRelay.find(inv.hash);
-                int nSendFlags = (inv.type == MSG_TX ? SERIALIZE_TRANSACTION_NO_WITNESS : 0);
-                if (mi != mapRelay.end()) {
-                    connman->PushMessage(pfrom, msgMaker.Make(nSendFlags, NetMsgType::TX, *mi->second));
+    std::deque<CInv>::iterator it = pfrom->vRecvGetData.begin();
+    std::vector<CInv> vNotFound;
+    const CNetMsgMaker msgMaker(pfrom->GetSendVersion());
+    {
+        LOCK(cs_main);
+
+        while (it != pfrom->vRecvGetData.end() && (it->type == MSG_TX || it->type == MSG_WITNESS_TX)) {
+            if (interruptMsgProc)
+                return;
+            // Don't bother if send buffer is too full to respond anyway
+            if (pfrom->fPauseSend)
+                break;
+
+            const CInv &inv = *it;
+            it++;
+
+            // Send stream from relay memory
+            bool push = false;
+            auto mi = mapRelay.find(inv.hash);
+            int nSendFlags = (inv.type == MSG_TX ? SERIALIZE_TRANSACTION_NO_WITNESS : 0);
+            if (mi != mapRelay.end()) {
+                connman->PushMessage(pfrom, msgMaker.Make(nSendFlags, NetMsgType::TX, *mi->second));
+                push = true;
+            } else if (pfrom->timeLastMempoolReq) {
+                auto txinfo = mempool.info(inv.hash);
+                // To protect privacy, do not answer getdata using the mempool when
+                // that TX couldn't have been INVed in reply to a MEMPOOL request.
+                if (txinfo.tx && txinfo.nTime <= pfrom->timeLastMempoolReq) {
+                    connman->PushMessage(pfrom, msgMaker.Make(nSendFlags, NetMsgType::TX, *txinfo.tx));
                     push = true;
-                } else if (pfrom->timeLastMempoolReq) {
-                    auto txinfo = mempool.info(inv.hash);
-                    // To protect privacy, do not answer getdata using the mempool when
-                    // that TX couldn't have been INVed in reply to a MEMPOOL request.
-                    if (txinfo.tx && txinfo.nTime <= pfrom->timeLastMempoolReq) {
-                        connman->PushMessage(pfrom, msgMaker.Make(nSendFlags, NetMsgType::TX, *txinfo.tx));
-                        push = true;
-                    }
-                }
-                if (!push) {
-                    vNotFound.push_back(inv);
                 }
             }
+            if (!push) {
+                vNotFound.push_back(inv);
+            }
 
             // Track requests for our stuff.
             GetMainSignals().Inventory(inv.hash);
+        }
+    } // release cs_main
 
-            if (inv.type == MSG_BLOCK || inv.type == MSG_FILTERED_BLOCK || inv.type == MSG_CMPCT_BLOCK || inv.type == MSG_WITNESS_BLOCK)
-                break;
+    if (it != pfrom->vRecvGetData.end()) {
+        const CInv &inv = *it;
+        it++;
+        if (inv.type == MSG_BLOCK || inv.type == MSG_FILTERED_BLOCK || inv.type == MSG_CMPCT_BLOCK || inv.type == MSG_WITNESS_BLOCK) {
+            ProcessGetBlockData(pfrom, consensusParams, inv, connman, interruptMsgProc);
         }
     }
 
@@ -2008,7 +2034,7 @@ bool static ProcessMessage(CNode* pfrom, const std::string& strCommand, CDataStr
             inv.type = State(pfrom->GetId())->fWantsCmpctWitness ? MSG_WITNESS_BLOCK : MSG_BLOCK;
             inv.hash = req.blockhash;
             pfrom->vRecvGetData.push_back(inv);
-            ProcessGetData(pfrom, chainparams.GetConsensus(), connman, interruptMsgProc);
+            // The message processing loop will go around again (without pausing) and we'll respond then (without cs_main)
             return true;
         }
 
@@ -2101,7 +2127,7 @@ bool static ProcessMessage(CNode* pfrom, const std::string& strCommand, CDataStr
         CInv inv(MSG_TX, tx.GetHash());
         pfrom->AddInventoryKnown(inv);
 
-        LOCK(cs_main);
+        LOCK2(cs_main, g_cs_orphans);
 
         bool fMissingInputs = false;
         CValidationState state;
@@ -2324,7 +2350,7 @@ bool static ProcessMessage(CNode* pfrom, const std::string& strCommand, CDataStr
         bool fBlockReconstructed = false;
 
         {
-        LOCK(cs_main);
+        LOCK2(cs_main, g_cs_orphans);
         // If AcceptBlockHeader returned true, it set pindex
         assert(pindex);
         UpdateBlockAvailability(pfrom->GetId(), pindex->GetBlockHash());

src/scheduler.cpp

@@ -206,3 +206,8 @@ void SingleThreadedSchedulerClient::EmptyQueue() {
         should_continue = !m_callbacks_pending.empty();
     }
 }
+
+size_t SingleThreadedSchedulerClient::CallbacksPending() {
+    LOCK(m_cs_callbacks_pending);
+    return m_callbacks_pending.size();
+}

src/scheduler.h

@@ -108,6 +108,8 @@ public:
     // Processes all remaining queue members on the calling thread, blocking until queue is empty
     // Must be called after the CScheduler has no remaining processing threads!
     void EmptyQueue();
+
+    size_t CallbacksPending();
 };
 
 #endif

src/test/miner_tests.cpp

@@ -216,7 +216,6 @@ BOOST_AUTO_TEST_CASE(CreateNewBlock_validity)
     entry.nFee = 11;
     entry.nHeight = 11;
 
-    LOCK(cs_main);
     fCheckpointsEnabled = false;
 
     // Simple block creation, nothing special yet:
@@ -229,27 +228,32 @@ BOOST_AUTO_TEST_CASE(CreateNewBlock_validity)
     for (unsigned int i = 0; i < sizeof(blockinfo)/sizeof(*blockinfo); ++i)
     {
         CBlock *pblock = &pblocktemplate->block; // pointer for convenience
-        pblock->nVersion = 1;
-        pblock->nTime = chainActive.Tip()->GetMedianTimePast()+1;
-        CMutableTransaction txCoinbase(*pblock->vtx[0]);
-        txCoinbase.nVersion = 1;
-        txCoinbase.vin[0].scriptSig = CScript();
-        txCoinbase.vin[0].scriptSig.push_back(blockinfo[i].extranonce);
-        txCoinbase.vin[0].scriptSig.push_back(chainActive.Height());
-        txCoinbase.vout.resize(1); // Ignore the (optional) segwit commitment added by CreateNewBlock (as the hardcoded nonces don't account for this)
-        txCoinbase.vout[0].scriptPubKey = CScript();
-        pblock->vtx[0] = MakeTransactionRef(std::move(txCoinbase));
-        if (txFirst.size() == 0)
-            baseheight = chainActive.Height();
-        if (txFirst.size() < 4)
-            txFirst.push_back(pblock->vtx[0]);
-        pblock->hashMerkleRoot = BlockMerkleRoot(*pblock);
-        pblock->nNonce = blockinfo[i].nonce;
+        {
+            LOCK(cs_main);
+            pblock->nVersion = 1;
+            pblock->nTime = chainActive.Tip()->GetMedianTimePast()+1;
+            CMutableTransaction txCoinbase(*pblock->vtx[0]);
+            txCoinbase.nVersion = 1;
+            txCoinbase.vin[0].scriptSig = CScript();
+            txCoinbase.vin[0].scriptSig.push_back(blockinfo[i].extranonce);
+            txCoinbase.vin[0].scriptSig.push_back(chainActive.Height());
+            txCoinbase.vout.resize(1); // Ignore the (optional) segwit commitment added by CreateNewBlock (as the hardcoded nonces don't account for this)
+            txCoinbase.vout[0].scriptPubKey = CScript();
+            pblock->vtx[0] = MakeTransactionRef(std::move(txCoinbase));
+            if (txFirst.size() == 0)
+                baseheight = chainActive.Height();
+            if (txFirst.size() < 4)
+                txFirst.push_back(pblock->vtx[0]);
+            pblock->hashMerkleRoot = BlockMerkleRoot(*pblock);
+            pblock->nNonce = blockinfo[i].nonce;
+        }
         std::shared_ptr<const CBlock> shared_pblock = std::make_shared<const CBlock>(*pblock);
         BOOST_CHECK(ProcessNewBlock(chainparams, shared_pblock, true, nullptr));
         pblock->hashPrevBlock = pblock->GetHash();
     }
 
+    LOCK(cs_main);
+
     // Just to make sure we can still make simple blocks
     BOOST_CHECK(pblocktemplate = AssemblerForTest(chainparams).CreateNewBlock(scriptPubKey));
 

src/test/test_bitcoin.cpp

@@ -69,9 +69,9 @@ TestingSetup::TestingSetup(const std::string& chainName) : BasicTestingSetup(cha
         fs::create_directories(pathTemp);
         gArgs.ForceSetArg("-datadir", pathTemp.string());
 
-        // Note that because we don't bother running a scheduler thread here,
-        // callbacks via CValidationInterface are unreliable, but that's OK,
-        // our unit tests aren't testing multiple parts of the code at once.
+        // We have to run a scheduler thread to prevent ActivateBestChain
+        // from blocking due to queue overrun.
+        threadGroup.create_thread(boost::bind(&CScheduler::serviceQueue, &scheduler));
         GetMainSignals().RegisterBackgroundSignalScheduler(scheduler);
 
         mempool.setSanityCheck(1.0);

src/test/txvalidationcache_tests.cpp

@@ -66,7 +66,6 @@ BOOST_FIXTURE_TEST_CASE(tx_mempool_block_doublespend, TestChain100Setup)
 
     // Test 1: block with both of those transactions should be rejected.
     block = CreateAndProcessBlock(spends, scriptPubKey);
-    LOCK(cs_main);
     BOOST_CHECK(chainActive.Tip()->GetBlockHash() != block.GetHash());
 
     // Test 2: ... and should be rejected if spend1 is in the memory pool
@@ -190,12 +189,12 @@ BOOST_FIXTURE_TEST_CASE(checkinputs_test, TestChain100Setup)
         spend_tx.vin[0].scriptSig << vchSig;
     }
 
-    LOCK(cs_main);
-
     // Test that invalidity under a set of flags doesn't preclude validity
     // under other (eg consensus) flags.
     // spend_tx is invalid according to DERSIG
     {
+        LOCK(cs_main);
+
         CValidationState state;
         PrecomputedTransactionData ptd_spend_tx(spend_tx);
 
@@ -213,15 +212,17 @@ BOOST_FIXTURE_TEST_CASE(checkinputs_test, TestChain100Setup)
         // test later that block validation works fine in the absence of cached
         // successes.
         ValidateCheckInputsForAllFlags(spend_tx, SCRIPT_VERIFY_DERSIG | SCRIPT_VERIFY_LOW_S | SCRIPT_VERIFY_STRICTENC, false);
+    }
 
-        // And if we produce a block with this tx, it should be valid (DERSIG not
-        // enabled yet), even though there's no cache entry.
-        CBlock block;
+    // And if we produce a block with this tx, it should be valid (DERSIG not
+    // enabled yet), even though there's no cache entry.
+    CBlock block;
 
-        block = CreateAndProcessBlock({spend_tx}, p2pk_scriptPubKey);
-        BOOST_CHECK(chainActive.Tip()->GetBlockHash() == block.GetHash());
-        BOOST_CHECK(pcoinsTip->GetBestBlock() == block.GetHash());
-    }
+    block = CreateAndProcessBlock({spend_tx}, p2pk_scriptPubKey);
+    BOOST_CHECK(chainActive.Tip()->GetBlockHash() == block.GetHash());
+    BOOST_CHECK(pcoinsTip->GetBestBlock() == block.GetHash());
+
+    LOCK(cs_main);
 
     // Test P2SH: construct a transaction that is valid without P2SH, and
     // then test validity with P2SH.

src/validation.cpp

@@ -40,6 +40,7 @@
 #include <validationinterface.h>
 #include <warnings.h>
 
+#include <future>
 #include <sstream>
 
 #include <boost/algorithm/string/replace.hpp>
@@ -2559,12 +2560,21 @@ bool CChainState::ActivateBestChain(CValidationState &state, const CChainParams&
     // far from a guarantee. Things in the P2P/RPC will often end up calling
     // us in the middle of ProcessNewBlock - do not assume pblock is set
     // sanely for performance or correctness!
+    AssertLockNotHeld(cs_main);
 
     CBlockIndex *pindexMostWork = nullptr;
     CBlockIndex *pindexNewTip = nullptr;
     int nStopAtHeight = gArgs.GetArg("-stopatheight", DEFAULT_STOPATHEIGHT);
     do {
         boost::this_thread::interruption_point();
+
+        if (GetMainSignals().CallbacksPending() > 10) {
+            // Block until the validation queue drains. This should largely
+            // never happen in normal operation, however may happen during
+            // reindex, causing memory blowup  if we run too far ahead.
+            SyncWithValidationInterfaceQueue();
+        }
+
         if (ShutdownRequested())
             break;
 
@@ -3383,6 +3393,8 @@ bool CChainState::AcceptBlock(const std::shared_ptr<const CBlock>& pblock, CVali
 
 bool ProcessNewBlock(const CChainParams& chainparams, const std::shared_ptr<const CBlock> pblock, bool fForceProcessing, bool *fNewBlock)
 {
+    AssertLockNotHeld(cs_main);
+
     {
         CBlockIndex *pindex = nullptr;
         if (fNewBlock) *fNewBlock = false;

src/validationinterface.cpp

@@ -11,9 +11,11 @@
 #include <sync.h>
 #include <txmempool.h>
 #include <util.h>
+#include <validation.h>
 
 #include <list>
 #include <atomic>
+#include <future>
 
 #include <boost/signals2/signal.hpp>
 
@@ -54,6 +56,11 @@ void CMainSignals::FlushBackgroundCallbacks() {
     }
 }
 
+size_t CMainSignals::CallbacksPending() {
+    if (!m_internals) return 0;
+    return m_internals->m_schedulerClient.CallbacksPending();
+}
+
 void CMainSignals::RegisterWithMempoolSignals(CTxMemPool& pool) {
     pool.NotifyEntryRemoved.connect(boost::bind(&CMainSignals::MempoolEntryRemoved, this, _1, _2));
 }
@@ -113,6 +120,16 @@ void CallFunctionInValidationInterfaceQueue(std::function<void ()> func) {
     g_signals.m_internals->m_schedulerClient.AddToProcessQueue(std::move(func));
 }
 
+void SyncWithValidationInterfaceQueue() {
+    AssertLockNotHeld(cs_main);
+    // Block until the validation queue drains
+    std::promise<void> promise;
+    CallFunctionInValidationInterfaceQueue([&promise] {
+        promise.set_value();
+    });
+    promise.get_future().wait();
+}
+
 void CMainSignals::MempoolEntryRemoved(CTransactionRef ptx, MemPoolRemovalReason reason) {
     if (reason != MemPoolRemovalReason::BLOCK && reason != MemPoolRemovalReason::CONFLICT) {
         m_internals->m_schedulerClient.AddToProcessQueue([ptx, this] {

src/validationinterface.h

@@ -42,6 +42,16 @@ void UnregisterAllValidationInterfaces();
  * will result in a deadlock (that DEBUG_LOCKORDER will miss).
  */
 void CallFunctionInValidationInterfaceQueue(std::function<void ()> func);
+/**
+ * This is a synonym for the following, which asserts certain locks are not
+ * held:
+ *     std::promise<void> promise;
+ *     CallFunctionInValidationInterfaceQueue([&promise] {
+ *         promise.set_value();
+ *     });
+ *     promise.get_future().wait();
+ */
+void SyncWithValidationInterfaceQueue();
 
 class CValidationInterface {
 protected:
@@ -131,6 +141,8 @@ public:
     /** Call any remaining callbacks on the calling thread */
     void FlushBackgroundCallbacks();
 
+    size_t CallbacksPending();
+
     /** Register with mempool to call TransactionRemovedFromMempool callbacks */
     void RegisterWithMempoolSignals(CTxMemPool& pool);
     /** Unregister with mempool */

src/wallet/test/wallet_tests.cpp

@@ -370,8 +370,6 @@ static void AddKey(CWallet& wallet, const CKey& key)
 
 BOOST_FIXTURE_TEST_CASE(rescan, TestChain100Setup)
 {
-    LOCK(cs_main);
-
     // Cap last block file size, and mine new block in a new block file.
     CBlockIndex* const nullBlock = nullptr;
     CBlockIndex* oldTip = chainActive.Tip();
@@ -379,6 +377,8 @@ BOOST_FIXTURE_TEST_CASE(rescan, TestChain100Setup)
     CreateAndProcessBlock({}, GetScriptForRawPubKey(coinbaseKey.GetPubKey()));
     CBlockIndex* newTip = chainActive.Tip();
 
+    LOCK(cs_main);
+
     // Verify ScanForWalletTransactions picks up transactions in both the old
     // and new block files.
     {
@@ -447,8 +447,6 @@ BOOST_FIXTURE_TEST_CASE(rescan, TestChain100Setup)
 // than or equal to key birthday.
 BOOST_FIXTURE_TEST_CASE(importwallet_rescan, TestChain100Setup)
 {
-    LOCK(cs_main);
-
     // Create two blocks with same timestamp to verify that importwallet rescan
     // will pick up both blocks, not just the first.
     const int64_t BLOCK_TIME = chainActive.Tip()->GetBlockTimeMax() + 5;
@@ -462,6 +460,8 @@ BOOST_FIXTURE_TEST_CASE(importwallet_rescan, TestChain100Setup)
     SetMockTime(KEY_TIME);
     coinbaseTxns.emplace_back(*CreateAndProcessBlock({}, GetScriptForRawPubKey(coinbaseKey.GetPubKey())).vtx[0]);
 
+    LOCK(cs_main);
+
     // Import key into wallet and call dumpwallet to create backup file.
     {
         CWallet wallet;
@@ -627,10 +627,15 @@ public:
         BOOST_CHECK(wallet->CreateTransaction({recipient}, wtx, reservekey, fee, changePos, error, dummy));
         CValidationState state;
         BOOST_CHECK(wallet->CommitTransaction(wtx, reservekey, nullptr, state));
+        CMutableTransaction blocktx;
+        {
+            LOCK(wallet->cs_wallet);
+            blocktx = CMutableTransaction(*wallet->mapWallet.at(wtx.GetHash()).tx);
+        }
+        CreateAndProcessBlock({CMutableTransaction(blocktx)}, GetScriptForRawPubKey(coinbaseKey.GetPubKey()));
         LOCK(wallet->cs_wallet);
         auto it = wallet->mapWallet.find(wtx.GetHash());
         BOOST_CHECK(it != wallet->mapWallet.end());
-        CreateAndProcessBlock({CMutableTransaction(*it->second.tx)}, GetScriptForRawPubKey(coinbaseKey.GetPubKey()));
         it->second.SetMerkleBranch(chainActive.Tip(), 1);
         return it->second;
     }
@@ -641,7 +646,6 @@ public:
 BOOST_FIXTURE_TEST_CASE(ListCoins, ListCoinsTestingSetup)
 {
     std::string coinbaseAddress = coinbaseKey.GetPubKey().GetID().ToString();
-    LOCK2(cs_main, wallet->cs_wallet);
 
     // Confirm ListCoins initially returns 1 coin grouped under coinbaseKey
     // address.
@@ -669,6 +673,7 @@ BOOST_FIXTURE_TEST_CASE(ListCoins, ListCoinsTestingSetup)
     BOOST_CHECK_EQUAL(available.size(), 2);
     for (const auto& group : list) {
         for (const auto& coin : group.second) {
+            LOCK(wallet->cs_wallet);
             wallet->LockCoin(COutPoint(coin.tx->GetHash(), coin.i));
         }
     }

src/wallet/wallet.cpp

@@ -1292,12 +1292,7 @@ void CWallet::BlockUntilSyncedToCurrentChain() {
     // ...otherwise put a callback in the validation interface queue and wait
     // for the queue to drain enough to execute it (indicating we are caught up
     // at least with the time we entered this function).
-
-    std::promise<void> promise;
-    CallFunctionInValidationInterfaceQueue([&promise] {
-        promise.set_value();
-    });
-    promise.get_future().wait();
+    SyncWithValidationInterfaceQueue();
 }