Parallelize script verification

* During block verification (when parallelism is requested), script
  check actions are stored instead of being executed immediately.
* After every processed transactions, its signature actions are
  pushed to a CScriptCheckQueue, which maintains a queue and some
  synchronization mechanism.
* Two or more threads (if enabled) start processing elements from
  this queue,
* When the block connection code is finished processing transactions,
  it joins the worker pool until the queue is empty.

As cs_main is held the entire time, and all verification must be
finished before the block continues processing, this does not reach
the best possible performance. It is a less drastic change than
some more advanced mechanisms (like doing verification out-of-band
entirely, and rolling back blocks when a failure is detected).

The -par=N flag controls the number of threads (1-16). 0 means auto,
and is the default.

bitcoin-qt.pro

@@ -156,6 +156,7 @@ HEADERS += src/qt/bitcoingui.h \
     src/init.h \
     src/irc.h \
     src/mruset.h \
+    src/checkqueue.h \
     src/json/json_spirit_writer_template.h \
     src/json/json_spirit_writer.h \
     src/json/json_spirit_value.h \

src/checkqueue.h

@@ -0,0 +1,206 @@
+// Copyright (c) 2012 The Bitcoin developers
+// Distributed under the MIT/X11 software license, see the accompanying
+// file COPYING or http://www.opensource.org/licenses/mit-license.php.
+#ifndef CHECKQUEUE_H
+#define CHECKQUEUE_H
+
+#include <boost/thread/mutex.hpp>
+#include <boost/thread/locks.hpp>
+#include <boost/thread/condition_variable.hpp>
+
+#include <vector>
+#include <algorithm>
+
+template<typename T> class CCheckQueueControl;
+
+/** Queue for verifications that have to be performed.
+  * The verifications are represented by a type T, which must provide an
+  * operator(), returning a bool.
+  *
+  * One thread (the master) is assumed to push batches of verifications
+  * onto the queue, where they are processed by N-1 worker threads. When
+  * the master is done adding work, it temporarily joins the worker pool
+  * as an N'th worker, until all jobs are done.
+  */
+template<typename T> class CCheckQueue {
+private:
+    // Mutex to protect the inner state
+    boost::mutex mutex;
+
+    // Worker threads block on this when out of work
+    boost::condition_variable condWorker;
+
+    // Master thread blocks on this when out of work
+    boost::condition_variable condMaster;
+
+    // Quit method blocks on this until all workers are gone
+    boost::condition_variable condQuit;
+
+    // The queue of elements to be processed.
+    // As the order of booleans doesn't matter, it is used as a LIFO (stack)
+    std::vector<T> queue;
+
+    // The number of workers (including the master) that are idle.
+    int nIdle;
+
+    // The total number of workers (including the master).
+    int nTotal;
+
+    // The temporary evaluation result.
+    bool fAllOk;
+
+    // Number of verifications that haven't completed yet.
+    // This includes elements that are not anymore in queue, but still in
+    // worker's own batches.
+    unsigned int nTodo;
+
+    // Whether we're shutting down.
+    bool fQuit;
+
+    // The maximum number of elements to be processed in one batch
+    unsigned int nBatchSize;
+
+    // Internal function that does bulk of the verification work.
+    bool Loop(bool fMaster = false) {
+        boost::condition_variable &cond = fMaster ? condMaster : condWorker;
+        std::vector<T> vChecks;
+        vChecks.reserve(nBatchSize);
+        unsigned int nNow = 0;
+        bool fOk = true;
+        do {
+            {
+                boost::unique_lock<boost::mutex> lock(mutex);
+                // first do the clean-up of the previous loop run (allowing us to do it in the same critsect)
+                if (nNow) {
+                    fAllOk &= fOk;
+                    nTodo -= nNow;
+                    if (nTodo == 0 && !fMaster)
+                        // We processed the last element; inform the master he can exit and return the result
+                        condMaster.notify_one();
+                } else {
+                    // first iteration
+                    nTotal++;
+                }
+                // logically, the do loop starts here
+                while (queue.empty()) {
+                    if ((fMaster || fQuit) && nTodo == 0) {
+                        nTotal--;
+                        if (nTotal==0)
+                            condQuit.notify_one();
+                        bool fRet = fAllOk;
+                        // reset the status for new work later
+                        if (fMaster)
+                            fAllOk = true;
+                        // return the current status
+                        return fRet;
+                    }
+                    nIdle++;
+                    cond.wait(lock); // wait
+                    nIdle--;
+                }
+                // Decide how many work units to process now.
+                // * Do not try to do everything at once, but aim for increasingly smaller batches so
+                //   all workers finish approximately simultaneously.
+                // * Try to account for idle jobs which will instantly start helping.
+                // * Don't do batches smaller than 1 (duh), or larger than nBatchSize.
+                nNow = std::max(1U, std::min(nBatchSize, (unsigned int)queue.size() / (nTotal + nIdle + 1)));
+                vChecks.resize(nNow);
+                for (unsigned int i = 0; i < nNow; i++) {
+                     // We want the lock on the mutex to be as short as possible, so swap jobs from the global
+                     // queue to the local batch vector instead of copying.
+                     vChecks[i].swap(queue.back());
+                     queue.pop_back();
+                }
+                // Check whether we need to do work at all
+                fOk = fAllOk;
+            }
+            // execute work
+            BOOST_FOREACH(T &check, vChecks)
+                if (fOk)
+                    fOk = check();
+            vChecks.clear();
+        } while(true);
+    }
+
+public:
+    // Create a new check queue
+    CCheckQueue(unsigned int nBatchSizeIn) :
+        nIdle(0), nTotal(0), fAllOk(true), nTodo(0), fQuit(false), nBatchSize(nBatchSizeIn) {}
+
+    // Worker thread
+    void Thread() {
+        Loop();
+    }
+
+    // Wait until execution finishes, and return whether all evaluations where succesful.
+    bool Wait() {
+        return Loop(true);
+    }
+
+    // Add a batch of checks to the queue
+    void Add(std::vector<T> &vChecks) {
+        boost::unique_lock<boost::mutex> lock(mutex);
+        BOOST_FOREACH(T &check, vChecks) {
+            queue.push_back(T());
+            check.swap(queue.back());
+        }
+        nTodo += vChecks.size();
+        if (vChecks.size() == 1)
+            condWorker.notify_one();
+        else if (vChecks.size() > 1)
+            condWorker.notify_all();
+    }
+
+    // Shut the queue down
+    void Quit() {
+        boost::unique_lock<boost::mutex> lock(mutex);
+        fQuit = true;
+        // No need to wake the master, as he will quit automatically when all jobs are
+        // done.
+        condWorker.notify_all(); 
+
+        while (nTotal > 0)
+            condQuit.wait(lock);
+    }
+
+    friend class CCheckQueueControl<T>;
+};
+
+/** RAII-style controller object for a CCheckQueue that guarantees the passed
+ *  queue is finished before continuing.
+ */
+template<typename T> class CCheckQueueControl {
+private:
+    CCheckQueue<T> *pqueue;
+    bool fDone;
+
+public:
+    CCheckQueueControl(CCheckQueue<T> *pqueueIn) : pqueue(pqueueIn), fDone(false) {
+        // passed queue is supposed to be unused, or NULL
+        if (pqueue != NULL) {
+            assert(pqueue->nTotal == pqueue->nIdle);
+            assert(pqueue->nTodo == 0);
+            assert(pqueue->fAllOk == true);
+        }
+    }
+
+    bool Wait() {
+        if (pqueue == NULL)
+            return true;
+        bool fRet = pqueue->Wait();
+        fDone = true;
+        return fRet;
+    }
+
+    void Add(std::vector<T> &vChecks) {
+        if (pqueue != NULL)
+            pqueue->Add(vChecks);
+    }
+
+    ~CCheckQueueControl() {
+        if (!fDone)
+            Wait();
+    }
+};
+
+#endif

src/init.cpp

@@ -84,6 +84,10 @@ void Shutdown(void* parg)
         fShutdown = true;
         nTransactionsUpdated++;
         bitdb.Flush(false);
+        {
+            LOCK(cs_main);
+            ThreadScriptCheckQuit();
+        }
         StopNode();
         {
             LOCK(cs_main);
@@ -303,6 +307,7 @@ std::string HelpMessage()
         "  -checklevel=<n>        " + _("How thorough the block verification is (0-6, default: 1)") + "\n" +
         "  -loadblock=<file>      " + _("Imports blocks from external blk000??.dat file") + "\n" +
         "  -reindex               " + _("Rebuild blockchain index from current blk000??.dat files") + "\n" +
+        "  -par=N                 " + _("Set the number of script verification threads (1-16, 0=auto, default: 0)") + "\n" +
 
         "\n" + _("Block creation options:") + "\n" +
         "  -blockminsize=<n>      "   + _("Set minimum block size in bytes (default: 0)") + "\n" +
@@ -484,6 +489,15 @@ bool AppInit2()
     fDebug = GetBoolArg("-debug");
     fBenchmark = GetBoolArg("-benchmark");
 
+    // -par=0 means autodetect, but nScriptCheckThreads==0 means no concurrency
+    nScriptCheckThreads = GetArg("-par", 0);
+    if (nScriptCheckThreads == 0)
+        nScriptCheckThreads = boost::thread::hardware_concurrency();
+    if (nScriptCheckThreads <= 1) 
+        nScriptCheckThreads = 0;
+    else if (nScriptCheckThreads > MAX_SCRIPTCHECK_THREADS)
+        nScriptCheckThreads = MAX_SCRIPTCHECK_THREADS;
+
     // -debug implies fDebug*
     if (fDebug)
         fDebugNet = true;
@@ -579,6 +593,12 @@ bool AppInit2()
     if (fDaemon)
         fprintf(stdout, "Bitcoin server starting\n");
 
+    if (nScriptCheckThreads) {
+        printf("Using %u threads for script verification\n", nScriptCheckThreads);
+        for (int i=0; i<nScriptCheckThreads-1; i++)
+            NewThread(ThreadScriptCheck, NULL);
+    }
+
     int64 nStart;
 
     // ********************************************************* Step 5: verify database integrity

src/main.cpp

@@ -10,6 +10,7 @@
 #include "net.h"
 #include "init.h"
 #include "ui_interface.h"
+#include "checkqueue.h"
 #include <boost/algorithm/string/replace.hpp>
 #include <boost/filesystem.hpp>
 #include <boost/filesystem/fstream.hpp>
@@ -40,6 +41,7 @@ uint256 hashBestChain = 0;
 CBlockIndex* pindexBest = NULL;
 set<CBlockIndex*, CBlockIndexWorkComparator> setBlockIndexValid; // may contain all CBlockIndex*'s that have validness >=BLOCK_VALID_TRANSACTIONS, and must contain those who aren't failed
 int64 nTimeBestReceived = 0;
+int nScriptCheckThreads = 0;
 bool fImporting = false;
 bool fReindex = false;
 bool fBenchmark = false;
@@ -1360,10 +1362,13 @@ bool VerifySignature(const CCoins& txFrom, const CTransaction& txTo, unsigned in
     return CScriptCheck(txFrom, txTo, nIn, flags, nHashType)();
 }
 
-bool CTransaction::CheckInputs(CCoinsViewCache &inputs, bool fScriptChecks, unsigned int flags) const
+bool CTransaction::CheckInputs(CCoinsViewCache &inputs, bool fScriptChecks, unsigned int flags, std::vector<CScriptCheck> *pvChecks) const
 {
     if (!IsCoinBase())
     {
+        if (pvChecks)
+            pvChecks->reserve(vin.size());
+
         // This doesn't trigger the DoS code on purpose; if it did, it would make it easier
         // for an attacker to attempt to split the network.
         if (!HaveInputs(inputs))
@@ -1416,8 +1421,12 @@ bool CTransaction::CheckInputs(CCoinsViewCache &inputs, bool fScriptChecks, unsi
                 const CCoins &coins = inputs.GetCoins(prevout.hash);
 
                 // Verify signature
-                if (!VerifySignature(coins, *this, i, flags, 0))
-                    return DoS(100,error("CheckInputs() : %s VerifySignature failed", GetHash().ToString().substr(0,10).c_str()));
+                CScriptCheck check(coins, *this, i, flags, 0);
+                if (pvChecks) {
+                    pvChecks->push_back(CScriptCheck());
+                    check.swap(pvChecks->back());
+                } else if (!check())
+                    return DoS(100,false);
             }
         }
     }
@@ -1567,6 +1576,19 @@ void static FlushBlockFile()
 
 bool FindUndoPos(int nFile, CDiskBlockPos &pos, unsigned int nAddSize);
 
+static CCheckQueue<CScriptCheck> scriptcheckqueue(128);
+
+void ThreadScriptCheck(void*) {
+    vnThreadsRunning[THREAD_SCRIPTCHECK]++;
+    RenameThread("bitcoin-scriptch");
+    scriptcheckqueue.Thread();
+    vnThreadsRunning[THREAD_SCRIPTCHECK]--;
+}
+
+void ThreadScriptCheckQuit() {
+    scriptcheckqueue.Quit();
+}
+
 bool CBlock::ConnectBlock(CBlockIndex* pindex, CCoinsViewCache &view, bool fJustCheck)
 {
     // Check it again in case a previous version let a bad block in
@@ -1607,6 +1629,8 @@ bool CBlock::ConnectBlock(CBlockIndex* pindex, CCoinsViewCache &view, bool fJust
 
     CBlockUndo blockundo;
 
+    CCheckQueueControl<CScriptCheck> control(fScriptChecks && nScriptCheckThreads ? &scriptcheckqueue : NULL);
+
     int64 nStart = GetTimeMicros();
     int64 nFees = 0;
     int nInputs = 0;
@@ -1638,8 +1662,10 @@ bool CBlock::ConnectBlock(CBlockIndex* pindex, CCoinsViewCache &view, bool fJust
 
             nFees += tx.GetValueIn(view)-tx.GetValueOut();
 
-            if (!tx.CheckInputs(view, fScriptChecks, fStrictPayToScriptHash ? SCRIPT_VERIFY_P2SH : SCRIPT_VERIFY_NONE))
+            std::vector<CScriptCheck> vChecks;
+            if (!tx.CheckInputs(view, fScriptChecks, fStrictPayToScriptHash ? SCRIPT_VERIFY_P2SH : SCRIPT_VERIFY_NONE, nScriptCheckThreads ? &vChecks : NULL))
                 return false;
+            control.Add(vChecks);
         }
 
         CTxUndo txundo;
@@ -1656,6 +1682,12 @@ bool CBlock::ConnectBlock(CBlockIndex* pindex, CCoinsViewCache &view, bool fJust
     if (vtx[0].GetValueOut() > GetBlockValue(pindex->nHeight, nFees))
         return error("ConnectBlock() : coinbase pays too much (actual=%"PRI64d" vs limit=%"PRI64d")", vtx[0].GetValueOut(), GetBlockValue(pindex->nHeight, nFees));
 
+    if (!control.Wait())
+        return DoS(100, false);
+    int64 nTime2 = GetTimeMicros() - nStart;
+    if (fBenchmark)
+        printf("- Verify %u txins: %.2fms (%.3fms/txin)\n", nInputs - 1, 0.001 * nTime2, nInputs <= 1 ? 0 : 0.001 * nTime2 / (nInputs-1));
+
     if (fJustCheck)
         return true;
 

src/main.h

@@ -53,6 +53,8 @@ inline bool MoneyRange(int64 nValue) { return (nValue >= 0 && nValue <= MAX_MONE
 static const int COINBASE_MATURITY = 100;
 /** Threshold for nLockTime: below this value it is interpreted as block number, otherwise as UNIX timestamp. */
 static const unsigned int LOCKTIME_THRESHOLD = 500000000; // Tue Nov  5 00:53:20 1985 UTC
+/** Maximum number of script-checking threads allowed */
+static const int MAX_SCRIPTCHECK_THREADS = 16;
 #ifdef USE_UPNP
 static const int fHaveUPnP = true;
 #else
@@ -90,6 +92,7 @@ extern unsigned char pchMessageStart[4];
 extern bool fImporting;
 extern bool fReindex;
 extern bool fBenchmark;
+extern int nScriptCheckThreads;
 extern unsigned int nCoinCacheSize;
 
 // Settings
@@ -107,6 +110,7 @@ class CCoins;
 class CTxUndo;
 class CCoinsView;
 class CCoinsViewCache;
+class CScriptCheck;
 
 /** Register a wallet to receive updates from core */
 void RegisterWallet(CWallet* pwalletIn);
@@ -136,6 +140,10 @@ bool ProcessMessages(CNode* pfrom);
 bool SendMessages(CNode* pto, bool fSendTrickle);
 /** Run the importer thread, which deals with reindexing, loading bootstrap.dat, and whatever is passed to -loadblock */
 void ThreadImport(void *parg);
+/** Run an instance of the script checking thread */
+void ThreadScriptCheck(void* parg);
+/** Stop the script checking threads */
+void ThreadScriptCheckQuit();
 /** Run the miner threads */
 void GenerateBitcoins(bool fGenerate, CWallet* pwallet);
 /** Generate a new block, without valid proof-of-work */
@@ -633,8 +641,11 @@ public:
     bool HaveInputs(CCoinsViewCache &view) const;
 
     // Check whether all inputs of this transaction are valid (no double spends, scripts & sigs, amounts)
-    // This does not modify the UTXO set
-    bool CheckInputs(CCoinsViewCache &view, bool fScriptChecks = true, unsigned int flags = SCRIPT_VERIFY_P2SH | SCRIPT_VERIFY_STRICTENC) const;
+    // This does not modify the UTXO set. If pvChecks is not NULL, script checks are pushed onto it
+    // instead of being performed inline.
+    bool CheckInputs(CCoinsViewCache &view, bool fScriptChecks = true,
+                     unsigned int flags = SCRIPT_VERIFY_P2SH | SCRIPT_VERIFY_STRICTENC,
+                     std::vector<CScriptCheck> *pvChecks = NULL) const;
 
     // Apply the effects of this transaction on the UTXO set represented by view
     bool UpdateCoins(CCoinsViewCache &view, CTxUndo &txundo, int nHeight, const uint256 &txhash) const;

src/net.h

@@ -82,6 +82,7 @@ enum threadId
     THREAD_DUMPADDRESS,
     THREAD_RPCHANDLER,
     THREAD_IMPORT,
+    THREAD_SCRIPTCHECK,
 
     THREAD_MAX
 };

src/test/test_bitcoin.cpp

@@ -33,9 +33,13 @@ struct TestingSetup {
         pwalletMain = new CWallet("wallet.dat");
         pwalletMain->LoadWallet(fFirstRun);
         RegisterWallet(pwalletMain);
+        nScriptCheckThreads = 3;
+        for (int i=0; i < nScriptCheckThreads-1; i++)
+            NewThread(ThreadScriptCheck, NULL);
     }
     ~TestingSetup()
     {
+        ThreadScriptCheckQuit();
         delete pwalletMain;
         pwalletMain = NULL;
         delete pcoinsTip;