Merge #16899: UTXO snapshot creation (dumptxoutset)

92b2f5306ba0b3f031293cb8f415b67cb002c2f1 test: add dumptxoutset RPC test (James O'Beirne)
c1ccbc3ddef931896a7e9dcfa6704e305a69fbff devtools: add utxo_snapshot.sh (James O'Beirne)
57cf74c9918d10c69a46e6ceb3cb1a5e04edf5bc rpc: add dumptxoutset (James O'Beirne)
92fafb3a7da66f737e960e541fcfbcadedf6043a coinstats: add coins_count (James O'Beirne)
707fde7b9ba522c22179e2db0ed7b462c65138d9 add unused SnapshotMetadata class (James O'Beirne)

Pull request description:

  This is part of the [assumeutxo project](https://github.com/bitcoin/bitcoin/projects/11):

  Parent PR: #15606
  Issue: #15605
  Specification: https://github.com/jamesob/assumeutxo-docs/tree/master/proposal

  ---

  This changeset defines the serialization format for UTXO snapshots and adds an RPC command for creating them, `dumptxoutset`. It also adds a convenience script for generating and verifying snapshots at a certain height, since that requires doing a hacky rewind of the chain via `invalidateblock`.

  All of this is unused at the moment.

ACKs for top commit:
  laanwj:
    ACK 92b2f5306ba0b3f031293cb8f415b67cb002c2f1

Tree-SHA512: 200dff87767f157d627e99506ec543465d9329860a6cd49363081619c437163a640a46d008faa92b1f44fd403bfc7a7c9e851c658b5a4849efa9a34ca976bf31

6 files changed, 174 insertions, 10 deletions

diff --git a/src/Makefile.am b/src/Makefile.am
index 507e5cbb9f..8308388331 100644
--- a/src/Makefile.am
+++ b/src/Makefile.am
@@ -161,6 +161,7 @@ BITCOIN_CORE_H = \
   node/context.h \
   node/psbt.h \
   node/transaction.h \
+  node/utxo_snapshot.h \
   noui.h \
   optional.h \
   outputtype.h \
diff --git a/src/node/coinstats.cpp b/src/node/coinstats.cpp
index 57fa158ad2..23269cd4b0 100644
--- a/src/node/coinstats.cpp
+++ b/src/node/coinstats.cpp
@@ -38,6 +38,7 @@ static void ApplyStats(CCoinsStats &stats, CHashWriter& ss, const uint256& hash,
 //! Calculate statistics about the unspent transaction output set
 bool GetUTXOStats(CCoinsView *view, CCoinsStats &stats)
 {
+    stats = CCoinsStats();
     std::unique_ptr<CCoinsViewCursor> pcursor(view->Cursor());
     assert(pcursor);
 
@@ -61,6 +62,7 @@ bool GetUTXOStats(CCoinsView *view, CCoinsStats &stats)
             }
             prevkey = key.hash;
             outputs[key.n] = std::move(coin);
+            stats.coins_count++;
         } else {
             return error("%s: unable to read value", __func__);
         }
diff --git a/src/node/coinstats.h b/src/node/coinstats.h
index 7c11aab8bd..a19af0fd1b 100644
--- a/src/node/coinstats.h
+++ b/src/node/coinstats.h
@@ -15,16 +15,17 @@ class CCoinsView;
 
 struct CCoinsStats
 {
-    int nHeight;
-    uint256 hashBlock;
-    uint64_t nTransactions;
-    uint64_t nTransactionOutputs;
-    uint64_t nBogoSize;
-    uint256 hashSerialized;
-    uint64_t nDiskSize;
-    CAmount nTotalAmount;
-
-    CCoinsStats() : nHeight(0), nTransactions(0), nTransactionOutputs(0), nBogoSize(0), nDiskSize(0), nTotalAmount(0) {}
+    int nHeight{0};
+    uint256 hashBlock{};
+    uint64_t nTransactions{0};
+    uint64_t nTransactionOutputs{0};
+    uint64_t nBogoSize{0};
+    uint256 hashSerialized{};
+    uint64_t nDiskSize{0};
+    CAmount nTotalAmount{0};
+
+    //! The number of coins contained.
+    uint64_t coins_count{0};
 };
 
 //! Calculate statistics about the unspent transaction output set
diff --git a/src/node/utxo_snapshot.h b/src/node/utxo_snapshot.h
new file mode 100644
index 0000000000..702a0cbe53
--- /dev/null
+++ b/src/node/utxo_snapshot.h
@@ -0,0 +1,50 @@
+// Copyright (c) 2009-2010 Satoshi Nakamoto
+// Copyright (c) 2009-2019 The Bitcoin Core developers
+// Distributed under the MIT software license, see the accompanying
+// file COPYING or http://www.opensource.org/licenses/mit-license.php.
+
+#ifndef BITCOIN_NODE_UTXO_SNAPSHOT_H
+#define BITCOIN_NODE_UTXO_SNAPSHOT_H
+
+#include <uint256.h>
+#include <serialize.h>
+
+//! Metadata describing a serialized version of a UTXO set from which an
+//! assumeutxo CChainState can be constructed.
+class SnapshotMetadata
+{
+public:
+    //! The hash of the block that reflects the tip of the chain for the
+    //! UTXO set contained in this snapshot.
+    uint256 m_base_blockhash;
+
+    //! The number of coins in the UTXO set contained in this snapshot. Used
+    //! during snapshot load to estimate progress of UTXO set reconstruction.
+    uint64_t m_coins_count = 0;
+
+    //! Necessary to "fake" the base nChainTx so that we can estimate progress during
+    //! initial block download for the assumeutxo chainstate.
+    unsigned int m_nchaintx = 0;
+
+    SnapshotMetadata() { }
+    SnapshotMetadata(
+        const uint256& base_blockhash,
+        uint64_t coins_count,
+        unsigned int nchaintx) :
+            m_base_blockhash(base_blockhash),
+            m_coins_count(coins_count),
+            m_nchaintx(nchaintx) { }
+
+    ADD_SERIALIZE_METHODS;
+
+    template <typename Stream, typename Operation>
+    inline void SerializationOp(Stream& s, Operation ser_action)
+    {
+        READWRITE(m_base_blockhash);
+        READWRITE(m_coins_count);
+        READWRITE(m_nchaintx);
+    }
+
+};
+
+#endif // BITCOIN_NODE_UTXO_SNAPSHOT_H
diff --git a/src/rpc/blockchain.cpp b/src/rpc/blockchain.cpp
index d08f852751..d2ac12f3cf 100644
--- a/src/rpc/blockchain.cpp
+++ b/src/rpc/blockchain.cpp
@@ -15,6 +15,7 @@
 #include <hash.h>
 #include <index/blockfilterindex.h>
 #include <node/coinstats.h>
+#include <node/utxo_snapshot.h>
 #include <policy/feerate.h>
 #include <policy/policy.h>
 #include <policy/rbf.h>
@@ -2245,6 +2246,113 @@ static UniValue getblockfilter(const JSONRPCRequest& request)
     return ret;
 }
 
+/**
+ * Serialize the UTXO set to a file for loading elsewhere.
+ *
+ * @see SnapshotMetadata
+ */
+UniValue dumptxoutset(const JSONRPCRequest& request)
+{
+    RPCHelpMan{
+        "dumptxoutset",
+        "\nWrite the serialized UTXO set to disk.\n"
+        "Incidentally flushes the latest coinsdb (leveldb) to disk.\n",
+        {
+            {"path",
+                RPCArg::Type::STR,
+                RPCArg::Optional::NO,
+                /* default_val */ "",
+                "path to the output file. If relative, will be prefixed by datadir."},
+        },
+        RPCResult{
+            "{\n"
+            "  \"coins_written\": n,   (numeric) the number of coins written in the snapshot\n"
+            "  \"base_hash\": \"...\",   (string) the hash of the base of the snapshot\n"
+            "  \"base_height\": n,     (string) the height of the base of the snapshot\n"
+            "  \"path\": \"...\"         (string) the absolute path that the snapshot was written to\n"
+            "]\n"
+        },
+        RPCExamples{
+            HelpExampleCli("dumptxoutset", "utxo.dat")
+        }
+    }.Check(request);
+
+    fs::path path = fs::absolute(request.params[0].get_str(), GetDataDir());
+    // Write to a temporary path and then move into `path` on completion
+    // to avoid confusion due to an interruption.
+    fs::path temppath = fs::absolute(request.params[0].get_str() + ".incomplete", GetDataDir());
+
+    if (fs::exists(path)) {
+        throw JSONRPCError(
+            RPC_INVALID_PARAMETER,
+            path.string() + " already exists. If you are sure this is what you want, "
+            "move it out of the way first");
+    }
+
+    FILE* file{fsbridge::fopen(temppath, "wb")};
+    CAutoFile afile{file, SER_DISK, CLIENT_VERSION};
+    std::unique_ptr<CCoinsViewCursor> pcursor;
+    CCoinsStats stats;
+    CBlockIndex* tip;
+
+    {
+        // We need to lock cs_main to ensure that the coinsdb isn't written to
+        // between (i) flushing coins cache to disk (coinsdb), (ii) getting stats
+        // based upon the coinsdb, and (iii) constructing a cursor to the
+        // coinsdb for use below this block.
+        //
+        // Cursors returned by leveldb iterate over snapshots, so the contents
+        // of the pcursor will not be affected by simultaneous writes during
+        // use below this block.
+        //
+        // See discussion here:
+        //   https://github.com/bitcoin/bitcoin/pull/15606#discussion_r274479369
+        //
+        LOCK(::cs_main);
+
+        ::ChainstateActive().ForceFlushStateToDisk();
+
+        if (!GetUTXOStats(&::ChainstateActive().CoinsDB(), stats)) {
+            throw JSONRPCError(RPC_INTERNAL_ERROR, "Unable to read UTXO set");
+        }
+
+        pcursor = std::unique_ptr<CCoinsViewCursor>(::ChainstateActive().CoinsDB().Cursor());
+        tip = LookupBlockIndex(stats.hashBlock);
+        CHECK_NONFATAL(tip);
+    }
+
+    SnapshotMetadata metadata{tip->GetBlockHash(), stats.coins_count, tip->nChainTx};
+
+    afile << metadata;
+
+    COutPoint key;
+    Coin coin;
+    unsigned int iter{0};
+
+    while (pcursor->Valid()) {
+        if (iter % 5000 == 0 && !IsRPCRunning()) {
+            throw JSONRPCError(RPC_CLIENT_NOT_CONNECTED, "Shutting down");
+        }
+        ++iter;
+        if (pcursor->GetKey(key) && pcursor->GetValue(coin)) {
+            afile << key;
+            afile << coin;
+        }
+
+        pcursor->Next();
+    }
+
+    afile.fclose();
+    fs::rename(temppath, path);
+
+    UniValue result(UniValue::VOBJ);
+    result.pushKV("coins_written", stats.coins_count);
+    result.pushKV("base_hash", tip->GetBlockHash().ToString());
+    result.pushKV("base_height", tip->nHeight);
+    result.pushKV("path", path.string());
+    return result;
+}
+
 // clang-format off
 static const CRPCCommand commands[] =
 { //  category              name                      actor (function)         argNames
@@ -2281,6 +2389,7 @@ static const CRPCCommand commands[] =
     { "hidden",             "waitforblock",           &waitforblock,           {"blockhash","timeout"} },
     { "hidden",             "waitforblockheight",     &waitforblockheight,     {"height","timeout"} },
     { "hidden",             "syncwithvalidationinterfacequeue", &syncwithvalidationinterfacequeue, {} },
+    { "hidden",             "dumptxoutset",           &dumptxoutset,           {"path"} },
 };
 // clang-format on
 
diff --git a/src/validation.h b/src/validation.h
index 7f9582adfd..e9127040bf 100644
--- a/src/validation.h
+++ b/src/validation.h
@@ -21,6 +21,7 @@
 #include <txmempool.h> // For CTxMemPool::cs
 #include <txdb.h>
 #include <versionbits.h>
+#include <serialize.h>
 
 #include <algorithm>
 #include <atomic>