rpc: Optimize serialization disk space of dumptxoutset

Co-authored-by: Aurèle Oulès <aurele@oules.com>
Co-authored-by: TheCharlatan <seb.kung@gmail.com>

Author: 3 people

src/rpc/blockchain.cpp

@@ -34,6 +34,7 @@
 #include <rpc/server_util.h>
 #include <rpc/util.h>
 #include <script/descriptor.h>
+#include <serialize.h>
 #include <streams.h>
 #include <sync.h>
 #include <txdb.h>
@@ -2695,24 +2696,48 @@ UniValue CreateUTXOSnapshot(
     afile << metadata;
 
     COutPoint key;
+    Txid last_hash;
     Coin coin;
     unsigned int iter{0};
+    size_t written_coins_count{0};
+    std::vector<std::pair<uint32_t, Coin>> coins;
+
+    auto write_coins_to_file = [&](AutoFile& afile, const Txid& last_hash, const std::vector<std::pair<uint32_t, Coin>>& coins, size_t& written_coins_count) {
+        afile << last_hash;
+        WriteCompactSize(afile, coins.size());
+        for (const auto& [n, coin] : coins) {
+            WriteCompactSize(afile, n);
+            afile << coin;
+            ++written_coins_count;
+        }
+    };
 
+    pcursor->GetKey(key);
+    last_hash = key.hash;
     while (pcursor->Valid()) {
         if (iter % 5000 == 0) node.rpc_interruption_point();
         ++iter;
         if (pcursor->GetKey(key) && pcursor->GetValue(coin)) {
-            afile << key;
-            afile << coin;
+            if (key.hash != last_hash) {
+                write_coins_to_file(afile, last_hash, coins, written_coins_count);
+                last_hash = key.hash;
+                coins.clear();
+            }
+            coins.emplace_back(key.n, coin);
         }
-
         pcursor->Next();
     }
 
+    if (!coins.empty()) {
+        write_coins_to_file(afile, last_hash, coins, written_coins_count);
+    }
+
+    CHECK_NONFATAL(written_coins_count == maybe_stats->coins_count);
+
     afile.fclose();
 
     UniValue result(UniValue::VOBJ);
-    result.pushKV("coins_written", maybe_stats->coins_count);
+    result.pushKV("coins_written", written_coins_count);
     result.pushKV("base_hash", tip->GetBlockHash().ToString());
     result.pushKV("base_height", tip->nHeight);
     result.pushKV("path", path.utf8string());

src/test/validation_chainstatemanager_tests.cpp

@@ -226,10 +226,13 @@ struct SnapshotTestSetup : TestChain100Setup {
                 // A UTXO is missing but count is correct
                 metadata.m_coins_count -= 1;
 
-                COutPoint outpoint;
+                Txid txid;
+                auto_infile >> txid;
+                // coins size
+                (void)ReadCompactSize(auto_infile);
+                // vout index
+                (void)ReadCompactSize(auto_infile);
                 Coin coin;
-
-                auto_infile >> outpoint;
                 auto_infile >> coin;
         }));
 

src/validation.cpp

@@ -5660,69 +5660,81 @@ bool ChainstateManager::PopulateAndValidateSnapshot(
         return false;
     }
 
-    COutPoint outpoint;
-    Coin coin;
     const uint64_t coins_count = metadata.m_coins_count;
     uint64_t coins_left = metadata.m_coins_count;
 
-    LogPrintf("[snapshot] loading coins from snapshot %s\n", base_blockhash.ToString());
+    LogPrintf("[snapshot] loading %d coins from snapshot %s\n", coins_left, base_blockhash.ToString());
     int64_t coins_processed{0};
 
     while (coins_left > 0) {
         try {
-            coins_file >> outpoint;
-            coins_file >> coin;
-        } catch (const std::ios_base::failure&) {
-            LogPrintf("[snapshot] bad snapshot format or truncated snapshot after deserializing %d coins\n",
-                      coins_count - coins_left);
-            return false;
-        }
-        if (coin.nHeight > base_height ||
-            outpoint.n >= std::numeric_limits<decltype(outpoint.n)>::max() // Avoid integer wrap-around in coinstats.cpp:ApplyHash
-        ) {
-            LogPrintf("[snapshot] bad snapshot data after deserializing %d coins\n",
-                      coins_count - coins_left);
-            return false;
-        }
-        if (!MoneyRange(coin.out.nValue)) {
-            LogPrintf("[snapshot] bad snapshot data after deserializing %d coins - bad tx out value\n",
-                      coins_count - coins_left);
-            return false;
-        }
+            Txid txid;
+            coins_file >> txid;
+            size_t coins_per_txid{0};
+            coins_per_txid = ReadCompactSize(coins_file);
+
+            if (coins_per_txid > coins_left) {
+                LogPrintf("[snapshot] mismatch in coins count in snapshot metadata and actual snapshot data\n");
+                return false;
+            }
 
-        coins_cache.EmplaceCoinInternalDANGER(std::move(outpoint), std::move(coin));
+            for (size_t i = 0; i < coins_per_txid; i++) {
+                COutPoint outpoint;
+                Coin coin;
+                outpoint.n = static_cast<uint32_t>(ReadCompactSize(coins_file));
+                outpoint.hash = txid;
+                coins_file >> coin;
+                if (coin.nHeight > base_height ||
+                    outpoint.n >= std::numeric_limits<decltype(outpoint.n)>::max() // Avoid integer wrap-around in coinstats.cpp:ApplyHash
+                ) {
+                    LogPrintf("[snapshot] bad snapshot data after deserializing %d coins\n",
+                              coins_count - coins_left);
+                    return false;
+                }
+                if (!MoneyRange(coin.out.nValue)) {
+                    LogPrintf("[snapshot] bad snapshot data after deserializing %d coins - bad tx out value\n",
+                              coins_count - coins_left);
+                    return false;
+                }
+                coins_cache.EmplaceCoinInternalDANGER(std::move(outpoint), std::move(coin));
 
-        --coins_left;
-        ++coins_processed;
+                --coins_left;
+                ++coins_processed;
 
-        if (coins_processed % 1000000 == 0) {
-            LogPrintf("[snapshot] %d coins loaded (%.2f%%, %.2f MB)\n",
-                coins_processed,
-                static_cast<float>(coins_processed) * 100 / static_cast<float>(coins_count),
-                coins_cache.DynamicMemoryUsage() / (1000 * 1000));
-        }
+                if (coins_processed % 1000000 == 0) {
+                    LogPrintf("[snapshot] %d coins loaded (%.2f%%, %.2f MB)\n",
+                        coins_processed,
+                        static_cast<float>(coins_processed) * 100 / static_cast<float>(coins_count),
+                        coins_cache.DynamicMemoryUsage() / (1000 * 1000));
+                }
 
-        // Batch write and flush (if we need to) every so often.
-        //
-        // If our average Coin size is roughly 41 bytes, checking every 120,000 coins
-        // means <5MB of memory imprecision.
-        if (coins_processed % 120000 == 0) {
-            if (m_interrupt) {
-                return false;
-            }
+                // Batch write and flush (if we need to) every so often.
+                //
+                // If our average Coin size is roughly 41 bytes, checking every 120,000 coins
+                // means <5MB of memory imprecision.
+                if (coins_processed % 120000 == 0) {
+                    if (m_interrupt) {
+                        return false;
+                    }
 
-            const auto snapshot_cache_state = WITH_LOCK(::cs_main,
-                return snapshot_chainstate.GetCoinsCacheSizeState());
+                    const auto snapshot_cache_state = WITH_LOCK(::cs_main,
+                        return snapshot_chainstate.GetCoinsCacheSizeState());
 
-            if (snapshot_cache_state >= CoinsCacheSizeState::CRITICAL) {
-                // This is a hack - we don't know what the actual best block is, but that
-                // doesn't matter for the purposes of flushing the cache here. We'll set this
-                // to its correct value (`base_blockhash`) below after the coins are loaded.
-                coins_cache.SetBestBlock(GetRandHash());
+                    if (snapshot_cache_state >= CoinsCacheSizeState::CRITICAL) {
+                        // This is a hack - we don't know what the actual best block is, but that
+                        // doesn't matter for the purposes of flushing the cache here. We'll set this
+                        // to its correct value (`base_blockhash`) below after the coins are loaded.
+                        coins_cache.SetBestBlock(GetRandHash());
 
-                // No need to acquire cs_main since this chainstate isn't being used yet.
-                FlushSnapshotToDisk(coins_cache, /*snapshot_loaded=*/false);
+                        // No need to acquire cs_main since this chainstate isn't being used yet.
+                        FlushSnapshotToDisk(coins_cache, /*snapshot_loaded=*/false);
+                    }
+                }
             }
+        } catch (const std::ios_base::failure&) {
+            LogPrintf("[snapshot] bad snapshot format or truncated snapshot after deserializing %d coins\n",
+                      coins_processed);
+            return false;
         }
     }
 
@@ -5735,7 +5747,8 @@ bool ChainstateManager::PopulateAndValidateSnapshot(
 
     bool out_of_coins{false};
     try {
-        coins_file >> outpoint;
+        Txid txid;
+        coins_file >> txid;
     } catch (const std::ios_base::failure&) {
         // We expect an exception since we should be out of coins.
         out_of_coins = true;

test/functional/feature_assumeutxo.py

@@ -94,15 +94,16 @@ def expected_error(log_msg="", rpc_details=""):
                 f.write(valid_snapshot_contents[32 + 8:])
             expected_error(log_msg=f"bad snapshot - coins left over after deserializing 298 coins" if off == -1 else f"bad snapshot format or truncated snapshot after deserializing 299 coins")
 
-        self.log.info("  - snapshot file with alternated UTXO data")
+
+        self.log.info("  - snapshot file with alternated but parsable UTXO data results in different hash")
         cases = [
             # (content, offset, wrong_hash, custom_message)
             [b"\xff" * 32, 0, "7d52155c9a9fdc4525b637ef6170568e5dad6fabd0b1fdbb9432010b8453095b", None],  # wrong outpoint hash
-            [(1).to_bytes(4, "little"), 32, "9f4d897031ab8547665b4153317ae2fdbf0130c7840b66427ebc48b881cb80ad", None],  # wrong outpoint index
-            [b"\x81", 36, "3da966ba9826fb6d2604260e01607b55ba44e1a5de298606b08704bc62570ea8", None],  # wrong coin code VARINT
-            [b"\x80", 36, "091e893b3ccb4334378709578025356c8bcb0a623f37c7c4e493133c988648e5", None],  # another wrong coin code
-            [b"\x84\x58", 36, None, "[snapshot] bad snapshot data after deserializing 0 coins"],  # wrong coin case with height 364 and coinbase 0
-            [b"\xCA\xD2\x8F\x5A", 41, None, "[snapshot] bad snapshot data after deserializing 0 coins - bad tx out value"],  # Amount exceeds MAX_MONEY
+            [b"\x01", 33, "9f4d897031ab8547665b4153317ae2fdbf0130c7840b66427ebc48b881cb80ad", None],  # wrong outpoint index
+            [b"\x81", 34, "3da966ba9826fb6d2604260e01607b55ba44e1a5de298606b08704bc62570ea8", None],  # wrong coin code VARINT
+            [b"\x80", 34, "091e893b3ccb4334378709578025356c8bcb0a623f37c7c4e493133c988648e5", None],  # another wrong coin code
+            [b"\x84\x58", 34, None, "[snapshot] bad snapshot data after deserializing 0 coins"],  # wrong coin case with height 364 and coinbase 0
+            [b"\xCA\xD2\x8F\x5A", 39, None, "[snapshot] bad snapshot data after deserializing 0 coins - bad tx out value"],  # Amount exceeds MAX_MONEY
         ]
 
         for content, offset, wrong_hash, custom_message in cases:

test/functional/rpc_dumptxoutset.py

@@ -43,7 +43,7 @@ def run_test(self):
         # UTXO snapshot hash should be deterministic based on mocked time.
         assert_equal(
             sha256sum_file(str(expected_path)).hex(),
-            'b1bacb602eacf5fbc9a7c2ef6eeb0d229c04e98bdf0c2ea5929012cd0eae3830')
+            '3263fc0311ea46415b85513a59ad8fe67806b3cdce66147175ecb9da768d4a99')
 
         assert_equal(
             out['txoutset_hash'], 'a0b7baa3bf5ccbd3279728f230d7ca0c44a76e9923fca8f32dbfd08d65ea496a')