Merge bitcoin/bitcoin#25527: [kernel 3c/n] Decouple validation cache initialization from ArgsManager

0f3a253 validationcaches: Use size_t for sizes (Carl Dong)
41c5201 validationcaches: Add and use ValidationCacheSizes (Carl Dong)
82d3058 cuckoocache: Check for uint32 overflow in setup_bytes (Carl Dong)
b370164 validationcaches: Abolish arbitrary limit (Carl Dong)
08dbc6e cuckoocache: Return approximate memory size (Carl Dong)
0dbce4b tests: Reduce calls to InitS*Cache() (Carl Dong)

Pull request description:

  This is part of the `libbitcoinkernel` project: #24303, https://github.com/bitcoin/bitcoin/projects/18

  This PR is **_NOT_** dependent on any other PRs.

  -----

  a.k.a. "Stop calling `gArgs.GetIntArg("-maxsigcachesize")` from validation code"

  This PR introduces the `ValidationCacheSizes` struct and its corresponding `ApplyArgsManOptions` function, removing the need to call `gArgs` from `Init{Signature,ScriptExecution}Cache()`. This serves to further decouple `ArgsManager` from `libbitcoinkernel` code.

  More context can be gleaned from the commit messages.

ACKs for top commit:
  glozow:
    re ACK 0f3a253
  theStack:
    Code-review ACK 0f3a253
  ryanofsky:
    Code review ACK 0f3a253. Rebase and comment tweak since last

Tree-SHA512: a492ca608466979807cac25ae3d8ef75d2f1345de52a156aa0d222c5a940f79f1b65db40090de69183cccdb12297ec060f6c64e57a26a155a94fec80e07ea0f7

Author: glozow

ci/test/06_script_b.sh

@@ -46,6 +46,7 @@ if [ "${RUN_TIDY}" = "true" ]; then
           " src/kernel"\
           " src/node/chainstate.cpp"\
           " src/node/mempool_args.cpp"\
+          " src/node/validation_cache_args.cpp"\
           " src/policy/feerate.cpp"\
           " src/policy/packages.cpp"\
           " src/policy/settings.cpp"\

src/Makefile.am

@@ -177,6 +177,7 @@ BITCOIN_CORE_H = \
   kernel/mempool_limits.h \
   kernel/mempool_options.h \
   kernel/mempool_persist.h \
+  kernel/validation_cache_sizes.h \
   key.h \
   key_io.h \
   logging.h \
@@ -207,6 +208,7 @@ BITCOIN_CORE_H = \
   node/psbt.h \
   node/transaction.h \
   node/utxo_snapshot.h \
+  node/validation_cache_args.h \
   noui.h \
   outputtype.h \
   policy/feerate.h \
@@ -390,6 +392,7 @@ libbitcoin_node_a_SOURCES = \
   node/minisketchwrapper.cpp \
   node/psbt.cpp \
   node/transaction.cpp \
+  node/validation_cache_args.cpp \
   noui.cpp \
   policy/fees.cpp \
   policy/fees_args.cpp \

src/bitcoin-chainstate.cpp

@@ -13,6 +13,7 @@
 
 #include <kernel/checks.h>
 #include <kernel/context.h>
+#include <kernel/validation_cache_sizes.h>
 
 #include <chainparams.h>
 #include <consensus/validation.h>
@@ -62,8 +63,9 @@ int main(int argc, char* argv[])
     // Necessary for CheckInputScripts (eventually called by ProcessNewBlock),
     // which will try the script cache first and fall back to actually
     // performing the check with the signature cache.
-    InitSignatureCache();
-    InitScriptExecutionCache();
+    kernel::ValidationCacheSizes validation_cache_sizes{};
+    Assert(InitSignatureCache(validation_cache_sizes.signature_cache_bytes));
+    Assert(InitScriptExecutionCache(validation_cache_sizes.script_execution_cache_bytes));
 
 
     // SETUP: Scheduling and Background Signals

src/cuckoocache.h

@@ -12,7 +12,9 @@
 #include <atomic>
 #include <cmath>
 #include <cstring>
+#include <limits>
 #include <memory>
+#include <optional>
 #include <utility>
 #include <vector>
 
@@ -326,7 +328,7 @@ class cache
     }
 
     /** setup initializes the container to store no more than new_size
-     * elements.
+     * elements and no less than 2 elements.
      *
      * setup should only be called once.
      *
@@ -336,8 +338,8 @@ class cache
     uint32_t setup(uint32_t new_size)
     {
         // depth_limit must be at least one otherwise errors can occur.
-        depth_limit = static_cast<uint8_t>(std::log2(static_cast<float>(std::max((uint32_t)2, new_size))));
         size = std::max<uint32_t>(2, new_size);
+        depth_limit = static_cast<uint8_t>(std::log2(static_cast<float>(size)));
         table.resize(size);
         collection_flags.setup(size);
         epoch_flags.resize(size);
@@ -357,12 +359,21 @@ class cache
      *
      * @param bytes the approximate number of bytes to use for this data
      * structure
-     * @returns the maximum number of elements storable (see setup()
-     * documentation for more detail)
+     * @returns A pair of the maximum number of elements storable (see setup()
+     * documentation for more detail) and the approxmiate total size of these
+     * elements in bytes or std::nullopt if the size requested is too large.
      */
-    uint32_t setup_bytes(size_t bytes)
+    std::optional<std::pair<uint32_t, size_t>> setup_bytes(size_t bytes)
     {
-        return setup(bytes/sizeof(Element));
+        size_t requested_num_elems = bytes / sizeof(Element);
+        if (std::numeric_limits<uint32_t>::max() < requested_num_elems) {
+            return std::nullopt;
+        }
+
+        auto num_elems = setup(bytes/sizeof(Element));
+
+        size_t approx_size_bytes = num_elems * sizeof(Element);
+        return std::make_pair(num_elems, approx_size_bytes);
     }
 
     /** insert loops at most depth_limit times trying to insert a hash

src/init.cpp

@@ -11,6 +11,7 @@
 
 #include <kernel/checks.h>
 #include <kernel/mempool_persist.h>
+#include <kernel/validation_cache_sizes.h>
 
 #include <addrman.h>
 #include <banman.h>
@@ -44,6 +45,7 @@
 #include <node/mempool_args.h>
 #include <node/mempool_persist_args.h>
 #include <node/miner.h>
+#include <node/validation_cache_args.h>
 #include <policy/feerate.h>
 #include <policy/fees.h>
 #include <policy/fees_args.h>
@@ -105,7 +107,9 @@
 #endif
 
 using kernel::DumpMempool;
+using kernel::ValidationCacheSizes;
 
+using node::ApplyArgsManOptions;
 using node::CacheSizes;
 using node::CalculateCacheSizes;
 using node::DEFAULT_PERSIST_MEMPOOL;
@@ -548,7 +552,7 @@ void SetupServerArgs(ArgsManager& argsman)
     argsman.AddArg("-addrmantest", "Allows to test address relay on localhost", ArgsManager::ALLOW_ANY | ArgsManager::DEBUG_ONLY, OptionsCategory::DEBUG_TEST);
     argsman.AddArg("-capturemessages", "Capture all P2P messages to disk", ArgsManager::ALLOW_ANY | ArgsManager::DEBUG_ONLY, OptionsCategory::DEBUG_TEST);
     argsman.AddArg("-mocktime=<n>", "Replace actual time with " + UNIX_EPOCH_TIME + " (default: 0)", ArgsManager::ALLOW_ANY | ArgsManager::DEBUG_ONLY, OptionsCategory::DEBUG_TEST);
-    argsman.AddArg("-maxsigcachesize=<n>", strprintf("Limit sum of signature cache and script execution cache sizes to <n> MiB (default: %u)", DEFAULT_MAX_SIG_CACHE_SIZE), ArgsManager::ALLOW_ANY | ArgsManager::DEBUG_ONLY, OptionsCategory::DEBUG_TEST);
+    argsman.AddArg("-maxsigcachesize=<n>", strprintf("Limit sum of signature cache and script execution cache sizes to <n> MiB (default: %u)", DEFAULT_MAX_SIG_CACHE_BYTES >> 20), ArgsManager::ALLOW_ANY | ArgsManager::DEBUG_ONLY, OptionsCategory::DEBUG_TEST);
     argsman.AddArg("-maxtipage=<n>", strprintf("Maximum tip age in seconds to consider node in initial block download (default: %u)", DEFAULT_MAX_TIP_AGE), ArgsManager::ALLOW_ANY | ArgsManager::DEBUG_ONLY, OptionsCategory::DEBUG_TEST);
     argsman.AddArg("-printpriority", strprintf("Log transaction fee rate in " + CURRENCY_UNIT + "/kvB when mining blocks (default: %u)", DEFAULT_PRINTPRIORITY), ArgsManager::ALLOW_ANY | ArgsManager::DEBUG_ONLY, OptionsCategory::DEBUG_TEST);
     argsman.AddArg("-uacomment=<cmt>", "Append comment to the user agent string", ArgsManager::ALLOW_ANY, OptionsCategory::DEBUG_TEST);
@@ -1115,8 +1119,13 @@ bool AppInitMain(NodeContext& node, interfaces::BlockAndHeaderTipInfo* tip_info)
                   args.GetArg("-datadir", ""), fs::PathToString(fs::current_path()));
     }
 
-    InitSignatureCache();
-    InitScriptExecutionCache();
+    ValidationCacheSizes validation_cache_sizes{};
+    ApplyArgsManOptions(args, validation_cache_sizes);
+    if (!InitSignatureCache(validation_cache_sizes.signature_cache_bytes)
+        || !InitScriptExecutionCache(validation_cache_sizes.script_execution_cache_bytes))
+    {
+        return InitError(strprintf(_("Unable to allocate memory for -maxsigcachesize: '%s' MiB"), args.GetIntArg("-maxsigcachesize", DEFAULT_MAX_SIG_CACHE_BYTES >> 20)));
+    }
 
     int script_threads = args.GetIntArg("-par", DEFAULT_SCRIPTCHECK_THREADS);
     if (script_threads <= 0) {

src/kernel/validation_cache_sizes.h

@@ -0,0 +1,20 @@
+// Copyright (c) 2022 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_KERNEL_VALIDATION_CACHE_SIZES_H
+#define BITCOIN_KERNEL_VALIDATION_CACHE_SIZES_H
+
+#include <script/sigcache.h>
+
+#include <cstddef>
+#include <limits>
+
+namespace kernel {
+struct ValidationCacheSizes {
+    size_t signature_cache_bytes{DEFAULT_MAX_SIG_CACHE_BYTES / 2};
+    size_t script_execution_cache_bytes{DEFAULT_MAX_SIG_CACHE_BYTES / 2};
+};
+}
+
+#endif // BITCOIN_KERNEL_VALIDATION_CACHE_SIZES_H

src/node/validation_cache_args.cpp

@@ -0,0 +1,34 @@
+// Copyright (c) 2022 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 <node/validation_cache_args.h>
+
+#include <kernel/validation_cache_sizes.h>
+
+#include <util/system.h>
+
+#include <algorithm>
+#include <cstddef>
+#include <cstdint>
+#include <memory>
+#include <optional>
+
+using kernel::ValidationCacheSizes;
+
+namespace node {
+void ApplyArgsManOptions(const ArgsManager& argsman, ValidationCacheSizes& cache_sizes)
+{
+    if (auto max_size = argsman.GetIntArg("-maxsigcachesize")) {
+        // 1. When supplied with a max_size of 0, both InitSignatureCache and
+        //    InitScriptExecutionCache create the minimum possible cache (2
+        //    elements). Therefore, we can use 0 as a floor here.
+        // 2. Multiply first, divide after to avoid integer truncation.
+        size_t clamped_size_each = std::max<int64_t>(*max_size, 0) * (1 << 20) / 2;
+        cache_sizes = {
+            .signature_cache_bytes = clamped_size_each,
+            .script_execution_cache_bytes = clamped_size_each,
+        };
+    }
+}
+} // namespace node

src/node/validation_cache_args.h

@@ -0,0 +1,17 @@
+// Copyright (c) 2022 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_VALIDATION_CACHE_ARGS_H
+#define BITCOIN_NODE_VALIDATION_CACHE_ARGS_H
+
+class ArgsManager;
+namespace kernel {
+struct ValidationCacheSizes;
+};
+
+namespace node {
+void ApplyArgsManOptions(const ArgsManager& argsman, kernel::ValidationCacheSizes& cache_sizes);
+} // namespace node
+
+#endif // BITCOIN_NODE_VALIDATION_CACHE_ARGS_H

src/script/sigcache.cpp

@@ -14,6 +14,7 @@
 
 #include <algorithm>
 #include <mutex>
+#include <optional>
 #include <shared_mutex>
 #include <vector>
 
@@ -75,7 +76,7 @@ class CSignatureCache
         std::unique_lock<std::shared_mutex> lock(cs_sigcache);
         setValid.insert(entry);
     }
-    uint32_t setup_bytes(size_t n)
+    std::optional<std::pair<uint32_t, size_t>> setup_bytes(size_t n)
     {
         return setValid.setup_bytes(n);
     }
@@ -92,14 +93,15 @@ static CSignatureCache signatureCache;
 
 // To be called once in AppInitMain/BasicTestingSetup to initialize the
 // signatureCache.
-void InitSignatureCache()
+bool InitSignatureCache(size_t max_size_bytes)
 {
-    // nMaxCacheSize is unsigned. If -maxsigcachesize is set to zero,
-    // setup_bytes creates the minimum possible cache (2 elements).
-    size_t nMaxCacheSize = std::min(std::max((int64_t)0, gArgs.GetIntArg("-maxsigcachesize", DEFAULT_MAX_SIG_CACHE_SIZE) / 2), MAX_MAX_SIG_CACHE_SIZE) * ((size_t) 1 << 20);
-    size_t nElems = signatureCache.setup_bytes(nMaxCacheSize);
-    LogPrintf("Using %zu MiB out of %zu/2 requested for signature cache, able to store %zu elements\n",
-            (nElems*sizeof(uint256)) >>20, (nMaxCacheSize*2)>>20, nElems);
+    auto setup_results = signatureCache.setup_bytes(max_size_bytes);
+    if (!setup_results) return false;
+
+    const auto [num_elems, approx_size_bytes] = *setup_results;
+    LogPrintf("Using %zu MiB out of %zu MiB requested for signature cache, able to store %zu elements\n",
+              approx_size_bytes >> 20, max_size_bytes >> 20, num_elems);
+    return true;
 }
 
 bool CachingTransactionSignatureChecker::VerifyECDSASignature(const std::vector<unsigned char>& vchSig, const CPubKey& pubkey, const uint256& sighash) const

src/script/sigcache.h

@@ -10,14 +10,13 @@
 #include <span.h>
 #include <util/hasher.h>
 
+#include <optional>
 #include <vector>
 
-// DoS prevention: limit cache size to 32MB (over 1000000 entries on 64-bit
+// DoS prevention: limit cache size to 32MiB (over 1000000 entries on 64-bit
 // systems). Due to how we count cache size, actual memory usage is slightly
-// more (~32.25 MB)
-static const unsigned int DEFAULT_MAX_SIG_CACHE_SIZE = 32;
-// Maximum sig cache size allowed
-static const int64_t MAX_MAX_SIG_CACHE_SIZE = 16384;
+// more (~32.25 MiB)
+static constexpr size_t DEFAULT_MAX_SIG_CACHE_BYTES{32 << 20};
 
 class CPubKey;
 
@@ -33,6 +32,6 @@ class CachingTransactionSignatureChecker : public TransactionSignatureChecker
     bool VerifySchnorrSignature(Span<const unsigned char> sig, const XOnlyPubKey& pubkey, const uint256& sighash) const override;
 };
 
-void InitSignatureCache();
+[[nodiscard]] bool InitSignatureCache(size_t max_size_bytes);
 
 #endif // BITCOIN_SCRIPT_SIGCACHE_H