fix: 0-initialize pubkey arrays to avoid non-deterministic data for c…

…ompressed pubkeys BREAKING CHANGE: Reparsing the blockchain is strongly recommended. In v0.6 the format of raw pubkeys was changed from a dedicated class (CPubKey) to an array (std::array<unsigned char, 65>). The dedicated class handled compressed pubkeys with 33 bytes correctly, while the new array left the remaining bytes uninitialized. This caused non-deterministic parser outputs and incorrect handling of multisig addresses. Closes #367
citp · Jan 31, 2020 · fc34ac3 · fc34ac3
1 parent bccf558
commit fc34ac3
Show file tree

Hide file tree

Showing 4 changed files with 27 additions and 5 deletions.
diff --git a/include/blocksci/core/script_data.hpp b/include/blocksci/core/script_data.hpp
@@ -12,6 +12,7 @@
 #include <blocksci/blocksci_export.h>
 #include <blocksci/core/bitcoin_uint256.hpp>
 #include <blocksci/core/raw_address.hpp>
+#include <blocksci/scripts/bitcoin_pubkey.hpp>
 
 #include <array>
 #include <limits>
@@ -107,8 +108,16 @@ namespace blocksci {
         };
         bool hasPubkey;
 
-        PubkeyData(uint32_t txNum, const RawPubkey &pubkey_) : ScriptDataBase(txNum), pubkey(pubkey_), hasPubkey(true) {}
-        PubkeyData(uint32_t txNum, const uint160 &address_) : ScriptDataBase(txNum), address(address_), hasPubkey(false) {}
+        PubkeyData(uint32_t txNum, const RawPubkey &pubkey_) : ScriptDataBase(txNum), hasPubkey(true) {
+            pubkey.fill(0);
+            auto itBegin = pubkey_.begin();
+            auto itEnd = itBegin + blocksci::CPubKey::GetLen(pubkey_[0]);
+            std::copy(itBegin, itEnd, pubkey.begin());
+        }
+        PubkeyData(uint32_t txNum, const uint160 &address_) : ScriptDataBase(txNum), hasPubkey(false) {
+            pubkey.fill(0);
+            address = address_;
+        }
 
         size_t size() {
             return sizeof(PubkeyData);
@@ -123,7 +132,10 @@ namespace blocksci {
         RawAddress wrappedAddress;
         bool isSegwit;
 
-        ScriptHashData(uint32_t txNum, uint160 hash160_, const RawAddress &wrappedAddress_) : ScriptDataBase(txNum), hash160(hash160_), wrappedAddress(wrappedAddress_), isSegwit(false) {}
+        ScriptHashData(uint32_t txNum, uint160 hash160_, const RawAddress &wrappedAddress_) : ScriptDataBase(txNum), wrappedAddress(wrappedAddress_), isSegwit(false) {
+            hash256.SetNull();
+            hash160 = hash160_;
+        }
 
         ScriptHashData(uint32_t txNum, uint256 hash256_, const RawAddress &wrappedAddress_) : ScriptDataBase(txNum), hash256(hash256_), wrappedAddress(wrappedAddress_), isSegwit(true) {}
 

diff --git a/tools/parser/script_input.cpp b/tools/parser/script_input.cpp
@@ -89,6 +89,7 @@ void ScriptInputData<blocksci::AddressType::Enum::SCRIPTHASH>::check(AddressStat
 }
 
 ScriptInputData<blocksci::AddressType::Enum::PUBKEYHASH>::ScriptInputData(const InputView &inputView, const blocksci::CScriptView &scriptView, const RawTransaction &, const SpendData<blocksci::AddressType::Enum::PUBKEYHASH> &) {
+    pubkey.fill(0);
     if (scriptView.size() > 0) {
         auto pc = scriptView.begin();
         blocksci::opcodetype opcode = blocksci::OP_0;
@@ -173,6 +174,7 @@ ScriptInputData<blocksci::AddressType::Enum::NULL_DATA>::ScriptInputData(const I
 }
 
 ScriptInputData<blocksci::AddressType::Enum::WITNESS_PUBKEYHASH>::ScriptInputData(const InputView &inputView, const blocksci::CScriptView &, const RawTransaction &, const SpendData<blocksci::AddressType::Enum::WITNESS_PUBKEYHASH> &) {
+    pubkey.fill(0);
     auto &pubkeyWitness = inputView.witnessStack[1];
     std::copy(reinterpret_cast<const unsigned char *>(pubkeyWitness.itemBegin), reinterpret_cast<const unsigned char *>(pubkeyWitness.itemBegin) + pubkeyWitness.length, pubkey.begin());
 }

diff --git a/tools/parser/script_output.cpp b/tools/parser/script_output.cpp
@@ -176,6 +176,7 @@ uint32_t AnyScriptOutput::resolve(AddressState &state) {
 // MARK: TX_PUBKEY
 
 ScriptOutputData<blocksci::AddressType::Enum::PUBKEY>::ScriptOutputData(const ranges::iterator_range<const unsigned char *> &vch1) {
+    pubkey.fill(0);
     std::copy(vch1.begin(), vch1.end(), pubkey.begin());
 }
 
@@ -284,6 +285,7 @@ blocksci::uint160 ScriptOutputData<blocksci::AddressType::Enum::MULTISIG>::getHa
 
 void ScriptOutputData<blocksci::AddressType::Enum::MULTISIG>::addAddress(const ranges::iterator_range<const unsigned char *> &vch1) {
     blocksci::RawPubkey pubkey;
+    pubkey.fill(0);
     std::copy(vch1.begin(), vch1.end(), pubkey.begin());
     addresses.emplace_back(pubkey);
     addressCount++;

diff --git a/tools/parser/script_output.hpp b/tools/parser/script_output.hpp
@@ -64,7 +64,10 @@ struct ScriptOutputData<blocksci::AddressType::Enum::PUBKEY> : public ScriptOutp
     blocksci::RawPubkey pubkey;
 
     ScriptOutputData(const ranges::iterator_range<const unsigned char *> &vch1);
-    ScriptOutputData(const blocksci::RawPubkey &pub) : pubkey(pub) {}
+    ScriptOutputData(const blocksci::RawPubkey &pub) {
+        pubkey.fill(0);
+        pubkey = pub;
+    }
     ScriptOutputData() = default;
 
     blocksci::uint160 getHash() const;
@@ -91,7 +94,10 @@ struct ScriptOutputData<blocksci::AddressType::Enum::MULTISIG_PUBKEY> : public S
     blocksci::RawPubkey pubkey;
 
     ScriptOutputData(const ranges::iterator_range<const unsigned char *> &vch1);
-    ScriptOutputData(const blocksci::RawPubkey &pub) : pubkey(pub) {}
+    ScriptOutputData(const blocksci::RawPubkey &pub) {
+        pubkey.fill(0);
+        pubkey = pub;
+    }
     ScriptOutputData() = default;
 
     blocksci::uint160 getHash() const;