Merge pull request #12958 from ethereum/yul-fuzzer-bound-memory-accesses

Yul grammar generator: Bound memory accesses.

test/libyul/yulInterpreterTests/and_create.yul

@@ -6,8 +6,8 @@
 }
 // ----
 // Trace:
-//   CREATE(0, 0xffffffffffffffffffffffffffffffffffffffff, 0)
+//   CREATE(0, 0, 0)
-//   CREATE(0, 0xffffffffffffffffffffffffffffffffffffffff, 0)
+//   CREATE(0, 0, 0)
 // Memory dump:
 //      0: 0000000000000000000000000000000000000000000000000000000000000001
 // Storage dump:

test/libyul/yulInterpreterTests/and_create2.yul

@@ -8,8 +8,8 @@
 // EVMVersion: >=constantinople
 // ----
 // Trace:
-//   CREATE2(0, 0xffffffffffffffffffffffffffffffffffffffff, 0, 0)
+//   CREATE2(0, 0, 0, 0)
-//   CREATE2(0, 0xffffffffffffffffffffffffffffffffffffffff, 0, 0)
+//   CREATE2(0, 0, 0, 0)
 // Memory dump:
 //      0: 0000000000000000000000000000000000000000000000000000000000000001
 // Storage dump:

test/libyul/yulInterpreterTests/zero_length_reads.yul

@@ -0,0 +1,40 @@
+{
+  returndatacopy(1, 1, 0)
+  calldatacopy(1, 1, 0)
+  extcodecopy(1, 1, 1, 0)
+  codecopy(1, 1, 0)
+  log0(1, 0)
+  log1(1, 0, 1)
+  log2(1, 0, 1, 1)
+  log3(1, 0, 1, 1, 1)
+  log4(1, 0, 1, 1, 1, 1)
+  pop(create(1, 1, 0))
+  pop(create2(1, 1, 0, 1))
+  pop(call(1, 1, 1, 1, 0, 1, 0))
+  pop(callcode(1, 1, 1, 1, 0, 1, 0))
+  pop(delegatecall(1, 1, 1, 0, 1, 0))
+  pop(staticcall(1, 1, 1, 0, 1, 0))
+  return(1, 0)
+}
+// ====
+// EVMVersion: >=constantinople
+// ----
+// Trace:
+//   RETURNDATACOPY(0, 1, 0)
+//   CALLDATACOPY(0, 1, 0)
+//   EXTCODECOPY(1, 0, 1, 0)
+//   CODECOPY(0, 1, 0)
+//   LOG0(0, 0)
+//   LOG1(0, 0, 1)
+//   LOG2(0, 0, 1, 1)
+//   LOG3(0, 0, 1, 1, 1)
+//   LOG4(0, 0, 1, 1, 1, 1)
+//   CREATE(1, 0, 0)
+//   CREATE2(1, 0, 0, 1)
+//   CALL(1, 1, 1, 0, 0, 1, 0)
+//   CALLCODE(1, 1, 1, 0, 0, 1, 0)
+//   DELEGATECALL(1, 1, 0, 0, 1, 0)
+//   STATICCALL(1, 1, 0, 0, 1, 0)
+//   RETURN(0, 0)
+// Memory dump:
+// Storage dump:

test/libyul/yulInterpreterTests/zero_length_reads_and_revert.yul

@@ -0,0 +1,40 @@
+{
+  returndatacopy(1, 1, 0)
+  calldatacopy(1, 1, 0)
+  extcodecopy(1, 1, 1, 0)
+  codecopy(1, 1, 0)
+  log0(1, 0)
+  log1(1, 0, 1)
+  log2(1, 0, 1, 1)
+  log3(1, 0, 1, 1, 1)
+  log4(1, 0, 1, 1, 1, 1)
+  pop(create(1, 1, 0))
+  pop(create2(1, 1, 0, 1))
+  pop(call(1, 1, 1, 1, 0, 1, 0))
+  pop(callcode(1, 1, 1, 1, 0, 1, 0))
+  pop(delegatecall(1, 1, 1, 0, 1, 0))
+  pop(staticcall(1, 1, 1, 0, 1, 0))
+  revert(1, 0)
+}
+// ====
+// EVMVersion: >=constantinople
+// ----
+// Trace:
+//   RETURNDATACOPY(0, 1, 0)
+//   CALLDATACOPY(0, 1, 0)
+//   EXTCODECOPY(1, 0, 1, 0)
+//   CODECOPY(0, 1, 0)
+//   LOG0(0, 0)
+//   LOG1(0, 0, 1)
+//   LOG2(0, 0, 1, 1)
+//   LOG3(0, 0, 1, 1, 1)
+//   LOG4(0, 0, 1, 1, 1, 1)
+//   CREATE(1, 0, 0)
+//   CREATE2(1, 0, 0, 1)
+//   CALL(1, 1, 1, 0, 0, 1, 0)
+//   CALLCODE(1, 1, 1, 0, 0, 1, 0)
+//   DELEGATECALL(1, 1, 0, 0, 1, 0)
+//   STATICCALL(1, 1, 0, 0, 1, 0)
+//   REVERT(0, 0)
+// Memory dump:
+// Storage dump:

test/libyul/yulInterpreterTests/zero_range.yul

@@ -5,6 +5,6 @@
 }
 // ----
 // Trace:
-//   CALLDATACOPY(32, 0, 0)
+//   CALLDATACOPY(0, 0, 0)
 // Memory dump:
 // Storage dump:

test/libyul/yulOptimizerTests/loadResolver/zero_length_reads.yul

@@ -0,0 +1,45 @@
+{
+  returndatacopy(1, 1, 0)
+  calldatacopy(1, 1, 0)
+  extcodecopy(1, 1, 1, 0)
+  codecopy(1, 1, 0)
+  log0(1, 0)
+  log1(1, 0, 1)
+  log2(1, 0, 1, 1)
+  log3(1, 0, 1, 1, 1)
+  log4(1, 0, 1, 1, 1, 1)
+  pop(create(1, 1, 0))
+  pop(create2(1, 1, 0, 1))
+  pop(call(1, 1, 1, 1, 0, 1, 0))
+  pop(callcode(1, 1, 1, 1, 0, 1, 0))
+  pop(delegatecall(1, 1, 1, 0, 1, 0))
+  pop(staticcall(1, 1, 1, 0, 1, 0))
+  return(1, 0)
+}
+// ====
+// EVMVersion: >=constantinople
+// ----
+// step: loadResolver
+//
+// {
+//     {
+//         let _1 := 0
+//         let _2 := 1
+//         returndatacopy(0, _2, _1)
+//         calldatacopy(0, _2, _1)
+//         extcodecopy(_2, 0, _2, _1)
+//         codecopy(0, _2, _1)
+//         log0(0, _1)
+//         log1(0, _1, _2)
+//         log2(0, _1, _2, _2)
+//         log3(0, _1, _2, _2, _2)
+//         log4(0, _1, _2, _2, _2, _2)
+//         pop(create(_2, 0, _1))
+//         pop(create2(_2, 0, _1, _2))
+//         pop(call(_2, _2, _2, 0, _1, _2, _1))
+//         pop(callcode(_2, _2, _2, 0, _1, _2, _1))
+//         pop(delegatecall(_2, _2, 0, _1, _2, _1))
+//         pop(staticcall(_2, _2, 0, _1, _2, _1))
+//         return(0, _1)
+//     }
+// }

test/tools/ossfuzz/protoToYul.cpp

@@ -351,9 +351,22 @@ void ProtoConverter::visit(BinaryOp const& _x)
 		break;
 	}
 	m_output << "(";
+	if (op == BinaryOp::KECCAK)
+	{
+		m_output << "mod(";
+		visit(_x.left());
+		m_output << ", " << to_string(s_maxMemory - s_maxSize) << ")";
+		m_output << ",";
+		m_output << "mod(";
+		visit(_x.right());
+		m_output << ", " << to_string(s_maxSize) << ")";
+	}
+	else
+	{
 		visit(_x.left());
 		m_output << ",";
 		visit(_x.right());
+	}
 	m_output << ")";
 }
 
@@ -623,6 +636,13 @@ void ProtoConverter::visit(UnaryOp const& _x)
 		break;
 	}
 	m_output << "(";
+	if (op == UnaryOp::MLOAD)
+	{
+		m_output << "mod(";
+		visit(_x.operand());
+		m_output << ", " << to_string(s_maxMemory) << ")";
+	}
+	else
 		visit(_x.operand());
 	m_output << ")";
 }
@@ -778,11 +798,15 @@ void ProtoConverter::visit(CopyFunc const& _x)
 		break;
 	}
 	m_output << "(";
+	m_output << "mod(";
 	visit(_x.target());
+	m_output << ", " << to_string(s_maxMemory - s_maxSize) << ")";
 	m_output << ", ";
 	visit(_x.source());
 	m_output << ", ";
+	m_output << "mod(";
 	visit(_x.size());
+	m_output << ", " << to_string(s_maxSize) << ")";
 	m_output << ")\n";
 }
 
@@ -792,32 +816,42 @@ void ProtoConverter::visit(ExtCodeCopy const& _x)
 	m_output << "(";
 	visit(_x.addr());
 	m_output << ", ";
+	m_output << "mod(";
 	visit(_x.target());
+	m_output << ", " << to_string(s_maxMemory - s_maxSize) << ")";
 	m_output << ", ";
 	visit(_x.source());
 	m_output << ", ";
+	m_output << "mod(";
 	visit(_x.size());
+	m_output << ", " << to_string(s_maxSize) << ")";
 	m_output << ")\n";
 }
 
 void ProtoConverter::visit(LogFunc const& _x)
 {
+	auto visitPosAndSize = [&](LogFunc const& _y) {
+		m_output << "mod(";
+		visit(_y.pos());
+		m_output << ", " << to_string(s_maxMemory - s_maxSize) << ")";
+		m_output << ", ";
+		m_output << "mod(";
+		visit(_y.size());
+		m_output << ", " << to_string(s_maxSize) << ")";
+	};
+
 	switch (_x.num_topics())
 	{
 	case LogFunc::ZERO:
 		m_output << "log0";
 		m_output << "(";
-		visit(_x.pos());
+		visitPosAndSize(_x);
-		m_output << ", ";
-		visit(_x.size());
 		m_output << ")\n";
 		break;
 	case LogFunc::ONE:
 		m_output << "log1";
 		m_output << "(";
-		visit(_x.pos());
+		visitPosAndSize(_x);
-		m_output << ", ";
-		visit(_x.size());
 		m_output << ", ";
 		visit(_x.t1());
 		m_output << ")\n";
@@ -825,9 +859,7 @@ void ProtoConverter::visit(LogFunc const& _x)
 	case LogFunc::TWO:
 		m_output << "log2";
 		m_output << "(";
-		visit(_x.pos());
+		visitPosAndSize(_x);
-		m_output << ", ";
-		visit(_x.size());
 		m_output << ", ";
 		visit(_x.t1());
 		m_output << ", ";
@@ -837,9 +869,7 @@ void ProtoConverter::visit(LogFunc const& _x)
 	case LogFunc::THREE:
 		m_output << "log3";
 		m_output << "(";
-		visit(_x.pos());
+		visitPosAndSize(_x);
-		m_output << ", ";
-		visit(_x.size());
 		m_output << ", ";
 		visit(_x.t1());
 		m_output << ", ";
@@ -851,9 +881,7 @@ void ProtoConverter::visit(LogFunc const& _x)
 	case LogFunc::FOUR:
 		m_output << "log4";
 		m_output << "(";
-		visit(_x.pos());
+		visitPosAndSize(_x);
-		m_output << ", ";
-		visit(_x.size());
 		m_output << ", ";
 		visit(_x.t1());
 		m_output << ", ";
@@ -1015,13 +1043,21 @@ void ProtoConverter::visit(LowLevelCall const& _x)
 		visit(_x.wei());
 		m_output << ", ";
 	}
+	m_output << "mod(";
 	visit(_x.in());
+	m_output << ", " << to_string(s_maxMemory - s_maxSize) << ")";
 	m_output << ", ";
+	m_output << "mod(";
 	visit(_x.insize());
+	m_output << ", " << to_string(s_maxSize) << ")";
 	m_output << ", ";
+	m_output << "mod(";
 	visit(_x.out());
+	m_output << ", " << to_string(s_maxMemory - s_maxSize) << ")";
 	m_output << ", ";
+	m_output << "mod(";
 	visit(_x.outsize());
+	m_output << ", " << to_string(s_maxSize) << ")";
 	m_output << ")";
 }
 
@@ -1048,9 +1084,13 @@ void ProtoConverter::visit(Create const& _x)
 	}
 	visit(_x.wei());
 	m_output << ", ";
+	m_output << "mod(";
 	visit(_x.position());
+	m_output << ", " << to_string(s_maxMemory - s_maxSize) << ")";
 	m_output << ", ";
+	m_output << "mod(";
 	visit(_x.size());
+	m_output << ", " << to_string(s_maxSize) << ")";
 	if (type == Create::CREATE2)
 	{
 		m_output << ", ";
@@ -1069,7 +1109,8 @@ void ProtoConverter::visit(IfStmt const& _x)
 
 void ProtoConverter::visit(StoreFunc const& _x)
 {
-	switch (_x.st())
+	auto storeType = _x.st();
+	switch (storeType)
 	{
 	case StoreFunc::MSTORE:
 		m_output << "mstore(";
@@ -1081,7 +1122,15 @@ void ProtoConverter::visit(StoreFunc const& _x)
 		m_output << "mstore8(";
 		break;
 	}
+	// Write to memory within bounds, storage is unbounded
+	if (storeType == StoreFunc::SSTORE)
 		visit(_x.loc());
+	else
+	{
+		m_output << "mod(";
+		visit(_x.loc());
+		m_output << ", " << to_string(s_maxMemory) << ")";
+	}
 	m_output << ", ";
 	visit(_x.val());
 	m_output << ")\n";
@@ -1262,9 +1311,13 @@ void ProtoConverter::visit(RetRevStmt const& _x)
 		break;
 	}
 	m_output << "(";
+	m_output << "mod(";
 	visit(_x.pos());
+	m_output << ", " << to_string(s_maxMemory - s_maxSize) << ")";
 	m_output << ", ";
+	m_output << "mod(";
 	visit(_x.size());
+	m_output << ", " << to_string(s_maxSize) << ")";
 	m_output << ")\n";
 }
 
@@ -1651,8 +1704,12 @@ void ProtoConverter::fillFunctionCallInput(unsigned _numInParams)
 			m_output << "calldataload(" << slot << ")";
 			break;
 		case 1:
+		{
+			// Access memory within stipulated bounds
+			slot = "mod(" + dictionaryToken() + ", " + to_string(s_maxMemory) + ")";
 			m_output << "mload(" << slot << ")";
 			break;
+		}
 		case 2:
 			m_output << "sload(" << slot << ")";
 			break;

test/tools/ossfuzz/protoToYul.h

@@ -344,6 +344,11 @@ private:
 	static unsigned constexpr s_modOutputParams = 5;
 	/// Hard-coded identifier for a Yul object's data block
 	static auto constexpr s_dataIdentifier = "datablock";
+	/// Upper bound on memory writes = 2**32 - 1
+	/// See: https://eips.ethereum.org/EIPS/eip-1985#memory-size
+	static unsigned constexpr s_maxMemory = 4294967295;
+	/// Upper bound on size for range copy functions
+	static unsigned constexpr s_maxSize = 65536;
 	/// Predicate to keep track of for body scope. If false, break/continue
 	/// statements can not be created.
 	bool m_inForBodyScope;

test/tools/yulInterpreter/EVMInstructionInterpreter.cpp

@@ -211,22 +211,22 @@ u256 EVMInstructionInterpreter::eval(
 	case Instruction::CALLDATASIZE:
 		return m_state.calldata.size();
 	case Instruction::CALLDATACOPY:
-		logTrace(_instruction, arg);
 		if (accessMemory(arg[0], arg[2]))
 			copyZeroExtended(
 				m_state.memory, m_state.calldata,
 				size_t(arg[0]), size_t(arg[1]), size_t(arg[2])
 			);
+		logTrace(_instruction, arg);
 		return 0;
 	case Instruction::CODESIZE:
 		return m_state.code.size();
 	case Instruction::CODECOPY:
-		logTrace(_instruction, arg);
 		if (accessMemory(arg[0], arg[2]))
 			copyZeroExtended(
 				m_state.memory, m_state.code,
 				size_t(arg[0]), size_t(arg[1]), size_t(arg[2])
 			);
+		logTrace(_instruction, arg);
 		return 0;
 	case Instruction::GASPRICE:
 		return m_state.gasprice;
@@ -239,23 +239,23 @@ u256 EVMInstructionInterpreter::eval(
 	case Instruction::EXTCODEHASH:
 		return u256(keccak256(h256(arg[0] + 1)));
 	case Instruction::EXTCODECOPY:
-		logTrace(_instruction, arg);
 		if (accessMemory(arg[1], arg[3]))
 			// TODO this way extcodecopy and codecopy do the same thing.
 			copyZeroExtended(
 				m_state.memory, m_state.code,
 				size_t(arg[1]), size_t(arg[2]), size_t(arg[3])
 			);
+		logTrace(_instruction, arg);
 		return 0;
 	case Instruction::RETURNDATASIZE:
 		return m_state.returndata.size();
 	case Instruction::RETURNDATACOPY:
-		logTrace(_instruction, arg);
 		if (accessMemory(arg[0], arg[2]))
 			copyZeroExtended(
 				m_state.memory, m_state.returndata,
 				size_t(arg[0]), size_t(arg[1]), size_t(arg[2])
 			);
+		logTrace(_instruction, arg);
 		return 0;
 	case Instruction::BLOCKHASH:
 		if (arg[0] >= m_state.blockNumber || arg[0] + 256 < m_state.blockNumber)
@@ -319,11 +319,17 @@ u256 EVMInstructionInterpreter::eval(
 	case Instruction::CREATE:
 		accessMemory(arg[1], arg[2]);
 		logTrace(_instruction, arg);
+		if (arg[2] != 0)
 			return (0xcccccc + arg[1]) & u256("0xffffffffffffffffffffffffffffffffffffffff");
+		else
+			return 0xcccccc;
 	case Instruction::CREATE2:
 		accessMemory(arg[1], arg[2]);
 		logTrace(_instruction, arg);
+		if (arg[2] != 0)
 			return (0xdddddd + arg[1]) & u256("0xffffffffffffffffffffffffffffffffffffffff");
+		else
+			return 0xdddddd;
 	case Instruction::CALL:
 	case Instruction::CALLCODE:
 		accessMemory(arg[3], arg[4]);
@@ -340,7 +346,6 @@ u256 EVMInstructionInterpreter::eval(
 		accessMemory(arg[2], arg[3]);
 		accessMemory(arg[4], arg[5]);
 		logTrace(_instruction, arg);
-
 		// Randomly fail based on the called address if it isn't a call to self.
 		// Used for fuzzing.
 		return (
@@ -359,7 +364,6 @@ u256 EVMInstructionInterpreter::eval(
 		accessMemory(arg[0], arg[1]);
 		logTrace(_instruction, arg);
 		m_state.storage.clear();
-		m_state.trace.clear();
 		BOOST_THROW_EXCEPTION(ExplicitlyTerminated());
 	case Instruction::INVALID:
 		logTrace(_instruction);
@@ -479,7 +483,10 @@ u256 EVMInstructionInterpreter::evalBuiltin(
 	else if (fun == "datacopy")
 	{
 		// This is identical to codecopy.
-		if (accessMemory(_evaluatedArguments.at(0), _evaluatedArguments.at(2)))
+		if (
+				_evaluatedArguments.at(2) != 0 &&
+				accessMemory(_evaluatedArguments.at(0), _evaluatedArguments.at(2))
+		)
 			copyZeroExtended(
 				m_state.memory,
 				m_state.code,
@@ -560,8 +567,13 @@ void EVMInstructionInterpreter::logTrace(
 	if (!(_writesToMemory && memWriteTracingDisabled()))
 	{
 		string message = _pseudoInstruction + "(";
+		std::pair<bool, size_t> inputMemoryPtrModified = isInputMemoryPtrModified(_pseudoInstruction, _arguments);
 		for (size_t i = 0; i < _arguments.size(); ++i)
-			message += (i > 0 ? ", " : "") + formatNumber(_arguments[i]);
+		{
+			bool printZero = inputMemoryPtrModified.first && inputMemoryPtrModified.second == i;
+			u256 arg = printZero ? 0 : _arguments[i];
+			message += (i > 0 ? ", " : "") + formatNumber(arg);
+		}
 		message += ")";
 		if (!_data.empty())
 			message += " [" + util::toHex(_data) + "]";
@@ -573,3 +585,65 @@ void EVMInstructionInterpreter::logTrace(
 		}
 	}
 }
+
+std::pair<bool, size_t> EVMInstructionInterpreter::isInputMemoryPtrModified(
+	std::string const& _pseudoInstruction,
+	std::vector<u256> const& _arguments
+)
+{
+	if (_pseudoInstruction == "RETURN" || _pseudoInstruction == "REVERT")
+	{
+		if (_arguments[1] == 0)
+			return {true, 0};
+		else
+			return {false, 0};
+	}
+	else if (
+		_pseudoInstruction == "RETURNDATACOPY" || _pseudoInstruction == "CALLDATACOPY"
+		|| _pseudoInstruction == "CODECOPY")
+	{
+		if (_arguments[2] == 0)
+			return {true, 0};
+		else
+			return {false, 0};
+	}
+	else if (_pseudoInstruction == "EXTCODECOPY")
+	{
+		if (_arguments[3] == 0)
+			return {true, 1};
+		else
+			return {false, 0};
+	}
+	else if (
+		_pseudoInstruction == "LOG0" || _pseudoInstruction == "LOG1" || _pseudoInstruction == "LOG2"
+		|| _pseudoInstruction == "LOG3" || _pseudoInstruction == "LOG4")
+	{
+		if (_arguments[1] == 0)
+			return {true, 0};
+		else
+			return {false, 0};
+	}
+	if (_pseudoInstruction == "CREATE" || _pseudoInstruction == "CREATE2")
+	{
+		if (_arguments[2] == 0)
+			return {true, 1};
+		else
+			return {false, 0};
+	}
+	if (_pseudoInstruction == "CALL" || _pseudoInstruction == "CALLCODE")
+	{
+		if (_arguments[4] == 0)
+			return {true, 3};
+		else
+			return {false, 0};
+	}
+	else if (_pseudoInstruction == "DELEGATECALL" || _pseudoInstruction == "STATICCALL")
+	{
+		if (_arguments[3] == 0)
+			return {true, 2};
+		else
+			return {false, 0};
+	}
+	else
+		return {false, 0};
+}

test/tools/yulInterpreter/EVMInstructionInterpreter.h

@@ -125,6 +125,19 @@ private:
 		std::vector<u256> const& _arguments = {},
 		bytes const& _data = {}
 	);
+
+	/// @returns a pair of boolean and size_t whose first value is true if @param _pseudoInstruction
+	/// is a Yul instruction that the Yul optimizer's loadResolver step rewrites the input
+	/// memory pointer value to zero if that instruction's read length (contained within @param
+	// _arguments) is zero, and whose second value is the positional index of the input memory
+	// pointer argument.
+	/// If the Yul instruction is unaffected or affected but read length is non-zero, the first
+	/// value is false.
+	std::pair<bool, size_t> isInputMemoryPtrModified(
+		std::string const& _pseudoInstruction,
+		std::vector<u256> const& _arguments
+	);
+
 	/// @returns disable trace flag.
 	bool memWriteTracingDisabled()
 	{