Add SWAPN/DUPN.

Author: ekpyron

libevmasm/Assembly.cpp

@@ -767,11 +767,21 @@ AssemblyItem Assembly::newImmutableAssignment(std::string const& _identifier)
 	return AssemblyItem{AssignImmutable, h};
 }
 
-AssemblyItem Assembly::newAuxDataLoadN(size_t _offset)
+AssemblyItem Assembly::newAuxDataLoadN(size_t _offset) const
 {
 	return AssemblyItem{AuxDataLoadN, _offset};
 }
 
+AssemblyItem Assembly::newSwapN(size_t _depth) const
+{
+	return AssemblyItem::swapN(_depth);
+}
+
+AssemblyItem Assembly::newDupN(size_t _depth) const
+{
+	return AssemblyItem::dupN(_depth);
+}
+
 Assembly& Assembly::optimise(OptimiserSettings const& _settings)
 {
 	optimiseInternal(_settings, {});
@@ -1558,7 +1568,9 @@ LinkerObject const& Assembly::assembleEOF() const
 					item.instruction() != Instruction::RJUMPI &&
 					item.instruction() != Instruction::CALLF &&
 					item.instruction() != Instruction::JUMPF &&
-					item.instruction() != Instruction::RETF
+					item.instruction() != Instruction::RETF &&
+					item.instruction() != Instruction::DUPN &&
+					item.instruction() != Instruction::SWAPN
 				);
 				solAssert(!(item.instruction() >= Instruction::PUSH0 && item.instruction() <= Instruction::PUSH32));
 				ret.bytecode += assembleOperation(item);
@@ -1636,6 +1648,12 @@ LinkerObject const& Assembly::assembleEOF() const
 			case RetF:
 				ret.bytecode.push_back(static_cast<uint8_t>(Instruction::RETF));
 				break;
+			case SwapN:
+			case DupN:
+				ret.bytecode.push_back(static_cast<uint8_t>(item.instruction()));
+				solAssert(item.data() >= 1 && item.data() <= 256);
+				ret.bytecode.push_back(static_cast<uint8_t>(item.data() - 1));
+				break;
 			default:
 				solAssert(false, "Unexpected opcode while assembling.");
 			}

libevmasm/Assembly.h

@@ -88,7 +88,9 @@ class Assembly
 	AssemblyItem newPushLibraryAddress(std::string const& _identifier);
 	AssemblyItem newPushImmutable(std::string const& _identifier);
 	AssemblyItem newImmutableAssignment(std::string const& _identifier);
-	AssemblyItem newAuxDataLoadN(size_t offset);
+	AssemblyItem newAuxDataLoadN(size_t offset) const;
+	AssemblyItem newSwapN(size_t _depth) const;
+	AssemblyItem newDupN(size_t _depth) const;
 
 	AssemblyItem const& append(AssemblyItem _i);
 	AssemblyItem const& append(bytes const& _data) { return append(newData(_data)); }
@@ -102,6 +104,8 @@ class Assembly
 	void appendImmutable(std::string const& _identifier) { append(newPushImmutable(_identifier)); }
 	void appendImmutableAssignment(std::string const& _identifier) { append(newImmutableAssignment(_identifier)); }
 	void appendAuxDataLoadN(uint16_t _offset) { append(newAuxDataLoadN(_offset));}
+	void appendSwapN(size_t _depth) { append(newSwapN(_depth)); }
+	void appendDupN(size_t _depth) { append(newDupN(_depth)); }
 
 	void appendVerbatim(bytes _data, size_t _arguments, size_t _returnVariables)
 	{

libevmasm/AssemblyItem.cpp

@@ -19,6 +19,7 @@
 #include <libevmasm/AssemblyItem.h>
 
 #include <libevmasm/Assembly.h>
+#include <libevmasm/SemanticInformation.h>
 
 #include <libsolutil/CommonData.h>
 #include <libsolutil/CommonIO.h>
@@ -90,6 +91,9 @@ std::pair<std::string, std::string> AssemblyItem::nameAndData(langutil::EVMVersi
 	case JumpF:
 	case RetF:
 		return {instructionInfo(instruction(), _evmVersion).name, ""};
+	case SwapN:
+	case DupN:
+		return {instructionInfo(instruction(), _evmVersion).name, util::toString(static_cast<size_t>(data())) };
 	case Push:
 		return {"PUSH", toStringInHex(data())};
 	case PushTag:
@@ -192,6 +196,10 @@ size_t AssemblyItem::bytesRequired(size_t _addressLength, langutil::EVMVersion _
 		return 2;
 	case ReturnContract:
 		return 2;
+	case SwapN:
+		return 2;
+	case DupN:
+		return 2;
 	case UndefinedItem:
 		solAssert(false);
 	}
@@ -203,6 +211,10 @@ size_t AssemblyItem::arguments() const
 {
 	if (type() == CallF || type() == JumpF)
 		return functionSignature().argsNum;
+	else if (type() == SwapN)
+		return static_cast<size_t>(data()) + 1;
+	else if (type() == DupN)
+		return static_cast<size_t>(data());
 	else if (hasInstruction())
 	{
 		solAssert(instruction() != Instruction::CALLF && instruction() != Instruction::JUMPF);
@@ -231,6 +243,9 @@ size_t AssemblyItem::returnValues() const
 		// The latest EVMVersion is used here, since the InstructionInfo is assumed to be
 		// the same across all EVM versions except for the instruction name.
 		return static_cast<size_t>(instructionInfo(instruction(), EVMVersion()).ret);
+	case SwapN:
+	case DupN:
+		return static_cast<size_t>(data()) + 1;
 	case Push:
 	case PushTag:
 	case PushData:
@@ -270,8 +285,10 @@ bool AssemblyItem::canBeFunctional() const
 	case ConditionalRelativeJump:
 	case CallF:
 	case JumpF:
+	case SwapN:
+	case DupN:
 	case RetF:
-		return !isDupInstruction(instruction()) && !isSwapInstruction(instruction());
+		return !SemanticInformation::isDupInstruction(*this) && !SemanticInformation::isSwapInstruction(*this);
 	case Push:
 	case PushTag:
 	case PushData:
@@ -410,6 +427,12 @@ std::string AssemblyItem::toAssemblyText(Assembly const& _assembly) const
 	case RetF:
 		text = "retf";
 		break;
+	case SwapN:
+		text = "swapn{" + std::to_string(static_cast<size_t>(data())) + "}";
+		break;
+	case DupN:
+		text = "dupn{" + std::to_string(static_cast<size_t>(data())) + "}";
+		break;
 	}
 	if (m_jumpType == JumpType::IntoFunction || m_jumpType == JumpType::OutOfFunction)
 	{
@@ -435,6 +458,8 @@ std::ostream& solidity::evmasm::operator<<(std::ostream& _out, AssemblyItem cons
 	case CallF:
 	case JumpF:
 	case RetF:
+	case SwapN:
+	case DupN:
 		_out << " " << instructionInfo(_item.instruction(), EVMVersion()).name;
 		if (_item.instruction() == Instruction::JUMP || _item.instruction() == Instruction::JUMPI)
 			_out << "\t" << _item.getJumpTypeAsString();

libevmasm/AssemblyItem.h

@@ -62,7 +62,9 @@ enum AssemblyItemType
 	CallF, ///< Jumps to a returning EOF function, adding a new frame to the return stack.
 	JumpF, ///< Jumps to a returning or non-returning EOF function without changing the return stack.
 	RetF, ///< Returns from an EOF function, removing a frame from the return stack.
-	VerbatimBytecode ///< Contains data that is inserted into the bytecode code section without modification.
+	VerbatimBytecode, ///< Contains data that is inserted into the bytecode code section without modification.
+	SwapN, ///< EOF SWAPN with immediate argument.
+	DupN, ///< EOF DUPN with immediate argument.
 };
 
 enum class Precision { Precise , Approximate };
@@ -147,6 +149,16 @@ class AssemblyItem
 		solAssert(_tag.type() == Tag);
 		return AssemblyItem(ConditionalRelativeJump, Instruction::RJUMPI, _tag.data(), _debugData);
 	}
+	static AssemblyItem swapN(size_t _depth, langutil::DebugData::ConstPtr _debugData = langutil::DebugData::create())
+	{
+		solAssert(_depth >= 1 && _depth <= 256);
+		return AssemblyItem(SwapN, Instruction::SWAPN, _depth, _debugData);
+	}
+	static AssemblyItem dupN(size_t _depth, langutil::DebugData::ConstPtr _debugData = langutil::DebugData::create())
+	{
+		solAssert(_depth >= 1 && _depth <= 256);
+		return AssemblyItem(DupN, Instruction::DUPN, _depth, _debugData);
+	}
 
 	AssemblyItem(AssemblyItem const&) = default;
 	AssemblyItem(AssemblyItem&&) = default;
@@ -191,7 +203,9 @@ class AssemblyItem
 			m_type == ConditionalRelativeJump ||
 			m_type == CallF ||
 			m_type == JumpF ||
-			m_type == RetF;
+			m_type == RetF ||
+			m_type == SwapN ||
+			m_type == DupN;
 	}
 	/// @returns the instruction of this item (only valid if hasInstruction returns true)
 	Instruction instruction() const

libevmasm/Instruction.cpp

@@ -172,6 +172,8 @@ std::map<std::string, Instruction> const solidity::evmasm::c_instructions =
 	{ "CALLF", Instruction::CALLF },
 	{ "RETF", Instruction::RETF },
 	{ "JUMPF", Instruction::JUMPF },
+	{ "DUPN", Instruction::DUPN },
+	{ "SWAPN", Instruction::SWAPN },
 	{ "RJUMP", Instruction::RJUMP },
 	{ "RJUMPI", Instruction::RJUMPI },
 	{ "EOFCREATE", Instruction::EOFCREATE },
@@ -339,6 +341,8 @@ static std::map<Instruction, InstructionInfo> const c_instructionInfo =
 	{Instruction::RETF,           {"RETF",            0,  0,   0,  true,       Tier::RetF}},
 	{Instruction::CALLF,          {"CALLF",           2,  0,   0,  true,       Tier::CallF}},
 	{Instruction::JUMPF,          {"JUMPF",           2,  0,   0,  true,       Tier::JumpF}},
+	{Instruction::SWAPN,          {"SWAPN",           1,  0,   0,  false,      Tier::VeryLow}},
+	{Instruction::DUPN,           {"DUPN",            1,  0,   0,  false,      Tier::VeryLow}},
 	{Instruction::EOFCREATE,      {"EOFCREATE",       1,  4,   1,  true,       Tier::Special}},
 	{Instruction::RETURNCONTRACT, {"RETURNCONTRACT",  1,  2,   0,  true,       Tier::Special}},
 	{Instruction::CREATE,         {"CREATE",          0,  3,   1,  true,       Tier::Special}},

libevmasm/Instruction.h

@@ -190,6 +190,8 @@ enum class Instruction: uint8_t
 	CALLF = 0xe3,             ///< call function in a EOF code section
 	RETF = 0xe4,              ///< return to caller from the code section of EOF container
 	JUMPF = 0xe5,             ///< jump to a code section of EOF container without adding a new return stack frame.
+	DUPN = 0xe6,              ///< copies a value at the stack depth given as immediate argument to the top of the stack
+	SWAPN = 0xe7,             ///< swaps the highest value with a value at a stack depth given as immediate argument
 	EOFCREATE = 0xec,         ///< create a new account with associated container code.
 	RETURNCONTRACT = 0xee,    ///< return container to be deployed with axiliary data filled in.
 	CREATE = 0xf0,            ///< create a new account with associated code
@@ -232,18 +234,6 @@ inline bool isPushInstruction(Instruction _inst)
 	return Instruction::PUSH0 <= _inst && _inst <= Instruction::PUSH32;
 }
 
-/// @returns true if the instruction is a DUP
-inline bool isDupInstruction(Instruction _inst)
-{
-	return Instruction::DUP1 <= _inst && _inst <= Instruction::DUP16;
-}
-
-/// @returns true if the instruction is a SWAP
-inline bool isSwapInstruction(Instruction _inst)
-{
-	return Instruction::SWAP1 <= _inst && _inst <= Instruction::SWAP16;
-}
-
 /// @returns true if the instruction is a LOG
 inline bool isLogInstruction(Instruction _inst)
 {
@@ -256,18 +246,6 @@ inline unsigned getPushNumber(Instruction _inst)
 	return static_cast<uint8_t>(_inst) - unsigned(Instruction::PUSH0);
 }
 
-/// @returns the number of DUP Instruction _inst
-inline unsigned getDupNumber(Instruction _inst)
-{
-	return static_cast<uint8_t>(_inst) - unsigned(Instruction::DUP1) + 1;
-}
-
-/// @returns the number of SWAP Instruction _inst
-inline unsigned getSwapNumber(Instruction _inst)
-{
-	return static_cast<uint8_t>(_inst) - unsigned(Instruction::SWAP1) + 1;
-}
-
 /// @returns the number of LOG Instruction _inst
 inline unsigned getLogNumber(Instruction _inst)
 {

libevmasm/KnownState.cpp

@@ -137,14 +137,14 @@ KnownState::StoreOperation KnownState::feedItem(AssemblyItem const& _item, bool
 			setStackElement(
 				m_stackHeight + 1,
 				stackElement(
-					m_stackHeight - static_cast<int>(instruction) + static_cast<int>(Instruction::DUP1),
+					m_stackHeight - (static_cast<int>(SemanticInformation::getDupNumber(_item)) - 1),
 					_item.debugData()
 				)
 			);
 		else if (SemanticInformation::isSwapInstruction(_item))
 			swapStackElements(
 				m_stackHeight,
-				m_stackHeight - 1 - static_cast<int>(instruction) + static_cast<int>(Instruction::SWAP1),
+				m_stackHeight - 1 - (static_cast<int>(SemanticInformation::getSwapNumber(_item)) - 1),
 				_item.debugData()
 			);
 		else if (instruction != Instruction::POP)

libevmasm/PeepholeOptimiser.cpp

@@ -290,7 +290,7 @@ struct DupSwap: SimplePeepholeOptimizerMethod<DupSwap>
 		if (
 			SemanticInformation::isDupInstruction(_dupN) &&
 			SemanticInformation::isSwapInstruction(_swapN) &&
-			getDupNumber(_dupN.instruction()) == getSwapNumber(_swapN.instruction())
+			SemanticInformation::getDupNumber(_dupN) == SemanticInformation::getSwapNumber(_swapN)
 		)
 		{
 			*_out = _dupN;

libevmasm/SemanticInformation.cpp

@@ -312,16 +312,22 @@ bool SemanticInformation::isCommutativeOperation(AssemblyItem const& _item)
 
 bool SemanticInformation::isDupInstruction(AssemblyItem const& _item)
 {
+	if (_item.type() == evmasm::DupN)
+		return true;
 	if (_item.type() != evmasm::Operation)
 		return false;
-	return evmasm::isDupInstruction(_item.instruction());
+	auto inst = _item.instruction();
+	return Instruction::DUP1 <= inst && inst <= Instruction::DUP16;
 }
 
 bool SemanticInformation::isSwapInstruction(AssemblyItem const& _item)
 {
+	if (_item.type() == evmasm::SwapN)
+		return true;
 	if (_item.type() != evmasm::Operation)
 		return false;
-	return evmasm::isSwapInstruction(_item.instruction());
+	auto inst = _item.instruction();
+	return Instruction::SWAP1 <= inst && inst <= Instruction::SWAP16;
 }
 
 bool SemanticInformation::altersControlFlow(AssemblyItem const& _item)
@@ -387,6 +393,24 @@ bool SemanticInformation::reverts(Instruction _instruction)
 	}
 }
 
+size_t SemanticInformation::getDupNumber(AssemblyItem const& _item)
+{
+	assertThrow(isDupInstruction(_item), OptimizerException, "Not a DUP instruction.");
+	if (_item.type() == evmasm::DupN)
+		return static_cast<size_t>(_item.data());
+	auto inst = _item.instruction();
+	return static_cast<uint8_t>(inst) - static_cast<size_t>(Instruction::DUP1) + 1;
+}
+
+size_t SemanticInformation::getSwapNumber(AssemblyItem const& _item)
+{
+	assertThrow(isSwapInstruction(_item), OptimizerException, "Not a swap instruction.");
+	if (_item.type() == evmasm::SwapN)
+		return static_cast<size_t>(_item.data());
+	auto inst = _item.instruction();
+	return static_cast<uint8_t>(inst) - static_cast<size_t>(Instruction::SWAP1) + 1;
+}
+
 bool SemanticInformation::isDeterministic(AssemblyItem const& _item)
 {
 	assertThrow(_item.type() != VerbatimBytecode, AssemblyException, "");

libevmasm/SemanticInformation.h

@@ -86,6 +86,10 @@ struct SemanticInformation
 	static bool terminatesControlFlow(AssemblyItem const& _item);
 	static bool terminatesControlFlow(Instruction _instruction);
 	static bool reverts(Instruction _instruction);
+	/// @returns the 1-based DUP depth a DUP AssembleItem @a _item
+	static size_t getDupNumber(AssemblyItem const& _item);
+	/// @returns the 1-based SWAP depth a DUP AssembleItem @a _item
+	static size_t getSwapNumber(AssemblyItem const& _item);
 	/// @returns false if the value put on the stack by _item depends on anything else than
 	/// the information in the current block header, memory, storage, transient storage or stack.
 	static bool isDeterministic(AssemblyItem const& _item);

liblangutil/EVMVersion.cpp

@@ -84,6 +84,8 @@ bool EVMVersion::hasOpcode(Instruction _opcode, std::optional<uint8_t> _eofVersi
 	case Instruction::RJUMPI:
 	case Instruction::CALLF:
 	case Instruction::JUMPF:
+	case Instruction::DUPN:
+	case Instruction::SWAPN:
 	case Instruction::RETF:
 	case Instruction::EXTCALL:
 	case Instruction::EXTSTATICCALL:

libyul/AsmAnalysis.cpp

@@ -741,7 +741,9 @@ bool AsmAnalyzer::validateInstructions(evmasm::Instruction _instr, SourceLocatio
 		_instr != evmasm::Instruction::RJUMPI &&
 		_instr != evmasm::Instruction::CALLF &&
 		_instr != evmasm::Instruction::JUMPF &&
-		_instr != evmasm::Instruction::RETF
+		_instr != evmasm::Instruction::RETF &&
+		_instr != evmasm::Instruction::DUPN &&
+		_instr != evmasm::Instruction::SWAPN
 	);
 
 	auto errorForVM = [&](ErrorId _errorId, std::string const& vmKindMessage) {

libyul/backends/evm/AbstractAssembly.h

@@ -149,6 +149,13 @@ class AbstractAssembly
 	/// EOF auxiliary data in data section and the auxiliary data are different things.
 	virtual void appendToAuxiliaryData(bytes const& _data) = 0;
 
+	/// Appends a SWAPN with 1-based indexing for @arg _depth. I.e. a depth of 1 would be equivalent to emitting SWAP1.
+	/// @arg _depth ranges from 1 to 256.
+	virtual void appendSwapN(size_t _depth) = 0;
+	/// Appends a DUPN with 1-based indexing for @arg _depth. I.e. a depth of 1 would be equivalent to emitting DUP1.
+	/// @arg _depth ranges from 1 to 256.
+	virtual void appendDupN(size_t _depth) = 0;
+
 	/// Mark this assembly as invalid. Any attempt to request bytecode from it should throw.
 	virtual void markAsInvalid() = 0;
 

libyul/backends/evm/EVMDialect.cpp

@@ -211,8 +211,8 @@ std::vector<std::optional<BuiltinFunctionForEVM>> createBuiltins(langutil::EVMVe
 		auto const opcode = instr.second;
 
 		if (
-			!evmasm::isDupInstruction(opcode) &&
-			!evmasm::isSwapInstruction(opcode) &&
+			!(opcode >= evmasm::Instruction::DUP1 && opcode <= evmasm::Instruction::DUP16) &&
+			!(opcode >= evmasm::Instruction::SWAP1 && opcode <= evmasm::Instruction::SWAP16) &&
 			!evmasm::isPushInstruction(opcode) &&
 			opcode != evmasm::Instruction::JUMP &&
 			opcode != evmasm::Instruction::JUMPI &&
@@ -224,6 +224,8 @@ std::vector<std::optional<BuiltinFunctionForEVM>> createBuiltins(langutil::EVMVe
 			opcode != evmasm::Instruction::RJUMPI &&
 			opcode != evmasm::Instruction::CALLF &&
 			opcode != evmasm::Instruction::JUMPF &&
+			opcode != evmasm::Instruction::DUPN &&
+			opcode != evmasm::Instruction::SWAPN &&
 			opcode != evmasm::Instruction::RETF &&
 			_evmVersion.hasOpcode(opcode, _eofVersion) &&
 			!prevRandaoException(name)