Merge pull request ethereum#13759 from ethereum/prevrandao

Add `block.prevrandao` as alias for `block.difficulty`

Changelog.md

@@ -7,7 +7,9 @@ Language Features:
 Compiler Features:
  * Commandline Interface: Return exit code ``2`` on uncaught exceptions.
  * Commandline Interface: Add `--no-cbor-metadata` that skips CBOR metadata from getting appended at the end of the bytecode.
- * EVM: Basic support for the EVM version "Paris".
+ * EVM: Deprecate ``block.difficulty`` and disallow ``difficulty()`` in inline assembly for EVM versions >= paris. The change is due to the renaming introduced by [EIP-4399](https://eips.ethereum.org/EIPS/eip-4399).
+ * EVM: Introduce ``block.prevrandao`` in Solidity and ``prevrandao()`` in inline assembly for EVM versions >= paris.
+ * EVM: Support for the EVM version "Paris".
  * Natspec: Add event Natspec inheritance for devdoc.
  * Standard JSON: Add a boolean field `settings.metadata.appendCBOR` that skips CBOR metadata from getting appended at the end of the bytecode.
  * Yul EVM Code Transform: Generate more optimal code for user-defined functions that always terminate a transaction. No return labels will be pushed for calls to functions that always terminate.

docs/cheatsheet.rst

@@ -8,7 +8,7 @@ Order of Precedence of Operators
 ================================
 .. include:: types/operator-precedence-table.rst
 
-.. index:: assert, block, coinbase, difficulty, number, block;number, timestamp, block;timestamp, msg, data, gas, sender, value, gas price, origin, revert, require, keccak256, ripemd160, sha256, ecrecover, addmod, mulmod, cryptography, this, super, selfdestruct, balance, codehash, send
+.. index:: assert, block, coinbase, difficulty, prevrandao, number, block;number, timestamp, block;timestamp, msg, data, gas, sender, value, gas price, origin, revert, require, keccak256, ripemd160, sha256, ecrecover, addmod, mulmod, cryptography, this, super, selfdestruct, balance, codehash, send
 
 Global Variables
 ================
@@ -32,9 +32,10 @@ Global Variables
 - ``block.basefee`` (``uint``): current block's base fee (`EIP-3198 <https://eips.ethereum.org/EIPS/eip-3198>`_ and `EIP-1559 <https://eips.ethereum.org/EIPS/eip-1559>`_)
 - ``block.chainid`` (``uint``): current chain id
 - ``block.coinbase`` (``address payable``): current block miner's address
-- ``block.difficulty`` (``uint``): current block difficulty
+- ``block.difficulty`` (``uint``): current block difficulty (``EVM < Paris``). For other EVM versions it behaves as a deprecated alias for ``block.prevrandao`` that will be removed in the next breaking release
 - ``block.gaslimit`` (``uint``): current block gaslimit
 - ``block.number`` (``uint``): current block number
+- ``block.prevrandao`` (``uint``): random number provided by the beacon chain (``EVM >= Paris``) (see `EIP-4399 <https://eips.ethereum.org/EIPS/eip-4399>`_ )
 - ``block.timestamp`` (``uint``): current block timestamp in seconds since Unix epoch
 - ``gasleft() returns (uint256)``: remaining gas
 - ``msg.data`` (``bytes``): complete calldata

docs/grammar/SolidityLexer.g4

@@ -291,8 +291,8 @@ YulEVMBuiltin:
 	| 'returndatacopy' | 'extcodehash' | 'create' | 'create2' | 'call' | 'callcode'
 	| 'delegatecall' | 'staticcall' | 'return' | 'revert' | 'selfdestruct' | 'invalid'
 	| 'log0' | 'log1' | 'log2' | 'log3' | 'log4' | 'chainid' | 'origin' | 'gasprice'
-	| 'blockhash' | 'coinbase' | 'timestamp' | 'number' | 'difficulty' | 'gaslimit'
-	| 'basefee';
+	| 'blockhash' | 'coinbase' | 'timestamp' | 'number' | 'difficulty' | 'prevrandao'
+	| 'gaslimit' | 'basefee';
 
 YulLBrace: '{' -> pushMode(YulMode);
 YulRBrace: '}' -> popMode;

docs/units-and-global-variables.rst

@@ -65,7 +65,7 @@ There are special variables and functions which always exist in the global
 namespace and are mainly used to provide information about the blockchain
 or are general-use utility functions.
 
-.. index:: abi, block, coinbase, difficulty, encode, number, block;number, timestamp, block;timestamp, msg, data, gas, sender, value, gas price, origin
+.. index:: abi, block, coinbase, difficulty, prevrandao, encode, number, block;number, timestamp, block;timestamp, msg, data, gas, sender, value, gas price, origin
 
 
 Block and Transaction Properties
@@ -75,9 +75,10 @@ Block and Transaction Properties
 - ``block.basefee`` (``uint``): current block's base fee (`EIP-3198 <https://eips.ethereum.org/EIPS/eip-3198>`_ and `EIP-1559 <https://eips.ethereum.org/EIPS/eip-1559>`_)
 - ``block.chainid`` (``uint``): current chain id
 - ``block.coinbase`` (``address payable``): current block miner's address
-- ``block.difficulty`` (``uint``): current block difficulty
+- ``block.difficulty`` (``uint``): current block difficulty (``EVM < Paris``). For other EVM versions it behaves as a deprecated alias for ``block.prevrandao`` (`EIP-4399 <https://eips.ethereum.org/EIPS/eip-4399>`_ )
 - ``block.gaslimit`` (``uint``): current block gaslimit
 - ``block.number`` (``uint``): current block number
+- ``block.prevrandao`` (``uint``): random number provided by the beacon chain (``EVM >= Paris``)
 - ``block.timestamp`` (``uint``): current block timestamp as seconds since unix epoch
 - ``gasleft() returns (uint256)``: remaining gas
 - ``msg.data`` (``bytes calldata``): complete calldata
@@ -394,4 +395,4 @@ These keywords are reserved in Solidity. They might become part of the syntax in
 ``define``, ``final``, ``implements``, ``in``, ``inline``, ``let``, ``macro``, ``match``,
 ``mutable``, ``null``, ``of``, ``partial``, ``promise``, ``reference``, ``relocatable``,
 ``sealed``, ``sizeof``, ``static``, ``supports``, ``switch``, ``typedef``, ``typeof``,
-``var``.
+``var``.

docs/yul.rst

@@ -751,8 +751,8 @@ This document does not want to be a full description of the Ethereum virtual mac
 Please refer to a different document if you are interested in the precise semantics.
 
 Opcodes marked with ``-`` do not return a result and all others return exactly one value.
-Opcodes marked with ``F``, ``H``, ``B``, ``C``, ``I`` and ``L`` are present since Frontier, Homestead,
-Byzantium, Constantinople, Istanbul or London respectively.
+Opcodes marked with ``F``, ``H``, ``B``, ``C``, ``I``, ``L`` and ``P`` are present since Frontier,
+Homestead, Byzantium, Constantinople, Istanbul, London or Paris respectively.
 
 In the following, ``mem[a...b)`` signifies the bytes of memory starting at position ``a`` up to
 but not including position ``b`` and ``storage[p]`` signifies the storage contents at slot ``p``.
@@ -931,6 +931,8 @@ the ``dup`` and ``swap`` instructions as well as ``jump`` instructions, labels a
 +-------------------------+-----+---+-----------------------------------------------------------------+
 | difficulty()            |     | F | difficulty of the current block (see note below)                |
 +-------------------------+-----+---+-----------------------------------------------------------------+
+| prevrandao()            |     | P | randomness provided by the beacon chain (see note below)        |
++-------------------------+-----+---+-----------------------------------------------------------------+
 | gaslimit()              |     | F | block gas limit of the current block                            |
 +-------------------------+-----+---+-----------------------------------------------------------------+
 
@@ -945,8 +947,10 @@ the ``dup`` and ``swap`` instructions as well as ``jump`` instructions, labels a
   The remaining bytes will retain their values as of before the call.
 
 .. note::
-  With the Paris network upgrade the semantics of ``difficulty`` have been changed.
-  It returns the value of ``prevrandao``, which is a 256-bit value, whereas the highest recorded
+  The `difficulty()` instruction is disallowed in EVM version >= Paris.
+  With the Paris network upgrade the semantics of the instruction that was previously called
+  ``difficulty`` have been changed and the instruction was renamed to ``prevrandao``.
+  It can now return arbitrary values in the full 256-bit range, whereas the highest recorded
   difficulty value within Ethash was ~54 bits.
   This change is described in `EIP-4399 <https://eips.ethereum.org/EIPS/eip-4399>`_.
   Please note that irrelevant to which EVM version is selected in the compiler, the semantics of

libevmasm/Assembly.cpp

@@ -237,7 +237,7 @@ Json::Value Assembly::assemblyJSON(map<string, unsigned> const& _sourceIndices,
 				sourceIndex = static_cast<int>(iter->second);
 		}
 
-		auto [name, data] = item.nameAndData();
+		auto [name, data] = item.nameAndData(m_evmVersion);
 		Json::Value jsonItem;
 		jsonItem["name"] = name;
 		jsonItem["begin"] = item.location().start;

libevmasm/Assembly.h

@@ -49,7 +49,7 @@ using AssemblyPointer = std::shared_ptr<Assembly>;
 class Assembly
 {
 public:
-	Assembly(bool _creation, std::string _name): m_creation(_creation), m_name(std::move(_name)) { }
+	Assembly(langutil::EVMVersion _evmVersion, bool _creation, std::string _name): m_evmVersion(_evmVersion), m_creation(_creation), m_name(std::move(_name)) { }
 
 	AssemblyItem newTag() { assertThrow(m_usedTags < 0xffffffff, AssemblyException, ""); return AssemblyItem(Tag, m_usedTags++); }
 	AssemblyItem newPushTag() { assertThrow(m_usedTags < 0xffffffff, AssemblyException, ""); return AssemblyItem(PushTag, m_usedTags++); }
@@ -112,6 +112,7 @@ class Assembly
 	/// Changes the source location used for each appended item.
 	void setSourceLocation(langutil::SourceLocation const& _location) { m_currentSourceLocation = _location; }
 	langutil::SourceLocation const& currentSourceLocation() const { return m_currentSourceLocation; }
+	langutil::EVMVersion const& evmVersion() const { return m_evmVersion; }
 
 	/// Assembles the assembly into bytecode. The assembly should not be modified after this call, since the assembled version is cached.
 	LinkerObject const& assemble() const;
@@ -208,6 +209,8 @@ class Assembly
 	mutable LinkerObject m_assembledObject;
 	mutable std::vector<size_t> m_tagPositionsInBytecode;
 
+	langutil::EVMVersion m_evmVersion;
+
 	int m_deposit = 0;
 	/// True, if the assembly contains contract creation code.
 	bool const m_creation = false;

libevmasm/AssemblyItem.cpp

@@ -69,12 +69,12 @@ pair<size_t, size_t> AssemblyItem::splitForeignPushTag() const
 	return make_pair(subId, tag);
 }
 
-pair<string, string> AssemblyItem::nameAndData() const
+pair<string, string> AssemblyItem::nameAndData(langutil::EVMVersion _evmVersion) const
 {
 	switch (type())
 	{
 	case Operation:
-		return {instructionInfo(instruction()).name, m_data != nullptr ? toStringInHex(*m_data) : ""};
+		return {instructionInfo(instruction(), _evmVersion).name, m_data != nullptr ? toStringInHex(*m_data) : ""};
 	case Push:
 		return {"PUSH", toStringInHex(data())};
 	case PushTag:
@@ -168,7 +168,9 @@ size_t AssemblyItem::bytesRequired(size_t _addressLength, Precision _precision)
 size_t AssemblyItem::arguments() const
 {
 	if (type() == Operation)
-		return static_cast<size_t>(instructionInfo(instruction()).args);
+		// 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()).args);
 	else if (type() == VerbatimBytecode)
 		return get<0>(*m_verbatimBytecode);
 	else if (type() == AssignImmutable)
@@ -182,7 +184,9 @@ size_t AssemblyItem::returnValues() const
 	switch (m_type)
 	{
 	case Operation:
-		return static_cast<size_t>(instructionInfo(instruction()).ret);
+		// 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 Push:
 	case PushTag:
 	case PushData:
@@ -251,7 +255,7 @@ string AssemblyItem::toAssemblyText(Assembly const& _assembly) const
 	case Operation:
 	{
 		assertThrow(isValidInstruction(instruction()), AssemblyException, "Invalid instruction.");
-		text = util::toLower(instructionInfo(instruction()).name);
+		text = util::toLower(instructionInfo(instruction(), _assembly.evmVersion()).name);
 		break;
 	}
 	case Push:
@@ -323,12 +327,13 @@ string AssemblyItem::toAssemblyText(Assembly const& _assembly) const
 	return text;
 }
 
+// Note: This method is exclusively used for debugging.
 ostream& solidity::evmasm::operator<<(ostream& _out, AssemblyItem const& _item)
 {
 	switch (_item.type())
 	{
 	case Operation:
-		_out << " " << instructionInfo(_item.instruction()).name;
+		_out << " " << instructionInfo(_item.instruction(), EVMVersion()).name;
 		if (_item.instruction() == Instruction::JUMP || _item.instruction() == Instruction::JUMPI)
 			_out << "\t" << _item.getJumpTypeAsString();
 		break;

libevmasm/AssemblyItem.h

@@ -108,10 +108,11 @@ class AssemblyItem
 
 	/// This function is used in `Assembly::assemblyJSON`.
 	/// It returns the name & data of the current assembly item.
+	/// @param _evmVersion the EVM version.
 	/// @returns a pair, where the first element is the json-assembly
 	/// item name, where second element is the string representation
 	/// of it's data.
-	std::pair<std::string, std::string> nameAndData() const;
+	std::pair<std::string, std::string> nameAndData(langutil::EVMVersion _evmVersion) const;
 
 	bytes const& verbatimData() const { assertThrow(m_type == VerbatimBytecode, util::Exception, ""); return std::get<2>(*m_verbatimBytecode); }
 

libevmasm/CommonSubexpressionEliminator.cpp

@@ -406,15 +406,15 @@ void CSECodeGenerator::generateClassElement(Id _c, bool _allowSequenced)
 		m_stack.erase(m_stackHeight - static_cast<int>(i));
 	}
 	appendItem(*expr.item);
-	if (expr.item->type() != Operation || instructionInfo(expr.item->instruction()).ret == 1)
+	if (expr.item->type() != Operation || instructionInfo(expr.item->instruction(), EVMVersion()).ret == 1)
 	{
 		m_stack[m_stackHeight] = _c;
 		m_classPositions[_c].insert(m_stackHeight);
 	}
 	else
 	{
 		assertThrow(
-			instructionInfo(expr.item->instruction()).ret == 0,
+			instructionInfo(expr.item->instruction(), EVMVersion()).ret == 0,
 			OptimizerException,
 			"Invalid number of return values."
 		);

libevmasm/ConstantOptimiser.cpp

@@ -79,18 +79,18 @@ unsigned ConstantOptimisationMethod::optimiseConstants(
 	return optimisations;
 }
 
-bigint ConstantOptimisationMethod::simpleRunGas(AssemblyItems const& _items)
+bigint ConstantOptimisationMethod::simpleRunGas(AssemblyItems const& _items, langutil::EVMVersion _evmVersion)
 {
 	bigint gas = 0;
 	for (AssemblyItem const& item: _items)
 		if (item.type() == Push)
-			gas += GasMeter::runGas(Instruction::PUSH1);
+			gas += GasMeter::runGas(Instruction::PUSH1, _evmVersion);
 		else if (item.type() == Operation)
 		{
 			if (item.instruction() == Instruction::EXP)
 				gas += GasCosts::expGas;
 			else
-				gas += GasMeter::runGas(item.instruction());
+				gas += GasMeter::runGas(item.instruction(), _evmVersion);
 		}
 	return gas;
 }
@@ -131,7 +131,7 @@ void ConstantOptimisationMethod::replaceConstants(
 bigint LiteralMethod::gasNeeded() const
 {
 	return combineGas(
-		simpleRunGas({Instruction::PUSH1}),
+		simpleRunGas({Instruction::PUSH1}, m_params.evmVersion),
 		// PUSHX plus data
 		(m_params.isCreation ? GasCosts::txDataNonZeroGas(m_params.evmVersion) : GasCosts::createDataGas) + dataGas(toCompactBigEndian(m_value, 1)),
 		0
@@ -142,7 +142,7 @@ bigint CodeCopyMethod::gasNeeded() const
 {
 	return combineGas(
 		// Run gas: we ignore memory increase costs
-		simpleRunGas(copyRoutine()) + GasCosts::copyGas,
+		simpleRunGas(copyRoutine(), m_params.evmVersion) + GasCosts::copyGas,
 		// Data gas for copy routines: Some bytes are zero, but we ignore them.
 		bytesRequired(copyRoutine()) * (m_params.isCreation ? GasCosts::txDataNonZeroGas(m_params.evmVersion) : GasCosts::createDataGas),
 		// Data gas for data itself
@@ -322,7 +322,7 @@ bigint ComputeMethod::gasNeeded(AssemblyItems const& _routine) const
 {
 	auto numExps = static_cast<size_t>(count(_routine.begin(), _routine.end(), Instruction::EXP));
 	return combineGas(
-		simpleRunGas(_routine) + numExps * (GasCosts::expGas + GasCosts::expByteGas(m_params.evmVersion)),
+		simpleRunGas(_routine, m_params.evmVersion) + numExps * (GasCosts::expGas + GasCosts::expByteGas(m_params.evmVersion)),
 		// Data gas for routine: Some bytes are zero, but we ignore them.
 		bytesRequired(_routine) * (m_params.isCreation ? GasCosts::txDataNonZeroGas(m_params.evmVersion) : GasCosts::createDataGas),
 		0

libevmasm/ConstantOptimiser.h

@@ -76,7 +76,7 @@ class ConstantOptimisationMethod
 
 protected:
 	/// @returns the run gas for the given items ignoring special gas costs
-	static bigint simpleRunGas(AssemblyItems const& _items);
+	static bigint simpleRunGas(AssemblyItems const& _items, langutil::EVMVersion _evmVersion);
 	/// @returns the gas needed to store the given data literally
 	bigint dataGas(bytes const& _data) const;
 	static size_t bytesRequired(AssemblyItems const& _items);

libevmasm/Disassemble.cpp

@@ -30,6 +30,7 @@ using namespace solidity::evmasm;
 
 void solidity::evmasm::eachInstruction(
 	bytes const& _mem,
+	langutil::EVMVersion _evmVersion,
 	function<void(Instruction,u256 const&)> const& _onInstruction
 )
 {
@@ -38,7 +39,7 @@ void solidity::evmasm::eachInstruction(
 		Instruction const instr{*it};
 		int additional = 0;
 		if (isValidInstruction(instr))
-			additional = instructionInfo(instr).additional;
+			additional = instructionInfo(instr, _evmVersion).additional;
 
 		u256 data{};
 
@@ -57,15 +58,15 @@ void solidity::evmasm::eachInstruction(
 	}
 }
 
-string solidity::evmasm::disassemble(bytes const& _mem, string const& _delimiter)
+string solidity::evmasm::disassemble(bytes const& _mem, langutil::EVMVersion _evmVersion, string const& _delimiter)
 {
 	stringstream ret;
-	eachInstruction(_mem, [&](Instruction _instr, u256 const& _data) {
+	eachInstruction(_mem, _evmVersion, [&](Instruction _instr, u256 const& _data) {
 		if (!isValidInstruction(_instr))
 			ret << "0x" << std::uppercase << std::hex << static_cast<int>(_instr) << _delimiter;
 		else
 		{
-			InstructionInfo info = instructionInfo(_instr);
+			InstructionInfo info = instructionInfo(_instr, _evmVersion);
 			ret << info.name;
 			if (info.additional)
 				ret << " 0x" << std::uppercase << std::hex << _data;

libevmasm/Disassemble.h

@@ -30,9 +30,9 @@ namespace solidity::evmasm
 {
 
 /// Iterate through EVM code and call a function on each instruction.
-void eachInstruction(bytes const& _mem, std::function<void(Instruction, u256 const&)> const& _onInstruction);
+void eachInstruction(bytes const& _mem, langutil::EVMVersion _evmVersion, std::function<void(Instruction, u256 const&)> const& _onInstruction);
 
 /// Convert from EVM code to simple EVM assembly language.
-std::string disassemble(bytes const& _mem, std::string const& _delimiter = " ");
+std::string disassemble(bytes const& _mem, langutil::EVMVersion _evmVersion, std::string const& _delimiter = " ");
 
 }