Merge pull request #8250 from ethereum/run-opt-on-user-asm-8031

Run yul optimizer on user code without refs

libsolidity/codegen/CompilerContext.cpp

@@ -423,18 +423,12 @@ void CompilerContext::appendInlineAssembly(
 	// so we essentially only optimize the ABI functions.
 	if (_optimiserSettings.runYulOptimiser && _localVariables.empty())
 	{
-		bool const isCreation = m_runtimeContext != nullptr;
-		yul::GasMeter meter(dialect, isCreation, _optimiserSettings.expectedExecutionsPerDeployment);
 		yul::Object obj;
 		obj.code = parserResult;
 		obj.analysisInfo = make_shared<yul::AsmAnalysisInfo>(analysisInfo);
-		yul::OptimiserSuite::run(
-			dialect,
-			&meter,
-			obj,
-			_optimiserSettings.optimizeStackAllocation,
-			externallyUsedIdentifiers
-		);
+
+		optimizeYul(obj, dialect, _optimiserSettings, externallyUsedIdentifiers);
+
 		analysisInfo = std::move(*obj.analysisInfo);
 		parserResult = std::move(obj.code);
 
@@ -462,6 +456,29 @@ void CompilerContext::appendInlineAssembly(
 	updateSourceLocation();
 }
 
+
+void CompilerContext::optimizeYul(yul::Object& _object, yul::EVMDialect const& _dialect, OptimiserSettings const& _optimiserSettings, std::set<yul::YulString> const& _externalIdentifiers)
+{
+#ifdef SOL_OUTPUT_ASM
+	cout << yul::AsmPrinter(*dialect)(*_object.code) << endl;
+#endif
+
+	bool const isCreation = runtimeContext() != nullptr;
+	yul::GasMeter meter(_dialect, isCreation, _optimiserSettings.expectedExecutionsPerDeployment);
+	yul::OptimiserSuite::run(
+		_dialect,
+		&meter,
+		_object,
+		_optimiserSettings.optimizeStackAllocation,
+		_externalIdentifiers
+	);
+
+#ifdef SOL_OUTPUT_ASM
+	cout << "After optimizer:" << endl;
+	cout << yul::AsmPrinter(*dialect)(*object.code) << endl;
+#endif
+}
+
 FunctionDefinition const& CompilerContext::resolveVirtualFunction(
 	FunctionDefinition const& _function,
 	vector<ContractDefinition const*>::const_iterator _searchStart

libsolidity/codegen/CompilerContext.h

@@ -32,9 +32,13 @@
 
 #include <libevmasm/Assembly.h>
 #include <libevmasm/Instruction.h>
+#include <liblangutil/ErrorReporter.h>
 #include <liblangutil/EVMVersion.h>
 #include <libsolutil/Common.h>
 
+#include <libyul/AsmAnalysisInfo.h>
+#include <libyul/backends/evm/EVMDialect.h>
+
 #include <functional>
 #include <ostream>
 #include <stack>
@@ -232,6 +236,8 @@ public:
 	/// Otherwise returns "revert(0, 0)".
 	std::string revertReasonIfDebug(std::string const& _message = "");
 
+	void optimizeYul(yul::Object& _object, yul::EVMDialect const& _dialect, OptimiserSettings const& _optimiserSetting, std::set<yul::YulString> const& _externalIdentifiers = {});
+
 	/// Appends arbitrary data to the end of the bytecode.
 	void appendAuxiliaryData(bytes const& _data) { m_asm->appendAuxiliaryDataToEnd(_data); }
 

libsolidity/codegen/ContractCompiler.cpp

@@ -20,6 +20,7 @@
  * Solidity compiler.
  */
 
+
 #include <libsolidity/ast/AST.h>
 #include <libsolidity/ast/ASTUtils.h>
 #include <libsolidity/ast/TypeProvider.h>
@@ -27,7 +28,16 @@
 #include <libsolidity/codegen/ContractCompiler.h>
 #include <libsolidity/codegen/ExpressionCompiler.h>
 
+#include <libyul/AsmAnalysisInfo.h>
+#include <libyul/AsmAnalysis.h>
+#include <libyul/AsmData.h>
 #include <libyul/backends/evm/AsmCodeGen.h>
+#include <libyul/backends/evm/EVMMetrics.h>
+#include <libyul/backends/evm/EVMDialect.h>
+#include <libyul/optimiser/Suite.h>
+#include <libyul/Object.h>
+#include <libyul/optimiser/ASTCopier.h>
+#include <libyul/YulString.h>
 
 #include <libevmasm/Instruction.h>
 #include <libevmasm/Assembly.h>
@@ -38,6 +48,7 @@
 #include <libsolutil/Whiskers.h>
 
 #include <boost/range/adaptor/reversed.hpp>
+
 #include <algorithm>
 
 using namespace std;
@@ -788,10 +799,36 @@ bool ContractCompiler::visit(InlineAssembly const& _inlineAssembly)
 			_assembly.appendInstruction(Instruction::POP);
 		}
 	};
-	solAssert(_inlineAssembly.annotation().analysisInfo, "");
+
+	yul::Block const* code = &_inlineAssembly.operations();
+	yul::AsmAnalysisInfo* analysisInfo = _inlineAssembly.annotation().analysisInfo.get();
+
+	// Only used in the scope below, but required to live outside to keep the
+	// shared_ptr's alive
+	yul::Object object = {};
+
+	// The optimiser cannot handle external references
+	if (
+		m_optimiserSettings.runYulOptimiser &&
+		_inlineAssembly.annotation().externalReferences.empty()
+	)
+	{
+		yul::EVMDialect const* dialect = dynamic_cast<decltype(dialect)>(&_inlineAssembly.dialect());
+		solAssert(dialect, "");
+
+		// Create a modifiable copy of the code and analysis
+		object.code = make_shared<yul::Block>(yul::ASTCopier().translate(*code));
+		object.analysisInfo = make_shared<yul::AsmAnalysisInfo>(yul::AsmAnalyzer::analyzeStrictAssertCorrect(*dialect, object));
+
+		m_context.optimizeYul(object, *dialect, m_optimiserSettings);
+
+		code = object.code.get();
+		analysisInfo = object.analysisInfo.get();
+	}
+
 	yul::CodeGenerator::assemble(
-		_inlineAssembly.operations(),
-		*_inlineAssembly.annotation().analysisInfo,
+		*code,
+		*analysisInfo,
 		*m_context.assemblyPtr(),
 		m_context.evmVersion(),
 		identifierAccess,

libyul/optimiser/ASTCopier.h

@@ -84,6 +84,7 @@ public:
 	virtual Expression translate(Expression const& _expression);
 	virtual Statement translate(Statement const& _statement);
 
+	Block translate(Block const& _block);
 protected:
 	template <typename T>
 	std::vector<T> translateVector(std::vector<T> const& _values);
@@ -94,7 +95,6 @@ protected:
 		return _v ? std::make_unique<T>(translate(*_v)) : nullptr;
 	}
 
-	Block translate(Block const& _block);
 	Case translate(Case const& _case);
 	virtual Identifier translate(Identifier const& _identifier);
 	Literal translate(Literal const& _literal);

scripts/check_style.sh

@@ -17,6 +17,7 @@ fi
 
 FORMATERROR=$(
 (
+	git grep -nIE "#include \"" -- '*.h' '*.cpp' | egrep -v -e "license.h" -e "BuildInfo.h"  # Use include with <> characters
 	git grep -nIE "\<(if|for|while|switch)\(" -- '*.h' '*.cpp' # no space after "if", "for", "while" or "switch"
 	git grep -nIE "\<for\>\s*\([^=]*\>\s:\s.*\)" -- '*.h' '*.cpp' # no space before range based for-loop
 	git grep -nIE "\<if\>\s*\(.*\)\s*\{\s*$" -- '*.h' '*.cpp' # "{\n" on same line as "if" / "for"

test/cmdlineTests.sh

@@ -155,6 +155,7 @@ function test_solc_behaviour()
         rm "$stdout_path.bak"
     else
         sed -i.bak -e '/^Warning: This is a pre-release compiler version, please do not use it in production./d' "$stderr_path"
+        sed -i.bak -e 's/\(^[ ]*auxdata: \)0x[0-9a-f]*$/\1AUXDATA REMOVED/' "$stdout_path"
         sed -i.bak -e 's/ Consider adding "pragma .*$//' "$stderr_path"
         # Remove trailing empty lines. Needs a line break to make OSX sed happy.
         sed -i.bak -e '1{/^$/d

test/cmdlineTests/optimizer_user_yul/args

@@ -0,0 +1 @@
+--optimize --asm --metadata-hash none

test/cmdlineTests/optimizer_user_yul/input.sol

@@ -0,0 +1,37 @@
+pragma solidity >=0.0;
+
+contract C
+{
+	constructor() public payable
+	{
+		int a;
+
+		// Can't be optimized due to external reference "a"
+		assembly
+		{
+			let x,y,z
+
+			sstore(0, 1)
+
+			for { } sload(4) { } {
+				z := exp(x, y)
+			}
+
+			a := 2
+		}
+
+		// Can be optimized due to no external references
+		assembly
+		{
+			let x,y,z
+
+			sstore(2, 3)
+
+			for { } sload(5) { } {
+				// Expected to be optimized out for yulOptimizer, but not for
+				// old optimizer
+				z := exp(x, y)
+			}
+		}
+	}
+}

test/cmdlineTests/optimizer_user_yul/output

@@ -0,0 +1,87 @@
+
+======= optimizer_user_yul/input.sol:C =======
+EVM assembly:
+    /* "optimizer_user_yul/input.sol":24:489  contract C... */
+  mstore(0x40, 0x80)
+    /* "optimizer_user_yul/input.sol":72:77  int a */
+  0x00
+    /* "optimizer_user_yul/input.sol":38:487  constructor() public payable... */
+  dup1
+  0x00
+  dup1
+    /* "optimizer_user_yul/input.sol":176:177  1 */
+  0x01
+    /* "optimizer_user_yul/input.sol":173:174  0 */
+  0x00
+    /* "optimizer_user_yul/input.sol":166:178  sstore(0, 1) */
+  sstore
+    /* "optimizer_user_yul/input.sol":183:229  for { } sload(4) { } {... */
+tag_3:
+    /* "optimizer_user_yul/input.sol":197:198  4 */
+  0x04
+    /* "optimizer_user_yul/input.sol":191:199  sload(4) */
+  sload
+    /* "optimizer_user_yul/input.sol":183:229  for { } sload(4) { } {... */
+  iszero
+  tag_5
+  jumpi
+  pop
+    /* "optimizer_user_yul/input.sol":215:224  exp(x, y) */
+  dup1
+  dup3
+  exp
+    /* "optimizer_user_yul/input.sol":183:229  for { } sload(4) { } {... */
+  jump(tag_3)
+tag_5:
+    /* "optimizer_user_yul/input.sol":187:190  { } */
+  pop
+  pop
+  pop
+    /* "optimizer_user_yul/input.sol":239:240  2 */
+  0x02
+    /* "optimizer_user_yul/input.sol":234:240  a := 2 */
+  swap1
+  pop
+    /* "optimizer_user_yul/input.sol":340:341  3 */
+  0x03
+    /* "optimizer_user_yul/input.sol":337:338  2 */
+  0x02
+    /* "optimizer_user_yul/input.sol":330:342  sstore(2, 3) */
+  sstore
+    /* "optimizer_user_yul/input.sol":347:480  for { } sload(5) { } {... */
+tag_6:
+    /* "optimizer_user_yul/input.sol":361:362  5 */
+  0x05
+    /* "optimizer_user_yul/input.sol":355:363  sload(5) */
+  sload
+  tag_9
+  jumpi
+  jump(tag_8)
+tag_9:
+    /* "optimizer_user_yul/input.sol":347:480  for { } sload(5) { } {... */
+  jump(tag_6)
+tag_8:
+    /* "optimizer_user_yul/input.sol":311:484  {... */
+  pop
+    /* "optimizer_user_yul/input.sol":24:489  contract C... */
+  dataSize(sub_0)
+  dup1
+  dataOffset(sub_0)
+  0x00
+  codecopy
+  0x00
+  return
+stop
+
+sub_0: assembly {
+        /* "optimizer_user_yul/input.sol":24:489  contract C... */
+      mstore(0x40, 0x80)
+        /* "--CODEGEN--":12:13   */
+      0x00
+        /* "--CODEGEN--":9:10   */
+      dup1
+        /* "--CODEGEN--":2:14   */
+      revert
+
+    auxdata: AUXDATA REMOVED
+}