Merge pull request #5716 from ethereum/fixDataSizeOffset

Fix data size offset and properly provide builtins.

6 files changed, 81 insertions, 7 deletions

diff --git a/test/libyul/ObjectParser.cpp b/test/libyul/ObjectParser.cpp
index bb88e4da..13a95788 100644
--- a/test/libyul/ObjectParser.cpp
+++ b/test/libyul/ObjectParser.cpp
@@ -250,6 +250,36 @@ BOOST_AUTO_TEST_CASE(to_string)
 	BOOST_CHECK_EQUAL(asmStack.print(), expectation);
 }
 
+BOOST_AUTO_TEST_CASE(arg_to_dataoffset_must_be_literal)
+{
+	string code = R"(
+		object "outer" {
+			code { let x := "outer" let y := dataoffset(x) }
+		}
+	)";
+	CHECK_ERROR(code, TypeError, "Function expects direct literals as arguments.");
+}
+
+BOOST_AUTO_TEST_CASE(arg_to_datasize_must_be_literal)
+{
+	string code = R"(
+		object "outer" {
+			code { let x := "outer" let y := datasize(x) }
+		}
+	)";
+	CHECK_ERROR(code, TypeError, "Function expects direct literals as arguments.");
+}
+
+BOOST_AUTO_TEST_CASE(args_to_datacopy_are_arbitrary)
+{
+	string code = R"(
+		object "outer" {
+			code { let x := 0 let y := 2 let s := 3 datacopy(x, y, s) }
+		}
+	)";
+	BOOST_CHECK(successParse(code));
+}
+
 BOOST_AUTO_TEST_SUITE_END()
 
 }
diff --git a/test/libyul/Parser.cpp b/test/libyul/Parser.cpp
index df7e32a1..897f18ae 100644
--- a/test/libyul/Parser.cpp
+++ b/test/libyul/Parser.cpp
@@ -331,7 +331,7 @@ BOOST_AUTO_TEST_CASE(builtins_analysis)
 		{
 			return _name == "builtin"_yulstring ? &f : nullptr;
 		}
-		BuiltinFunction f{"builtin"_yulstring, vector<Type>(2), vector<Type>(3), false};
+		BuiltinFunction f{"builtin"_yulstring, vector<Type>(2), vector<Type>(3), false, false};
 	};
 
 	shared_ptr<Dialect> dialect = make_shared<SimpleDialect>();
diff --git a/test/libyul/YulOptimizerTest.cpp b/test/libyul/YulOptimizerTest.cpp
index 1b392e0f..0e3203cb 100644
--- a/test/libyul/YulOptimizerTest.cpp
+++ b/test/libyul/YulOptimizerTest.cpp
@@ -122,7 +122,7 @@ bool YulOptimizerTest::run(ostream& _stream, string const& _linePrefix, bool con
 	else if (m_optimizerStep == "expressionSplitter")
 	{
 		NameDispenser nameDispenser{*m_dialect, *m_ast};
-		ExpressionSplitter{nameDispenser}(*m_ast);
+		ExpressionSplitter{*m_dialect, nameDispenser}(*m_ast);
 	}
 	else if (m_optimizerStep == "expressionJoiner")
 	{
@@ -133,7 +133,7 @@ bool YulOptimizerTest::run(ostream& _stream, string const& _linePrefix, bool con
 	{
 		disambiguate();
 		NameDispenser nameDispenser{*m_dialect, *m_ast};
-		ExpressionSplitter{nameDispenser}(*m_ast);
+		ExpressionSplitter{*m_dialect, nameDispenser}(*m_ast);
 		ExpressionJoiner::run(*m_ast);
 		ExpressionJoiner::run(*m_ast);
 	}
@@ -158,7 +158,7 @@ bool YulOptimizerTest::run(ostream& _stream, string const& _linePrefix, bool con
 		(FunctionHoister{})(*m_ast);
 		(FunctionGrouper{})(*m_ast);
 		NameDispenser nameDispenser{*m_dialect, *m_ast};
-		ExpressionSplitter{nameDispenser}(*m_ast);
+		ExpressionSplitter{*m_dialect, nameDispenser}(*m_ast);
 		FullInliner(*m_ast, nameDispenser).run();
 		ExpressionJoiner::run(*m_ast);
 	}
@@ -182,7 +182,7 @@ bool YulOptimizerTest::run(ostream& _stream, string const& _linePrefix, bool con
 	{
 		disambiguate();
 		NameDispenser nameDispenser{*m_dialect, *m_ast};
-		ExpressionSplitter{nameDispenser}(*m_ast);
+		ExpressionSplitter{*m_dialect, nameDispenser}(*m_ast);
 		CommonSubexpressionEliminator{*m_dialect}(*m_ast);
 		ExpressionSimplifier::run(*m_dialect, *m_ast);
 		UnusedPruner::runUntilStabilised(*m_dialect, *m_ast);
@@ -260,7 +260,7 @@ void YulOptimizerTest::printIndented(ostream& _stream, string const& _output, st
 
 bool YulOptimizerTest::parse(ostream& _stream, string const& _linePrefix, bool const _formatted)
 {
-	m_dialect = m_yul ? yul::Dialect::yul() : yul::EVMDialect::strictAssemblyForEVM();
+	m_dialect = m_yul ? yul::Dialect::yul() : yul::EVMDialect::strictAssemblyForEVMObjects();
 	ErrorList errors;
 	ErrorReporter errorReporter(errors);
 	shared_ptr<Scanner> scanner = make_shared<Scanner>(CharStream(m_source, ""));
diff --git a/test/libyul/yulOptimizerTests/commonSubexpressionEliminator/object_access.yul b/test/libyul/yulOptimizerTests/commonSubexpressionEliminator/object_access.yul
new file mode 100644
index 00000000..5cfa3e6e
--- /dev/null
+++ b/test/libyul/yulOptimizerTests/commonSubexpressionEliminator/object_access.yul
@@ -0,0 +1,23 @@
+{
+    // Arguments to ``datasize`` and ``dataoffset`` need to be
+    // literals. We cannot simplify their arguments, but we can
+    // simplify them as a full expression.
+    // ``datacopy`` does not have this restriction.
+    let r := "abc"
+    let a := datasize("abc")
+    let x := dataoffset("abc")
+    // should be replaced by a
+    let y := datasize("abc")
+    datacopy("abc", x, y)
+    mstore(a, x)
+}
+// ----
+// commonSubexpressionEliminator
+// {
+//     let r := "abc"
+//     let a := datasize("abc")
+//     let x := dataoffset("abc")
+//     let y := a
+//     datacopy(r, x, a)
+//     mstore(a, x)
+// }
diff --git a/test/libyul/yulOptimizerTests/expressionSplitter/object_access.yul b/test/libyul/yulOptimizerTests/expressionSplitter/object_access.yul
new file mode 100644
index 00000000..2689ab6f
--- /dev/null
+++ b/test/libyul/yulOptimizerTests/expressionSplitter/object_access.yul
@@ -0,0 +1,21 @@
+{
+    // We should never split arguments to ``dataoffset``
+    // or ``datasize`` because they need to be literals
+    let x := dataoffset("abc")
+    let y := datasize("abc")
+    // datacopy is fine, though
+    datacopy(mload(0), mload(1), mload(2))
+}
+// ----
+// expressionSplitter
+// {
+//     let x := dataoffset("abc")
+//     let y := datasize("abc")
+//     let _1 := 2
+//     let _2 := mload(_1)
+//     let _3 := 1
+//     let _4 := mload(_3)
+//     let _5 := 0
+//     let _6 := mload(_5)
+//     datacopy(_6, _4, _2)
+// }
diff --git a/test/tools/yulopti.cpp b/test/tools/yulopti.cpp
index 8ef56ffe..ac21fd91 100644
--- a/test/tools/yulopti.cpp
+++ b/test/tools/yulopti.cpp
@@ -149,7 +149,7 @@ public:
 				(VarDeclInitializer{})(*m_ast);
 				break;
 			case 'x':
-				ExpressionSplitter{*m_nameDispenser}(*m_ast);
+				ExpressionSplitter{*m_dialect, *m_nameDispenser}(*m_ast);
 				break;
 			case 'j':
 				ExpressionJoiner::run(*m_ast);