Merge pull request #15595 from ipsilon/eof-container-id-fix

eof: Contract creation fix and tests.
ethereum · Nov 29, 2024 · 0b01a8a · 0b01a8a
2 parents d2a7eb3 + 6b347a1
commit 0b01a8a
Show file tree

Hide file tree

Showing 6 changed files with 4,505 additions and 20 deletions.
diff --git a/libevmasm/Assembly.cpp b/libevmasm/Assembly.cpp
@@ -924,7 +924,7 @@ void appendBigEndianUint16(bytes& _dest, ValueT _value)
 }
 }
 
-std::tuple<bytes, std::vector<size_t>, size_t> Assembly::createEOFHeader(std::set<uint16_t> const& _referencedSubIds) const
+std::tuple<bytes, std::vector<size_t>, size_t> Assembly::createEOFHeader(std::set<ContainerID> const& _referencedSubIds) const
 {
 	bytes retBytecode;
 	std::vector<size_t> codeSectionSizePositions;
@@ -1330,9 +1330,9 @@ LinkerObject const& Assembly::assembleLegacy() const
 	return ret;
 }
 
-std::map<uint16_t, uint16_t> Assembly::findReferencedContainers() const
+std::map<ContainerID, ContainerID> Assembly::findReferencedContainers() const
 {
-	std::set<uint16_t> referencedSubcontainersIds;
+	std::set<ContainerID> referencedSubcontainersIds;
 	solAssert(m_subs.size() <= 0x100); // According to EOF spec
 
 	for (auto&& codeSection: m_codeSections)
@@ -1344,12 +1344,13 @@ std::map<uint16_t, uint16_t> Assembly::findReferencedContainers() const
 				referencedSubcontainersIds.insert(containerId);
 			}
 
-	std::map<uint16_t, uint16_t> replacements;
+	std::map<ContainerID, ContainerID> replacements;
 	uint8_t nUnreferenced = 0;
-	for (uint8_t i = 0; i < static_cast<uint16_t>(m_subs.size()); ++i)
+	for (size_t i = 0; i < m_subs.size(); ++i)
 	{
-		if (referencedSubcontainersIds.count(i) > 0)
-			replacements[i] = static_cast<uint16_t>(i - nUnreferenced);
+		solAssert(i <= std::numeric_limits<ContainerID>::max());
+		if (referencedSubcontainersIds.count(static_cast<ContainerID>(i)) > 0)
+			replacements[static_cast<ContainerID>(i)] = static_cast<ContainerID>(i - nUnreferenced);
 		else
 			nUnreferenced++;
 	}
@@ -1441,13 +1442,19 @@ LinkerObject const& Assembly::assembleEOF() const
 			case EOFCreate:
 			{
 				ret.bytecode.push_back(static_cast<uint8_t>(Instruction::EOFCREATE));
-				ret.bytecode.push_back(static_cast<uint8_t>(item.data()));
+				solAssert(item.data() <= std::numeric_limits<ContainerID>::max());
+				auto const containerID = static_cast<ContainerID>(item.data());
+				solAssert(subIdsReplacements.count(containerID) == 1);
+				ret.bytecode.push_back(subIdsReplacements.at(containerID));
 				break;
 			}
 			case ReturnContract:
 			{
 				ret.bytecode.push_back(static_cast<uint8_t>(Instruction::RETURNCONTRACT));
-				ret.bytecode.push_back(static_cast<uint8_t>(item.data()));
+				solAssert(item.data() <= std::numeric_limits<ContainerID>::max());
+				auto const containerID = static_cast<ContainerID>(item.data());
+				solAssert(subIdsReplacements.count(containerID) == 1);
+				ret.bytecode.push_back(subIdsReplacements.at(containerID));
 				break;
 			}
 			case VerbatimBytecode:

diff --git a/libevmasm/Assembly.h b/libevmasm/Assembly.h
@@ -247,13 +247,13 @@ class Assembly
 	std::shared_ptr<std::string const> sharedSourceName(std::string const& _name) const;
 
 	/// Returns EOF header bytecode | code section sizes offsets | data section size offset
-	std::tuple<bytes, std::vector<size_t>, size_t> createEOFHeader(std::set<uint16_t> const& _referencedSubIds) const;
+	std::tuple<bytes, std::vector<size_t>, size_t> createEOFHeader(std::set<ContainerID> const& _referencedSubIds) const;
 
 	LinkerObject const& assembleLegacy() const;
 	LinkerObject const& assembleEOF() const;
 
 	/// Returns map from m_subs to an index of subcontainer in the final EOF bytecode
-	std::map<uint16_t, uint16_t> findReferencedContainers() const;
+	std::map<ContainerID, ContainerID> findReferencedContainers() const;
 	/// Returns max AuxDataLoadN offset for the assembly.
 	std::optional<uint16_t> findMaxAuxDataLoadNOffset() const;
 

diff --git a/libyul/AsmAnalysis.cpp b/libyul/AsmAnalysis.cpp
@@ -800,13 +800,27 @@ bool AsmAnalyzer::validateInstructions(FunctionCall const& _functionCall)
 
 void AsmAnalyzer::validateObjectStructure(langutil::SourceLocation _astRootLocation)
 {
-	if (m_eofVersion.has_value() && util::contains(m_objectStructure.objectName, '.')) // No dots in object name for EOF
-		m_errorReporter.syntaxError(
-			9822_error,
-			_astRootLocation,
-			fmt::format(
-				"The object name \"{objectName}\" is invalid in EOF context. Object names must not contain 'dot' character.",
-				fmt::arg("objectName", m_objectStructure.objectName)
-			)
-		);
+	if (m_eofVersion.has_value())
+	{
+		if (util::contains(m_objectStructure.objectName, '.')) // No dots in object name for EOF
+			m_errorReporter.syntaxError(
+				9822_error,
+				_astRootLocation,
+				fmt::format(
+					"The object name \"{objectName}\" is invalid in EOF context. Object names must not contain 'dot' character.",
+					fmt::arg("objectName", m_objectStructure.objectName)
+				)
+			);
+		else if (m_objectStructure.topLevelSubObjectNames().size() > 256)
+		{
+			m_errorReporter.syntaxError(
+				1305_error,
+				_astRootLocation,
+				fmt::format(
+					"Too many subobjects in \"{objectName}\". At most 256 subobjects allowed when compiling to EOF",
+					fmt::arg("objectName", m_objectStructure.objectName)
+				)
+			);
+		}
+	}
 }