[BUG] Anonymous union initialized using 'from_bytes' in some cases #447 (#467)

alexey-utkin · web-flow · commit 7787136c3674 · 2022-09-19T19:12:08.000+03:00
diff --git a/server/src/Tests.cpp b/server/src/Tests.cpp
@@ -252,33 +252,81 @@ std::shared_ptr<StructValueView> KTestObjectParser::structView(const std::vector
                                                                size_t offsetInBits,
                                                                types::PointerUsage usage) {
     std::vector<InitReference> tmpInitReferences;
-    return structView(byteArray, curStruct, offsetInBits, usage, {}, "", {}, tmpInitReferences);
+    return structView(byteArray, curStruct, offsetInBits, usage, {}, false, "", {}, tmpInitReferences);
 }
 
 std::shared_ptr<StructValueView> KTestObjectParser::structView(const std::vector<char> &byteArray,
                                                                const StructInfo &curStruct,
                                                                size_t offsetInBits,
                                                                PointerUsage usage,
                                                                const std::optional<const Tests::MethodDescription> &testingMethod,
+                                                               const bool anonymousField,
                                                                const std::string &name,
                                                                const MapAddressName &fromAddressToName,
                                                                std::vector<InitReference> &initReferences) {
     std::vector<std::shared_ptr<AbstractValueView>> subViews;
 
+    size_t fieldIndexToInitUnion = SIZE_MAX;
+    size_t sizeOfFieldToInitUnion = 0;
+    size_t prevFieldEndOffset = offsetInBits;
+    size_t structEndOffset = offsetInBits + curStruct.size;
+    size_t fieldIndex = 0;
+    bool dirtyInitializedStruct = false;
     for (const auto &field: curStruct.fields) {
+        bool dirtyInitializedField = false;
         size_t fieldLen = typesHandler.typeSize(field.type);
-        size_t fieldOffset = offsetInBits + field.offset;
+        size_t fieldStartOffset = offsetInBits + field.offset;
+        size_t fieldEndOffset = fieldStartOffset + fieldLen;
+        if (curStruct.subType == types::SubType::Union) {
+            prevFieldEndOffset = offsetInBits;
+        }
+
+        auto dirtyCheck = [&](int i) {
+            if (i >= byteArray.size()) {
+                LOG_S(ERROR) << "Bad type size info: " << field.name << " index: " << fieldIndex;
+            } else if (byteArray[i] == 0) {
+                return false;
+            }
+            // the field cannot init the union in this state
+            dirtyInitializedField = true;
+            return true;
+        };
+
+        if (prevFieldEndOffset < fieldStartOffset) {
+            // check an alignment gap
+            for (int i = prevFieldEndOffset/8; i < fieldStartOffset/8; ++i) {
+                if (dirtyCheck(i)) {
+                    break;
+                }
+            }
+        }
+        if (!dirtyInitializedField && curStruct.subType == types::SubType::Union) {
+            // check the rest of the union
+            for (int i = fieldEndOffset/8; i < structEndOffset/8; ++i) {
+                if (dirtyCheck(i)) {
+                    break;
+                }
+            }
+        }
 
         switch (typesHandler.getTypeKind(field.type)) {
             case TypeKind::STRUCT_LIKE:
-                subViews.push_back(structView(byteArray, typesHandler.getStructInfo(field.type), fieldOffset, usage, testingMethod,
-                                              PrinterUtils::getFieldAccess(name, field), fromAddressToName, initReferences));
+                {
+                    auto sv = structView(byteArray, typesHandler.getStructInfo(field.type),
+                                         fieldStartOffset, usage, testingMethod, field.anonymous,
+                                         PrinterUtils::getFieldAccess(name, field), fromAddressToName,
+                                         initReferences);
+                    dirtyInitializedField |= sv->isDirtyInit();
+                    subViews.push_back(sv);
+                }
                 break;
             case TypeKind::ENUM:
-                subViews.push_back(enumView(byteArray, typesHandler.getEnumInfo(field.type), fieldOffset, fieldLen));
+                subViews.push_back(enumView(byteArray, typesHandler.getEnumInfo(field.type),
+                                            fieldStartOffset, fieldLen));
                 break;
             case TypeKind::PRIMITIVE:
-                subViews.push_back(primitiveView(byteArray, field.type.baseTypeObj(), fieldOffset,
+                subViews.push_back(primitiveView(byteArray, field.type.baseTypeObj(),
+                                                 fieldStartOffset,
                                                  std::min(field.size, fieldLen)));
                 break;
             case TypeKind::ARRAY: {
@@ -296,21 +344,22 @@ std::shared_ptr<StructValueView> KTestObjectParser::structView(const std::vector
                     }
                     if (onlyArrays) {
                         size *= typesHandler.typeSize(field.type.baseTypeObj());
-                        subViews.push_back(multiArrayView(byteArray, field.type, size,
-                                                          fieldOffset, usage));
+                        subViews.push_back(multiArrayView(byteArray, field.type, size, fieldStartOffset, usage));
                     } else {
                         std::vector<std::shared_ptr<AbstractValueView>> nullViews(
                             size, std::make_shared<JustValueView>(PrinterUtils::C_NULL));
                         subViews.push_back(std::make_shared<ArrayValueView>(nullViews));
                     }
                 } else {
-                    auto view = arrayView(byteArray, field.type.baseTypeObj(), fieldLen, fieldOffset, usage);
+                    auto view = arrayView(byteArray, field.type.baseTypeObj(), fieldLen,
+                                          fieldStartOffset, usage);
                     subViews.push_back(view);
                 }
             }
                 break;
             case TypeKind::OBJECT_POINTER: {
-                std::string res = readBytesAsValueForType(byteArray, PointerWidthType, fieldOffset,
+                std::string res = readBytesAsValueForType(byteArray, PointerWidthType,
+                                                          fieldStartOffset,
                                                           PointerWidthSizeInBits);
                 subViews.push_back(getLazyPointerView(fromAddressToName, initReferences,
                                                       PrinterUtils::getFieldAccess(name, field), res, field.type));
@@ -330,15 +379,35 @@ std::shared_ptr<StructValueView> KTestObjectParser::structView(const std::vector
                 LOG_S(ERROR) << message;
                 throw NoSuchTypeException(message);
         }
+
+        if (!dirtyInitializedField && sizeOfFieldToInitUnion < fieldLen) {
+            fieldIndexToInitUnion = fieldIndex;
+            sizeOfFieldToInitUnion = fieldLen;
+        } else {
+            dirtyInitializedStruct = true;
+        }
+        prevFieldEndOffset = fieldEndOffset;
+        ++fieldIndex;
     }
 
     std::optional<std::string> entryValue;
-    if (curStruct.hasAnonymousStructOrUnion) {
-        auto bytesType = types::Type::createSimpleTypeFromName("utbot_byte");
-        const std::shared_ptr<AbstractValueView> rawDataView = arrayView(byteArray, bytesType, curStruct.size, offsetInBits, usage);
-        entryValue = PrinterUtils::convertBytesToUnion(curStruct.name, rawDataView->getEntryValue(nullptr));
+    if (curStruct.subType == types::SubType::Union) {
+        if (fieldIndexToInitUnion == SIZE_MAX && !curStruct.name.empty()) {
+            // init by memory copy
+            entryValue = PrinterUtils::convertBytesToUnion(
+                curStruct.name,
+                arrayView(byteArray,
+                          types::Type::createSimpleTypeFromName("utbot_byte"),
+                          curStruct.size,
+                          offsetInBits, usage)->getEntryValue(nullptr));
+            dirtyInitializedStruct = false;
+        }
+        if (fieldIndexToInitUnion != SIZE_MAX) {
+            dirtyInitializedStruct = false;
+        }
     }
-    return std::make_shared<StructValueView>(curStruct, subViews, entryValue);
+    return std::make_shared<StructValueView>(curStruct, subViews, entryValue,
+                                             anonymousField, dirtyInitializedStruct, fieldIndexToInitUnion);
 }
 
 std::string KTestObjectParser::primitiveCharView(const types::Type &type, std::string value) {
@@ -962,7 +1031,7 @@ std::shared_ptr<AbstractValueView> KTestObjectParser::testParameterView(
     switch (typesHandler.getTypeKind(paramType)) {
         case TypeKind::STRUCT_LIKE:
             return structView(rawData, typesHandler.getStructInfo(paramType), 0,
-                              usage, testingMethod, param.varName, fromAddressToName, initReferences);
+                              usage, testingMethod, false, param.varName, fromAddressToName, initReferences);
         case TypeKind::ENUM:
             return enumView(rawData, typesHandler.getEnumInfo(paramType), 0, SizeUtils::bytesToBits(rawData.size()));
         case TypeKind::PRIMITIVE:
diff --git a/server/src/Tests.h b/server/src/Tests.h
@@ -3,12 +3,12 @@
 
 #include "Include.h"
 #include "LineInfo.h"
+#include "NameDecorator.h"
 #include "types/Types.h"
 #include "utils/CollectionUtils.h"
 #include "utils/PrinterUtils.h"
 #include "utils/SizeUtils.h"
 #include "json.hpp"
-
 #include <klee/KTest.h>
 #include <klee/TestCase.h>
 #include <tsl/ordered_map.h>
@@ -21,8 +21,8 @@
 #include <iterator>
 #include <memory>
 #include <optional>
-#include <type_traits>
 #include <queue>
+#include <type_traits>
 #include <utility>
 #include <vector>
 
@@ -233,9 +233,27 @@ namespace tests {
     struct StructValueView : AbstractValueView {
         explicit StructValueView(const types::StructInfo &_structInfo,
                                  std::vector<std::shared_ptr<AbstractValueView>> _subViews,
-                                 std::optional<std::string> _entryValue)
-            : AbstractValueView(std::move(_subViews)), entryValue(std::move(_entryValue)),
-              structInfo(_structInfo){
+                                 std::optional<std::string> _entryValue,
+                                 bool _anonymous,
+                                 bool _dirtyInit,
+                                 size_t _fieldIndexToInitUnion)
+            : AbstractValueView(std::move(_subViews))
+            , entryValue(std::move(_entryValue))
+            , structInfo(_structInfo)
+            , anonymous(_anonymous)
+            , dirtyInit(_dirtyInit)
+            , fieldIndexToInitUnion(_fieldIndexToInitUnion){}
+
+        bool isDirtyInit() const {
+            return dirtyInit;
+        }
+
+        bool isAnonymous() const {
+            return anonymous;
+        }
+
+        size_t getFieldIndexToInitUnion() const {
+            return fieldIndexToInitUnion;
         }
 
         [[nodiscard]] const std::vector<std::shared_ptr<AbstractValueView>> &getSubViews() const override {
@@ -254,7 +272,7 @@ namespace tests {
             std::vector<std::string> entries;
             size_t i = 0;
             for (const auto &subView : subViews) {
-                if (structInfo.subType == types::SubType::Struct || structInfo.longestFieldIndexForUnionInit == i) {
+                if (structInfo.subType == types::SubType::Struct || fieldIndexToInitUnion == i) {
                     entries.push_back(subView->getEntryValue(nullptr));
                 }
                 ++i;
@@ -273,6 +291,9 @@ namespace tests {
         }
 
         [[nodiscard]] std::string getFieldPrefix(int i) const {
+            if (structInfo.fields[i].name.empty())
+                return "";
+
             std::string prefix = "." + structInfo.fields[i].name + " = ";
             if (structInfo.isCLike) {
                 return prefix;
@@ -291,6 +312,10 @@ namespace tests {
     private:
         const types::StructInfo structInfo;
         std::optional<std::string> entryValue;
+
+        bool anonymous;
+        bool dirtyInit;
+        size_t fieldIndexToInitUnion;
     };
 
     struct InitReference {
@@ -679,6 +704,7 @@ namespace tests {
                                                     size_t offsetInBits,
                                                     types::PointerUsage usage,
                                                     const std::optional<const Tests::MethodDescription> &testingMethod,
+                                                    const bool anonymous,
                                                     const std::string &name,
                                                     const MapAddressName &fromAddressToName,
                                                     std::vector<InitReference> &initReferences);
diff --git a/server/src/printers/TestsPrinter.cpp b/server/src/printers/TestsPrinter.cpp
@@ -713,12 +713,12 @@ std::string printer::MultiLinePrinter::print(TestsPrinter *printer,
                                              const tests::StructValueView *view) {
     auto subViews = view->getSubViews();
     std::stringstream structuredValuesWithPrefixes;
-    structuredValuesWithPrefixes << "{" << NL;
+
+    structuredValuesWithPrefixes << (view->isAnonymous() ? "/* { */" : "{") << NL;
     ++printer->tabsDepth;
 
-    const types::StructInfo &structInfo = view->getStructInfo();
-    const size_t longestFieldIndexForUnionInit = structInfo.longestFieldIndexForUnionInit;
-    const bool isStruct = structInfo.subType == types::SubType::Struct;
+    const size_t fieldIndexToInitUnion = view->getFieldIndexToInitUnion();
+    const bool isStruct = view->getStructInfo().subType == types::SubType::Struct;
 
     size_t i = 0;
     for (const auto &sview : subViews) {
@@ -728,7 +728,7 @@ std::string printer::MultiLinePrinter::print(TestsPrinter *printer,
             structuredValuesWithPrefixes << NL;
         }
 
-        bool printInComment = !(isStruct || longestFieldIndexForUnionInit == i);
+        bool printInComment = !(isStruct || fieldIndexToInitUnion == i);
         if (printInComment) {
             ++printer->commentDepth;
         }
@@ -743,7 +743,9 @@ std::string printer::MultiLinePrinter::print(TestsPrinter *printer,
     }
 
     --printer->tabsDepth;
-    structuredValuesWithPrefixes << NL << printer->LINE_INDENT() << "}";
+    structuredValuesWithPrefixes << NL
+                                 << printer->LINE_INDENT()
+                                 << (view->isAnonymous() ? "/* } */" : "}");
 
     return structuredValuesWithPrefixes.str();
 }
diff --git a/server/src/types/Types.h b/server/src/types/Types.h
@@ -258,6 +258,7 @@ namespace types {
 
     struct Field {
         types::Type type;
+        bool anonymous;
         std::string name;
         /// size in @b bits
         size_t size;
@@ -291,7 +292,6 @@ namespace types {
 
     struct StructInfo: TypeInfo {
         std::vector<Field> fields{};
-        size_t longestFieldIndexForUnionInit;
         FPointerMap functionFields{};
         bool hasAnonymousStructOrUnion;
         bool isCLike;
diff --git a/server/src/types/TypesResolver.cpp b/server/src/types/TypesResolver.cpp
@@ -121,16 +121,15 @@ void TypesResolver::resolveStructEx(const clang::RecordDecl *D, const std::strin
        << "\tFile path: " << structInfo.filePath.string() << "";
     std::vector<types::Field> fields;
 
-    structInfo.longestFieldIndexForUnionInit = SIZE_MAX;
-    size_t i = 0;
-    size_t maxFieldSize = 0;
     for (const clang::FieldDecl *F : D->fields()) {
         if (F->isUnnamedBitfield()) {
             continue;
         }
-        structInfo.hasAnonymousStructOrUnion |= F->isAnonymousStructOrUnion();
         types::Field field;
+        field.anonymous = F->isAnonymousStructOrUnion();
         field.name = F->getNameAsString();
+        structInfo.hasAnonymousStructOrUnion |= field.anonymous;
+
         const clang::QualType paramType = F->getType().getCanonicalType();
         field.type = types::Type(paramType, paramType.getAsString(), sourceManager);
         if (field.type.isPointerToFunction()) {
@@ -175,11 +174,6 @@ void TypesResolver::resolveStructEx(const clang::RecordDecl *D, const std::strin
             field.accessSpecifier = types::Field::AS_pubic;
         }
         fields.push_back(field);
-        if (subType == types::SubType::Union && maxFieldSize < field.size) {
-            structInfo.longestFieldIndexForUnionInit = i;
-            maxFieldSize = field.size;
-        }
-        ++i;
     }
     structInfo.fields = fields;
     structInfo.size = getRecordSize(D);
diff --git a/server/test/framework/Syntax_Tests.cpp b/server/test/framework/Syntax_Tests.cpp
