Misc fixes (#743)

* Make TaintConfigData compatible with C++20

* Compatibility with opaque pointers

* Allow setting the LLVM version from higher-level projects

* Getting rid of UB in CHAResolver

* Better IntraMonoSolver dump

* minor

* improve intra mono dump

* minor

* expose getNonPureVirtualVFTEntry as namespace-scope function + fix build with LLVM 14

* Fix solveIFDSProblem

---------

Co-authored-by: Fabian Schiebel <fabianbs@mail.uni-paderborn.de>

Author: fabianbs96

CMakeLists.txt

@@ -8,6 +8,10 @@ set(CMAKE_POLICY_DEFAULT_CMP0069 NEW)
 cmake_policy(SET CMP0077 NEW)
 set(CMAKE_POLICY_DEFAULT_CMP0077 NEW)
 
+# Allow overwriting cache variables of external projects from this CMakeLists file
+cmake_policy(SET CMP0126 NEW)
+set(CMAKE_POLICY_DEFAULT_CMP0126 NEW)
+
 # Allow portable use of CMAKE_VISIBILITY_INLINES_HIDDEN not only for shared libraries
 cmake_policy(SET CMP0063 NEW)
 set(CMAKE_POLICY_DEFAULT_CMP0063 NEW)
@@ -303,6 +307,9 @@ endif()
 option(USE_LLVM_FAT_LIB  "Link against libLLVM.so instead of the individual LLVM libraries if possible (default is OFF; always on if BUILD_SHARED_LIBS is ON)" OFF)
 
 # LLVM
+if (NOT PHASAR_LLVM_VERSION)
+  set(PHASAR_LLVM_VERSION 14)
+endif()
 include(add_llvm)
 add_llvm()
 

cmake/add_llvm.cmake

@@ -2,7 +2,7 @@ macro(add_llvm)
 
   if (NOT PHASAR_IN_TREE)
     # Only search for LLVM if we build out of tree
-    find_package(LLVM 14 REQUIRED CONFIG)
+    find_package(LLVM ${PHASAR_LLVM_VERSION} REQUIRED CONFIG)
     find_library(LLVM_LIBRARY NAMES LLVM PATHS ${LLVM_LIBRARY_DIRS} NO_DEFAULT_PATH)
 
     if(USE_LLVM_FAT_LIB AND ${LLVM_LIBRARY} STREQUAL "LLVM_LIBRARY-NOTFOUND")

cmake/phasar_macros.cmake

@@ -125,10 +125,21 @@ function(generate_ll_file)
   endif()
 
   if(GEN_LL_MEM2REG)
+    if(CMAKE_CXX_COMPILER_ID MATCHES "Clang")
+      get_filename_component(COMPILER_PATH_STR ${CMAKE_CXX_COMPILER} DIRECTORY)
+      find_program(OPT_TOOL opt HINTS ${COMPILER_PATH_STR})
+    else()
+      find_program(OPT_TOOL opt)
+    endif()
+
+    if(NOT OPT_TOOL)
+      set(OPT_TOOL opt)
+    endif()
+
     add_custom_command(
       OUTPUT ${test_code_ll_file}
       COMMAND ${GEN_CMD} ${test_code_file_path} -o ${test_code_ll_file}
-      COMMAND ${CMAKE_CXX_COMPILER_LAUNCHER} opt -mem2reg -S -opaque-pointers=0 ${test_code_ll_file} -o ${test_code_ll_file}
+      COMMAND ${CMAKE_CXX_COMPILER_LAUNCHER} ${OPT_TOOL} -mem2reg -S -opaque-pointers=0 ${test_code_ll_file} -o ${test_code_ll_file}
       COMMENT ${GEN_CMD_COMMENT}
       DEPENDS ${GEN_LL_FILE}
       VERBATIM

include/phasar/DataFlow/IfdsIde/Solver/IFDSSolver.h

@@ -21,7 +21,6 @@
 #include "phasar/DataFlow/IfdsIde/Solver/IDESolver.h"
 #include "phasar/Domain/BinaryDomain.h"
 
-#include <memory>
 #include <set>
 #include <type_traits>
 #include <unordered_map>
@@ -118,8 +117,7 @@ using IFDSSolver_P = IFDSSolver<typename Problem::ProblemAnalysisDomain,
 
 template <typename AnalysisDomainTy, typename Container>
 OwningSolverResults<typename AnalysisDomainTy::n_t,
-                    typename AnalysisDomainTy::d_t,
-                    typename AnalysisDomainTy::l_t>
+                    typename AnalysisDomainTy::d_t, BinaryDomain>
 solveIFDSProblem(IFDSTabulationProblem<AnalysisDomainTy, Container> &Problem,
                  const typename AnalysisDomainTy::i_t &ICF) {
   IFDSSolver<AnalysisDomainTy, Container> Solver(Problem, &ICF);

include/phasar/DataFlow/Mono/Solver/IntraMonoSolver.h

@@ -18,7 +18,9 @@
 #define PHASAR_DATAFLOW_MONO_SOLVER_INTRAMONOSOLVER_H
 
 #include "phasar/DataFlow/Mono/IntraMonoProblem.h"
-#include "phasar/Utils/BitVectorSet.h"
+#include "phasar/PhasarLLVM/Utils/LLVMShorthands.h"
+#include "phasar/Utils/DefaultValue.h"
+#include "phasar/Utils/Utilities.h"
 
 #include <deque>
 #include <unordered_map>
@@ -109,20 +111,47 @@ template <typename AnalysisDomainTy> class IntraMonoSolver {
     }
   }
 
-  mono_container_t getResultsAt(n_t Stmt) { return Analysis[Stmt]; }
+  [[nodiscard]] const mono_container_t &getResultsAt(n_t Stmt) const {
+    auto It = Analysis.find(Stmt);
+    if (It != Analysis.end()) {
+      return It->second;
+    }
+    return getDefaultValue<mono_container_t>();
+  }
 
   virtual void dumpResults(llvm::raw_ostream &OS = llvm::outs()) {
-    OS << "Intra-Monotone solver results:\n"
-          "------------------------------\n";
-    for (auto &[Node, FlowFacts] : this->Analysis) {
-      OS << "Instruction:\n" << NToString(Node);
-      OS << "\nFacts:\n";
+    OS << "\n***************************************************************\n"
+       << "*                 Raw IntraMonoSolver results                 *\n"
+       << "***************************************************************\n";
+
+    if (Analysis.empty()) {
+      OS << "No results computed!" << '\n';
+      return;
+    }
+
+    std::vector<std::pair<n_t, mono_container_t>> Cells;
+    Cells.reserve(Analysis.size());
+    Cells.insert(Cells.end(), Analysis.begin(), Analysis.end());
+
+    std::sort(Cells.begin(), Cells.end(), [](const auto &Lhs, const auto &Rhs) {
+      if constexpr (std::is_same_v<n_t, const llvm::Instruction *>) {
+        return StringIDLess{}(getMetaDataID(Lhs.first),
+                              getMetaDataID(Rhs.first));
+      } else {
+        // If non-LLVM IR is used
+        return Lhs.first < Rhs.first;
+      }
+    });
+
+    for (const auto &[Node, FlowFacts] : Cells) {
+      OS << "Instruction: " << NToString(Node);
+      OS << "\nFacts: ";
       if (FlowFacts.empty()) {
-        OS << "\tEMPTY\n";
+        OS << "EMPTY\n";
       } else {
         IMProblem.printContainer(OS, FlowFacts);
       }
-      OS << "\n\n";
+      OS << "\n";
     }
   }
 

include/phasar/PhasarLLVM/ControlFlow/Resolver/CHAResolver.h

@@ -31,7 +31,9 @@ class CHAResolver : public Resolver {
   CHAResolver(const LLVMProjectIRDB *IRDB, const LLVMVFTableProvider *VTP,
               const LLVMTypeHierarchy *TH);
 
-  ~CHAResolver() override = default;
+  // Deleting an incomplete type (LLVMTypeHierarchy) is UB, so instantiate the
+  // dtor in CHAResolver.cpp
+  ~CHAResolver() override;
 
   FunctionSetTy resolveVirtualCall(const llvm::CallBase *CallSite) override;
 

include/phasar/PhasarLLVM/ControlFlow/Resolver/Resolver.h

@@ -38,14 +38,29 @@ class LLVMVFTableProvider;
 class LLVMTypeHierarchy;
 enum class CallGraphAnalysisType;
 
+/// Assuming that `CallSite` is a virtual call through a vtable, retrieves the
+/// index in the vtable of the virtual function called.
 [[nodiscard]] std::optional<unsigned>
 getVFTIndex(const llvm::CallBase *CallSite);
 
+/// Assuming that `CallSite` is a vall to a non-static member function,
+/// retrieves the type of the receiver. Returns nullptr, if the receiver-type
+/// could not be extracted
 [[nodiscard]] const llvm::StructType *
 getReceiverType(const llvm::CallBase *CallSite);
 
+/// Assuming that `CallSite` is a virtual call, where `Idx` is retrieved through
+/// `getVFTIndex()` and `T` through `getReceiverType()`
+[[nodiscard]] const llvm::Function *
+getNonPureVirtualVFTEntry(const llvm::StructType *T, unsigned Idx,
+                          const llvm::CallBase *CallSite,
+                          const psr::LLVMVFTableProvider &VTP);
+
 [[nodiscard]] std::string getReceiverTypeName(const llvm::CallBase *CallSite);
 
+/// Checks whether the signature of `DestFun` matches the required withature of
+/// `CallSite`, such that `DestFun` qualifies as callee-candidate, if `CallSite`
+/// is an indirect/virtual call.
 [[nodiscard]] bool isConsistentCall(const llvm::CallBase *CallSite,
                                     const llvm::Function *DestFun);
 
@@ -59,7 +74,12 @@ class Resolver {
 
   const llvm::Function *
   getNonPureVirtualVFTEntry(const llvm::StructType *T, unsigned Idx,
-                            const llvm::CallBase *CallSite);
+                            const llvm::CallBase *CallSite) {
+    if (!VTP) {
+      return nullptr;
+    }
+    return psr::getNonPureVirtualVFTEntry(T, Idx, CallSite, *VTP);
+  }
 
 public:
   using FunctionSetTy = llvm::SmallDenseSet<const llvm::Function *, 4>;

include/phasar/PhasarLLVM/DB/LLVMProjectIRDB.h

@@ -18,7 +18,6 @@
 #include "llvm/ADT/DenseMap.h"
 #include "llvm/ADT/STLExtras.h"
 #include "llvm/ADT/StringRef.h"
-#include "llvm/ADT/iterator_range.h"
 #include "llvm/IR/Instruction.h"
 #include "llvm/IR/LLVMContext.h"
 #include "llvm/IR/Module.h"
@@ -44,7 +43,8 @@ class LLVMProjectIRDB : public ProjectIRDBBase<LLVMProjectIRDB> {
   /// Reads and parses the given LLVM IR file and owns the resulting IR Module.
   /// If an error occurs, an error message is written to stderr and subsequent
   /// calls to isValid() return false.
-  explicit LLVMProjectIRDB(const llvm::Twine &IRFileName);
+  explicit LLVMProjectIRDB(const llvm::Twine &IRFileName,
+                           bool EnableOpaquePointers = LLVM_VERSION_MAJOR > 14);
   /// Initializes the new ProjectIRDB with the given IR Module _without_ taking
   /// ownership. The module is optionally being preprocessed.
   ///
@@ -58,7 +58,8 @@ class LLVMProjectIRDB : public ProjectIRDBBase<LLVMProjectIRDB> {
   /// Parses the given LLVM IR file and owns the resulting IR Module.
   /// If an error occurs, an error message is written to stderr and subsequent
   /// calls to isValid() return false.
-  explicit LLVMProjectIRDB(llvm::MemoryBufferRef Buf);
+  explicit LLVMProjectIRDB(llvm::MemoryBufferRef Buf,
+                           bool EnableOpaquePointers = LLVM_VERSION_MAJOR > 14);
 
   LLVMProjectIRDB(const LLVMProjectIRDB &) = delete;
   LLVMProjectIRDB &operator=(LLVMProjectIRDB &) = delete;

include/phasar/PhasarLLVM/TaintConfig/TaintConfigData.h

@@ -18,7 +18,9 @@ namespace psr {
 enum class TaintCategory;
 
 struct FunctionData {
+#if __cplusplus < 202002L
   FunctionData() noexcept = default;
+#endif
 
   std::string Name;
   TaintCategory ReturnCat{};
@@ -29,7 +31,9 @@ struct FunctionData {
 };
 
 struct VariableData {
+#if __cplusplus < 202002L
   VariableData() noexcept = default;
+#endif
 
   size_t Line{};
   std::string Name;

lib/PhasarLLVM/ControlFlow/Resolver/CHAResolver.cpp

@@ -40,6 +40,8 @@ CHAResolver::CHAResolver(const LLVMProjectIRDB *IRDB,
   }
 }
 
+CHAResolver::~CHAResolver() = default;
+
 auto CHAResolver::resolveVirtualCall(const llvm::CallBase *CallSite)
     -> FunctionSetTy {
   PHASAR_LOG_LEVEL(DEBUG, "Call virtual function: ");