19337095106

Committed 13 Nov 2025 03:41PM UTC coverage: 72.7% (-10.7%) from 83.355%

Build # 19337095106

Build Type

push

github

Committed by

web-flow

Commit Message

Make TBAA optional (#45)

- No default TBAA usage
- LLVM 20 support

Run Details

1701 of 2973 branches covered (57.21%)

Branch coverage included in aggregate %.

207 of 213 new or added lines in 8 files covered. (97.18%)

56 existing lines in 6 files now uncovered.

2219 of 2419 relevant lines covered (91.73%)

7456.57 hits per line

Source File
Press 'n' to go to next uncovered line, 'b' for previous

82.51

/lib/type/DITypeExtractor.cpp


//  llvm-dimeta library
//  Copyright (c) 2022-2025 llvm-dimeta authors
//  Distributed under the BSD 3-Clause license.
//  (See accompanying file LICENSE)
//  SPDX-License-Identifier: BSD-3-Clause
//

#include "DIFinder.h"
#include "DIRootType.h"
#include "DIUtil.h"
#include "DataflowAnalysis.h"
#include "DefUseAnalysis.h"
#include "GEP.h"
#include "TBAA.h"
#include "Util.h"
#include "ValuePath.h"
#include "support/Logger.h"

#include "llvm/ADT/ArrayRef.h"
#include "llvm/ADT/STLExtras.h"
#include "llvm/ADT/SmallVector.h"
#include "llvm/ADT/ilist_iterator.h"
#include "llvm/BinaryFormat/Dwarf.h"
#include "llvm/Config/llvm-config.h"
#include "llvm/IR/Argument.h"
#include "llvm/IR/Constants.h"
#include "llvm/IR/DebugInfoMetadata.h"
#include "llvm/IR/Function.h"
#include "llvm/IR/GlobalVariable.h"
#include "llvm/IR/InstIterator.h"
#include "llvm/IR/InstrTypes.h"
#include "llvm/IR/Instructions.h"
#include "llvm/IR/IntrinsicInst.h"
#include "llvm/IR/Metadata.h"
#include "llvm/IR/Operator.h"
#include "llvm/IR/Value.h"
#include "llvm/Support/Casting.h"
#include "llvm/Support/Debug.h"
#include "llvm/Support/ErrorHandling.h"
#include "llvm/Support/raw_ostream.h"

#include <cassert>
#include <iterator>
#include <optional>
#include <type_traits>
#include <unordered_map>
#include <utility>

namespace dimeta::type {

namespace reset {

using GepToDIMemberMap = std::unordered_map<const llvm::GEPOperator*, llvm::DIDerivedType*>;

namespace detail {

template <typename T, typename InstTy>
std::optional<const T*> get_operand_to(const InstTy* memory_instruction) {
  static_assert(std::is_same_v<InstTy, llvm::StoreInst> || std::is_same_v<InstTy, llvm::LoadInst>,
                "Expected load or store instruction");
  const auto* memory_target = memory_instruction->getPointerOperand();
  if (llvm::isa<T>(memory_target)) {
    return llvm::dyn_cast<T>(memory_target);
  }

  if (auto bcast = llvm::dyn_cast<llvm::BitCastInst>(memory_target)) {
    if (llvm::isa<T>(bcast->getOperand(0))) {
      return llvm::dyn_cast<T>(bcast->getOperand(0));
    }
  }

  return {};
}

bool is_array_gep_with_non_const_indices(const llvm::GetElementPtrInst* gep) {
#if LLVM_VERSION_MAJOR > 12
  auto indices = gep->indices();
#else
  auto indices = llvm::make_range(gep->idx_begin(), gep->idx_end());
#endif
  for (const auto& index : indices) {
    if (!llvm::isa<llvm::ConstantInt>(index.get())) {
      return true;
    }
  }
  return false;
}

bool is_array_gep(const llvm::GetElementPtrInst* gep) {
  if (!gep) {
    return false;
  }
  return detail::is_array_gep_with_non_const_indices(gep) || gep->getSourceElementType()->isArrayTy();
}

}  // namespace detail

template <typename T>
bool store_to(const llvm::StoreInst* store) {
  return detail::get_operand_to<T>(store).has_value();
}

template <typename T>
bool load_to(const llvm::LoadInst* load) {
  return detail::get_operand_to<T>(load).has_value();
}

bool load_for_array_gep(const llvm::LoadInst* load) {
  for (const auto* user : load->users()) {
    if (const auto* gep = llvm::dyn_cast<llvm::GetElementPtrInst>(user)) {
      return detail::is_array_gep(gep);
    }
  }
  return false;
}

bool load_of_array_gep(const llvm::LoadInst* load) {
  return detail::is_array_gep(llvm::dyn_cast<llvm::GetElementPtrInst>(load->getPointerOperand()));
}

bool store_to_array_gep(const llvm::StoreInst* store) {
  auto gep = detail::get_operand_to<llvm::GetElementPtrInst>(store);
  if (!gep) {
    return false;
  }
  return detail::is_array_gep_with_non_const_indices(gep.value());
}

namespace dipath {

struct IRMapping {
  const llvm::Value* value{nullptr};
  llvm::DIType* mapped{nullptr};
  std::string reason;
};

struct ValueToDiPath {
  llvm::SmallVector<IRMapping, 8> path_to_ditype;

  void emplace_back(const llvm::Value* val, llvm::DIType* mapped_di_type, const std::string reason = "") {
    path_to_ditype.emplace_back(IRMapping{val, mapped_di_type, std::move(reason)});
  }

  std::optional<llvm::DIType*> final_type() const {
    if (path_to_ditype.empty()) {
      return {};
    }
    const auto& ditype = path_to_ditype.back();
    return ditype.mapped != nullptr ? std::optional{ditype.mapped} : std::nullopt;
  }
};

llvm::raw_ostream& operator<<(llvm::raw_ostream& os, const ValueToDiPath& vdp) {
#if DIMETA_LOG_LEVEL > 2  // FIXME: For coverage
  const auto& mappings = vdp.path_to_ditype;
  // os << "ValueToDiPath: ";  // Prefix to identify the type being printed
  if (mappings.empty()) {
    os << "[]";
    return os;
  }
  const auto mapping_to_string = [](const IRMapping& mapping) -> std::string {
    std::string str_buffer;
    llvm::raw_string_ostream stream(str_buffer);

    stream << "{IR: ";
    if (mapping.value) {
      // mapping.value->printAsOperand(stream, true);
      mapping.value->print(stream, true);
    } else {
      stream << "null";
    }

    stream << "; DI: ";
    if (mapping.mapped) {
      stream << log::ditype_str(mapping.mapped);
    } else {
      stream << "null";
    }

    if (!mapping.reason.empty()) {
      stream << ", Reason: \"" << mapping.reason << "\"}";
    }
    return stream.str();
  };

  os << "[" << mapping_to_string(mappings.front());
  std::for_each(std::next(mappings.begin()), mappings.end(), [&](const IRMapping& mapping_item) {
    os << " --> ";
    os << mapping_to_string(mapping_item);
  });

  os << "]";
#endif
  return os;
}

}  // namespace dipath

std::optional<llvm::DIType*> reset_load_related_basic(const dataflow::ValuePath& path, llvm::DIType* type_to_reset,
                                                      const llvm::LoadInst* load) {
  auto* type = type_to_reset;

  if (load_to<llvm::GlobalVariable>(load) || load_to<llvm::AllocaInst>(load)) {
    LOG_DEBUG("Do not reset DIType based on load to global,alloca")
    return type;
  }

  if (di::util::is_array_member(*type)) {
    auto* base_type = llvm::dyn_cast<llvm::DIDerivedType>(type)->getBaseType();
    LOG_DEBUG("Load of array-like " << log::ditype_str(base_type))
    // auto type_de = di::util::desugar(*base_type, 1);
    if (auto underlying_type = llvm::dyn_cast<llvm::DICompositeType>(base_type)->getBaseType()) {
      return underlying_type;
    }
  }

  const auto try_resolve_to_first_member = [&](llvm::DIType* candidate_type) -> std::optional<llvm::DIType*> {
    auto comp = di::util::desugar(*candidate_type);
    LOG_DEBUG("Desugared load to " << log::ditype_str(comp.value_or(nullptr)));
    if (!comp) {
      return {};
    }
    LOG_DEBUG("Loading first pointer member?");
    auto result = di::util::resolve_byte_offset_to_member_of(comp.value(), 0);
    if (result) {
      LOG_DEBUG("Return type of load " << log::ditype_str(result->type_of_member.value_or(nullptr)));
      return result->type_of_member;
    }
    return {};
  };

  // a (last?) load to a GEP of a composite likely loads the first member in an optimized context:
  const bool last_load     = path.start_value().value_or(nullptr) == load;
  const bool is_not_member = !di::util::is_member(*type);
  if (is_not_member || last_load) {
    const bool is_not_arg_load  = !load_to<llvm::Argument>(load);
    const bool is_not_array_gep = !load_of_array_gep(load) && !load_for_array_gep(load);
    if (is_not_array_gep && (is_not_arg_load || last_load)) {
      if (auto resolved = try_resolve_to_first_member(type)) {
        return resolved.value();
      }
    }
  }

  if (auto* maybe_ptr_to_type = llvm::dyn_cast<llvm::DIDerivedType>(type)) {
    if (di::util::is_pointer(*maybe_ptr_to_type)) {
      LOG_DEBUG("Load of pointer-like " << log::ditype_str(maybe_ptr_to_type))
    }
    auto* base_type = maybe_ptr_to_type->getBaseType();

#if DIMETA_USE_TBAA == 1
    if (auto* composite = llvm::dyn_cast<llvm::DICompositeType>(base_type)) {
      LOG_DEBUG("Have ptr to composite " << log::ditype_str(composite))
      auto type_tbaa = tbaa::resolve_tbaa(base_type, *load);
      if (type_tbaa) {
        return type_tbaa;
      }
    }
#endif
    return base_type;
  }

  return type;
}

std::optional<llvm::DIType*> reset_store_related_basic(const dataflow::ValuePath&, llvm::DIType* type_to_reset,
                                                       const llvm::StoreInst* store_inst) {
  auto* type = type_to_reset;

  if (store_to<llvm::GlobalVariable>(store_inst) || store_to<llvm::AllocaInst>(store_inst)) {
    // Relevant in "heap_lulesh_mock_char.cpp"
    LOG_DEBUG("Store to alloca/global, return " << log::ditype_str(type))
    return type;
  }

  if (di::util::is_array(*type)) {
    return llvm::cast<llvm::DICompositeType>(type)->getBaseType();
  }

  if (!di::util::is_array_member(*type)) {
    // !di::util::is_array(*type) &&
    const auto is_non_pointer_member = [&](auto& type_) {
      return di::util::is_non_static_member(type_) &&
             !di::util::is_pointer_like(*llvm::cast<llvm::DIDerivedType>(&type_)->getBaseType());
    };

    auto desugared_composite = di::util::desugar(*type);
    LOG_DEBUG("Desugared " << log::ditype_str(desugared_composite.value_or(nullptr)))

    const bool is_load_target     = store_to<llvm::LoadInst>(store_inst);
    const bool is_arg_target      = store_to<llvm::Argument>(store_inst);
    const bool is_array_gep_store = store_to_array_gep(store_inst);
    const bool is_non_ptr_member  = is_non_pointer_member(*type);

    if (desugared_composite && (is_non_ptr_member || is_array_gep_store || is_arg_target || is_load_target)) {
      LOG_DEBUG("Storing to first pointer member?")
      auto result = di::util::resolve_byte_offset_to_member_of(desugared_composite.value(), 0);
      if (result) {
#if DIMETA_USE_TBAA == 1
        if (result->member) {
          const auto member_name = result->member.value()->getName();
          const bool is_vptr     = dimeta::util::starts_with_any_of(member_name, "_vptr");
          // Let this be handled by TBAA is available, see test 10_lulesh_ad_tbaa_static_member.ll
          if (is_vptr) {
            return type;
          }
        }
#endif
        LOG_DEBUG("Return type of store " << log::ditype_str(result->type_of_member.value_or(nullptr)))
        return result->type_of_member;
      }
    }
  }

  if (!llvm::isa<llvm::DIDerivedType>(type)) {
    LOG_DEBUG("Store resolved, return " << log::ditype_str(type))
    return type;
  }

  auto* derived_type = llvm::cast<llvm::DIDerivedType>(type);

  if (di::util::is_array_member(*type)) {
    auto* member_base               = derived_type->getBaseType();
    const bool is_array_type_member = member_base->getTag() == llvm::dwarf::DW_TAG_array_type;
    if (is_array_type_member) {
      return llvm::cast<llvm::DICompositeType>(member_base)->getBaseType();
    }
  }

  if (di::util::is_non_static_member(*derived_type)) {
    auto* member_base = derived_type->getBaseType();
    return member_base;
  }

  if (di::util::is_pointer(*derived_type)) {
    if (auto* may_be_ptr_to_ptr = llvm::dyn_cast<llvm::DIDerivedType>(derived_type->getBaseType())) {
      // Pointer to pointer by default remove one level for RHS assignment type w.r.t. store:
      const auto is_ptr_to_ptr = di::util::is_pointer(*may_be_ptr_to_ptr);
      if (is_ptr_to_ptr) {
        LOG_DEBUG("Store to ptr-ptr, return " << log::ditype_str(may_be_ptr_to_ptr))
        return may_be_ptr_to_ptr;
      }
    }
  }

  LOG_DEBUG("Store resolved, return " << log::ditype_str(type))
  return type;
}

template <typename Iter>
std::optional<llvm::DIType*> reset_ditype(llvm::DIType* type_to_reset, const dataflow::ValuePath& path,
                                          const Iter& path_iter, dipath::ValueToDiPath& logged_dipath) {
  std::optional<llvm::DIType*> type = type_to_reset;

  const auto& current_value = path_iter;
  LOG_DEBUG("Type to reset: " << log::ditype_str(*type));
  LOG_DEBUG(">> based on IR: " << **current_value);

  if (llvm::isa<llvm::GEPOperator>(*current_value)) {
    LOG_DEBUG("Reset based on GEP")
    auto* gep             = llvm::cast<llvm::GEPOperator>(*current_value);
    const auto gep_result = gep::extract_gep_dereferenced_type(type.value(), *gep);
    if (gep_result.member && !gep_result.use_type) {
      LOG_DEBUG("Using gep member type result")
      type = gep_result.member;
    } else {
      type = gep_result.type;
    }
  } else if (const auto* load = llvm::dyn_cast<llvm::LoadInst>(*current_value)) {
    LOG_DEBUG("Reset based on load")
    type = reset::reset_load_related_basic(path, type.value(), load);
  } else if (const auto* store_inst = llvm::dyn_cast<llvm::StoreInst>(*current_value)) {
    LOG_DEBUG("Reset based on store")
    type = reset::reset_store_related_basic(path, type.value(), store_inst);
  } else {
    LOG_DEBUG(">> skipping");
    logged_dipath.emplace_back(*current_value, type.value_or(nullptr));
    return type;
  }

  logged_dipath.emplace_back(*current_value, type.value_or(nullptr));

  return type;
}

}  // namespace reset

std::optional<llvm::DIType*> find_type(const dataflow::CallValuePath& call_path) {
  auto type = root::find_type_root(call_path);

  if (!type) {
    LOG_DEBUG("find_type_root failed to find a type for path " << call_path.path)
    return {};
  }

  reset::dipath::ValueToDiPath dipath;

  const auto path_end = call_path.path.path_to_value.rend();
  for (auto path_iter = call_path.path.path_to_value.rbegin(); path_iter != path_end; ++path_iter) {
    LOG_DEBUG("Extracted type: " << log::ditype_str(*type));
    type = reset::reset_ditype(type.value(), call_path.path, path_iter, dipath).value_or(type.value());
    LOG_DEBUG("reset_ditype result " << log::ditype_str(type.value_or(nullptr)) << "\n")
    if (!type) {
      break;
    }
  }

#if DIMETA_USE_TBAA == 1
  if (type) {
    // If last node is a store inst, try to extract type via TBAA
    const auto* const start_node = llvm::dyn_cast_or_null<llvm::StoreInst>(*call_path.path.start_value());
    if (start_node) {
      auto type_tbaa = tbaa::resolve_tbaa(type.value(), *llvm::dyn_cast<llvm::Instruction>(start_node));
      if (type_tbaa) {
        dipath.emplace_back(start_node, type_tbaa.value(), "TBAA");
      }
    }
  }
#endif

  LOG_DEBUG("Final mapping\n" << dipath)

  return dipath.final_type();
}
}  // namespace dimeta::type

1
2	// llvm-dimeta library
3	// Copyright (c) 2022-2025 llvm-dimeta authors
4	// Distributed under the BSD 3-Clause license.
5	// (See accompanying file LICENSE)
6	// SPDX-License-Identifier: BSD-3-Clause
7	//
8
9	#include "DIFinder.h"
10	#include "DIRootType.h"
11	#include "DIUtil.h"
12	#include "DataflowAnalysis.h"
13	#include "DefUseAnalysis.h"
14	#include "GEP.h"
15	#include "TBAA.h"
16	#include "Util.h"
17	#include "ValuePath.h"
18	#include "support/Logger.h"
19
20	#include "llvm/ADT/ArrayRef.h"
21	#include "llvm/ADT/STLExtras.h"
22	#include "llvm/ADT/SmallVector.h"
23	#include "llvm/ADT/ilist_iterator.h"
24	#include "llvm/BinaryFormat/Dwarf.h"
25	#include "llvm/Config/llvm-config.h"
26	#include "llvm/IR/Argument.h"
27	#include "llvm/IR/Constants.h"
28	#include "llvm/IR/DebugInfoMetadata.h"
29	#include "llvm/IR/Function.h"
30	#include "llvm/IR/GlobalVariable.h"
31	#include "llvm/IR/InstIterator.h"
32	#include "llvm/IR/InstrTypes.h"
33	#include "llvm/IR/Instructions.h"
34	#include "llvm/IR/IntrinsicInst.h"
35	#include "llvm/IR/Metadata.h"
36	#include "llvm/IR/Operator.h"
37	#include "llvm/IR/Value.h"
38	#include "llvm/Support/Casting.h"
39	#include "llvm/Support/Debug.h"
40	#include "llvm/Support/ErrorHandling.h"
41	#include "llvm/Support/raw_ostream.h"
42
43	#include <cassert>
44	#include <iterator>
45	#include <optional>
46	#include <type_traits>
47	#include <unordered_map>
48	#include <utility>
49
50	namespace dimeta::type {
51
52	namespace reset {
53
54	using GepToDIMemberMap = std::unordered_map<const llvm::GEPOperator, llvm::DIDerivedType>;
55
56	namespace detail {
57
58	template <typename T, typename InstTy>
59	std::optional<const T> get_operand_to(const InstTy memory_instruction) {	11,832✔
60	static_assert(std::is_same_v<InstTy, llvm::StoreInst> \|\| std::is_same_v<InstTy, llvm::LoadInst>,
61	"Expected load or store instruction");
62	const auto* memory_target = memory_instruction->getPointerOperand();	11,832✔
63	if (llvm::isa<T>(memory_target)) {	11,832✔
64	return llvm::dyn_cast<T>(memory_target);	2,655✔
65	}
66
67	if (auto bcast = llvm::dyn_cast<llvm::BitCastInst>(memory_target)) {	9,823!
68	if (llvm::isa<T>(bcast->getOperand(0))) {	646!
69	return llvm::dyn_cast<T>(bcast->getOperand(0));	106✔
70	}
71	}	540✔
72
73	return {};	9,071✔
74	}	11,832✔
75
76	bool is_array_gep_with_non_const_indices(const llvm::GetElementPtrInst* gep) {	1,246✔
77	#if LLVM_VERSION_MAJOR > 12
78	auto indices = gep->indices();	1,246✔
79	#else
80	auto indices = llvm::make_range(gep->idx_begin(), gep->idx_end());
81	#endif
82	for (const auto& index : indices) {	3,112✔
83	if (!llvm::isa<llvm::ConstantInt>(index.get())) {	1,866✔
84	return true;	366✔
85	}
86	}	1,866✔
87	return false;	880✔
88	}	1,246✔
89
90	bool is_array_gep(const llvm::GetElementPtrInst* gep) {	474✔
91	if (!gep) {	474✔
92	return false;	106✔
93	}
94	return detail::is_array_gep_with_non_const_indices(gep) \|\| gep->getSourceElementType()->isArrayTy();	368✔
95	}	474✔
96
97	} // namespace detail
98
99	template <typename T>
100	bool store_to(const llvm::StoreInst* store) {	6,984✔
101	return detail::get_operand_to<T>(store).has_value();	6,984✔
102	}
103
104	template <typename T>
105	bool load_to(const llvm::LoadInst* load) {	3,420✔
106	return detail::get_operand_to<T>(load).has_value();	3,420✔
107	}
108
109	bool load_for_array_gep(const llvm::LoadInst* load) {	184✔
110	for (const auto* user : load->users()) {	368!
111	if (const auto* gep = llvm::dyn_cast<llvm::GetElementPtrInst>(user)) {	266✔
112	return detail::is_array_gep(gep);	82✔
113	}
114	}	184✔
115	return false;	102✔
116	}	184✔
117
118	bool load_of_array_gep(const llvm::LoadInst* load) {	392✔
119	return detail::is_array_gep(llvm::dyn_cast<llvm::GetElementPtrInst>(load->getPointerOperand()));	392✔
120	}
121
122	bool store_to_array_gep(const llvm::StoreInst* store) {	1,428✔
123	auto gep = detail::get_operand_to<llvm::GetElementPtrInst>(store);	1,428✔
124	if (!gep) {	1,428✔
125	return false;	550✔
126	}
127	return detail::is_array_gep_with_non_const_indices(gep.value());	878✔
128	}	1,428✔
129
130	namespace dipath {
131
132	struct IRMapping {
133	const llvm::Value* value{nullptr};
134	llvm::DIType* mapped{nullptr};
135	std::string reason;
136	};
137
138	struct ValueToDiPath {
139	llvm::SmallVector<IRMapping, 8> path_to_ditype;
140
141	void emplace_back(const llvm::Value* val, llvm::DIType* mapped_di_type, const std::string reason = "") {	9,711✔
142	path_to_ditype.emplace_back(IRMapping{val, mapped_di_type, std::move(reason)});	9,711!
143	}	9,711✔
144
145	std::optional<llvm::DIType*> final_type() const {	2,576✔
146	if (path_to_ditype.empty()) {	2,576!
NEW 147	return {};	×
148	}
149	const auto& ditype = path_to_ditype.back();	2,576✔
150	return ditype.mapped != nullptr ? std::optional{ditype.mapped} : std::nullopt;	2,576!
151	}	2,576✔
152	};
153
NEW 154	llvm::raw_ostream& operator<<(llvm::raw_ostream& os, const ValueToDiPath& vdp) {	×
155	#if DIMETA_LOG_LEVEL > 2 // FIXME: For coverage
156	const auto& mappings = vdp.path_to_ditype;
157	// os << "ValueToDiPath: "; // Prefix to identify the type being printed
158	if (mappings.empty()) {
159	os << "[]";
160	return os;
161	}
162	const auto mapping_to_string = [](const IRMapping& mapping) -> std::string {
163	std::string str_buffer;
164	llvm::raw_string_ostream stream(str_buffer);
165
166	stream << "{IR: ";
167	if (mapping.value) {
168	// mapping.value->printAsOperand(stream, true);
169	mapping.value->print(stream, true);
170	} else {
171	stream << "null";
172	}
173
174	stream << "; DI: ";
175	if (mapping.mapped) {
176	stream << log::ditype_str(mapping.mapped);
177	} else {
178	stream << "null";
179	}
180
181	if (!mapping.reason.empty()) {
182	stream << ", Reason: \"" << mapping.reason << "\"}";
183	}
184	return stream.str();
185	};
186
187	os << "[" << mapping_to_string(mappings.front());
188	std::for_each(std::next(mappings.begin()), mappings.end(), [&](const IRMapping& mapping_item) {
189	os << " --> ";
190	os << mapping_to_string(mapping_item);
191	});
192
193	os << "]";
194	#endif
NEW 195	return os;	×
196	}
197
198	} // namespace dipath
199
200	std::optional<llvm::DIType> reset_load_related_basic(const dataflow::ValuePath& path, llvm::DIType type_to_reset,	1,595✔
201	const llvm::LoadInst* load) {
202	auto* type = type_to_reset;	1,595✔
203
204	if (load_to<llvm::GlobalVariable>(load) \|\| load_to<llvm::AllocaInst>(load)) {	1,595✔
205	LOG_DEBUG("Do not reset DIType based on load to global,alloca")
206	return type;	1,041✔
207	}
208
209	if (di::util::is_array_member(*type)) {	554✔
210	auto* base_type = llvm::dyn_cast<llvm::DIDerivedType>(type)->getBaseType();	108✔
211	LOG_DEBUG("Load of array-like " << log::ditype_str(base_type))
212	// auto type_de = di::util::desugar(*base_type, 1);
213	if (auto underlying_type = llvm::dyn_cast<llvm::DICompositeType>(base_type)->getBaseType()) {	216!
214	return underlying_type;	108✔
215	}
216	}	108!
217
218	const auto try_resolve_to_first_member = [&](llvm::DIType* candidate_type) -> std::optional<llvm::DIType*> {	540✔
219	auto comp = di::util::desugar(*candidate_type);	94✔
220	LOG_DEBUG("Desugared load to " << log::ditype_str(comp.value_or(nullptr)));
221	if (!comp) {	94✔
222	return {};	76✔
223	}
224	LOG_DEBUG("Loading first pointer member?");
225	auto result = di::util::resolve_byte_offset_to_member_of(comp.value(), 0);	18✔
226	if (result) {	18!
227	LOG_DEBUG("Return type of load " << log::ditype_str(result->type_of_member.value_or(nullptr)));
228	return result->type_of_member;	18✔
229	}
NEW 230	return {};	×
231	};	94✔
232
233	// a (last?) load to a GEP of a composite likely loads the first member in an optimized context:
234	const bool last_load = path.start_value().value_or(nullptr) == load;	446✔
235	const bool is_not_member = !di::util::is_member(*type);	446✔
236	if (is_not_member \|\| last_load) {	446✔
237	const bool is_not_arg_load = !load_to<llvm::Argument>(load);	392✔
238	const bool is_not_array_gep = !load_of_array_gep(load) && !load_for_array_gep(load);	392✔
239	if (is_not_array_gep && (is_not_arg_load \|\| last_load)) {	392!
240	if (auto resolved = try_resolve_to_first_member(type)) {	112✔
241	return resolved.value();	18✔
242	}
243	}	76✔
244	}	392✔
245
246	if (auto* maybe_ptr_to_type = llvm::dyn_cast<llvm::DIDerivedType>(type)) {	856!
247	if (di::util::is_pointer(*maybe_ptr_to_type)) {	428✔
248	LOG_DEBUG("Load of pointer-like " << log::ditype_str(maybe_ptr_to_type))
249	}	320✔
250	auto* base_type = maybe_ptr_to_type->getBaseType();	428✔
251
252	#if DIMETA_USE_TBAA == 1
253	if (auto* composite = llvm::dyn_cast<llvm::DICompositeType>(base_type)) {	162✔
254	LOG_DEBUG("Have ptr to composite " << log::ditype_str(composite))
255	auto type_tbaa = tbaa::resolve_tbaa(base_type, *load);	18✔
256	if (type_tbaa) {	18✔
257	return type_tbaa;	12✔
258	}
259	}	6✔
260	#endif
261	return base_type;	416✔
262	}	428✔
263
264	return type;	×
265	}	1,595✔
266
267	std::optional<llvm::DIType> reset_store_related_basic(const dataflow::ValuePath&, llvm::DIType type_to_reset,	2,080✔
268	const llvm::StoreInst* store_inst) {
269	auto* type = type_to_reset;	2,080✔
270
271	if (store_to<llvm::GlobalVariable>(store_inst) \|\| store_to<llvm::AllocaInst>(store_inst)) {	2,080✔
272	// Relevant in "heap_lulesh_mock_char.cpp"
273	LOG_DEBUG("Store to alloca/global, return " << log::ditype_str(type))
274	return type;	274✔
275	}
276
277	if (di::util::is_array(*type)) {	1,806✔
278	return llvm::cast<llvm::DICompositeType>(type)->getBaseType();	90✔
279	}
280
281	if (!di::util::is_array_member(*type)) {	1,716✔
282	// !di::util::is_array(*type) &&
283	const auto is_non_pointer_member = [&](auto& type_) {	2,856✔
284	return di::util::is_non_static_member(type_) &&	1,992✔
285	!di::util::is_pointer_like(*llvm::cast<llvm::DIDerivedType>(&type_)->getBaseType());	564✔
286	};
287
288	auto desugared_composite = di::util::desugar(*type);	1,428✔
289	LOG_DEBUG("Desugared " << log::ditype_str(desugared_composite.value_or(nullptr)))
290
291	const bool is_load_target = store_to<llvm::LoadInst>(store_inst);	1,428✔
292	const bool is_arg_target = store_to<llvm::Argument>(store_inst);	1,428✔
293	const bool is_array_gep_store = store_to_array_gep(store_inst);	1,428✔
294	const bool is_non_ptr_member = is_non_pointer_member(*type);	1,428✔
295
296	if (desugared_composite && (is_non_ptr_member \|\| is_array_gep_store \|\| is_arg_target \|\| is_load_target)) {	1,428✔
297	LOG_DEBUG("Storing to first pointer member?")
298	auto result = di::util::resolve_byte_offset_to_member_of(desugared_composite.value(), 0);	220✔
299	if (result) {	220✔
300	#if DIMETA_USE_TBAA == 1
301	if (result->member) {	74!
302	const auto member_name = result->member.value()->getName();	74✔
303	const bool is_vptr = dimeta::util::starts_with_any_of(member_name, "_vptr");	74✔
304	// Let this be handled by TBAA is available, see test 10_lulesh_ad_tbaa_static_member.ll
305	if (is_vptr) {	74✔
306	return type;	6✔
307	}
308	}	74✔
309	#endif
310	LOG_DEBUG("Return type of store " << log::ditype_str(result->type_of_member.value_or(nullptr)))
311	return result->type_of_member;	196✔
312	}
313	}	220✔
314	}	1,428✔
315
316	if (!llvm::isa<llvm::DIDerivedType>(type)) {	1,514✔
317	LOG_DEBUG("Store resolved, return " << log::ditype_str(type))
318	return type;	12✔
319	}
320
321	auto* derived_type = llvm::cast<llvm::DIDerivedType>(type);	1,502✔
322
323	if (di::util::is_array_member(*type)) {	1,502✔
324	auto* member_base = derived_type->getBaseType();	288✔
325	const bool is_array_type_member = member_base->getTag() == llvm::dwarf::DW_TAG_array_type;	288✔
326	if (is_array_type_member) {	288!
327	return llvm::cast<llvm::DICompositeType>(member_base)->getBaseType();	288✔
328	}
329	}	288!
330
331	if (di::util::is_non_static_member(*derived_type)) {	1,214✔
332	auto* member_base = derived_type->getBaseType();	558✔
333	return member_base;	558✔
334	}	558✔
335
336	if (di::util::is_pointer(*derived_type)) {	656✔
337	if (auto* may_be_ptr_to_ptr = llvm::dyn_cast<llvm::DIDerivedType>(derived_type->getBaseType())) {	1,222✔
338	// Pointer to pointer by default remove one level for RHS assignment type w.r.t. store:
339	const auto is_ptr_to_ptr = di::util::is_pointer(*may_be_ptr_to_ptr);	584✔
340	if (is_ptr_to_ptr) {	584!
341	LOG_DEBUG("Store to ptr-ptr, return " << log::ditype_str(may_be_ptr_to_ptr))
342	return may_be_ptr_to_ptr;	584✔
343	}
344	}	584!
345	}	54✔
346
347	LOG_DEBUG("Store resolved, return " << log::ditype_str(type))
348	return type;	72✔
349	}	2,080✔
350
351	template <typename Iter>
352	std::optional<llvm::DIType> reset_ditype(llvm::DIType type_to_reset, const dataflow::ValuePath& path,	9,015✔
353	const Iter& path_iter, dipath::ValueToDiPath& logged_dipath) {
354	std::optional<llvm::DIType*> type = type_to_reset;	9,015✔
355
356	const auto& current_value = path_iter;	9,015✔
357	LOG_DEBUG("Type to reset: " << log::ditype_str(*type));
358	LOG_DEBUG(">> based on IR: " << **current_value);
359
360	if (llvm::isa<llvm::GEPOperator>(*current_value)) {	9,015✔
361	LOG_DEBUG("Reset based on GEP")
362	auto* gep = llvm::cast<llvm::GEPOperator>(*current_value);	2,244✔
363	const auto gep_result = gep::extract_gep_dereferenced_type(type.value(), *gep);	2,244✔
364	if (gep_result.member && !gep_result.use_type) {	2,244!
365	LOG_DEBUG("Using gep member type result")
366	type = gep_result.member;	1,354✔
367	} else {	1,354✔
368	type = gep_result.type;	890✔
369	}
370	} else if (const auto* load = llvm::dyn_cast<llvm::LoadInst>(*current_value)) {	15,786✔
371	LOG_DEBUG("Reset based on load")
372	type = reset::reset_load_related_basic(path, type.value(), load);	1,595✔
373	} else if (const auto* store_inst = llvm::dyn_cast<llvm::StoreInst>(*current_value)) {	11,947✔
374	LOG_DEBUG("Reset based on store")
375	type = reset::reset_store_related_basic(path, type.value(), store_inst);	2,080✔
376	} else {	2,080✔
377	LOG_DEBUG(">> skipping");
378	logged_dipath.emplace_back(*current_value, type.value_or(nullptr));	3,096!
379	return type;	3,096✔
380	}
381
382	logged_dipath.emplace_back(*current_value, type.value_or(nullptr));	5,919!
383
384	return type;	5,919✔
385	}	9,015✔
386
387	} // namespace reset
388
389	std::optional<llvm::DIType*> find_type(const dataflow::CallValuePath& call_path) {	2,602✔
390	auto type = root::find_type_root(call_path);	2,602✔
391
392	if (!type) {	2,602✔
393	LOG_DEBUG("find_type_root failed to find a type for path " << call_path.path)
394	return {};	26✔
395	}
396
397	reset::dipath::ValueToDiPath dipath;	2,576✔
398
399	const auto path_end = call_path.path.path_to_value.rend();	2,576!
400	for (auto path_iter = call_path.path.path_to_value.rbegin(); path_iter != path_end; ++path_iter) {	11,591!
401	LOG_DEBUG("Extracted type: " << log::ditype_str(*type));
402	type = reset::reset_ditype(type.value(), call_path.path, path_iter, dipath).value_or(type.value());	9,015!
403	LOG_DEBUG("reset_ditype result " << log::ditype_str(type.value_or(nullptr)) << "\n")
404	if (!type) {	9,015!
405	break;	×
406	}
407	}	9,015✔
408
409	#if DIMETA_USE_TBAA == 1
410	if (type) {	872!
411	// If last node is a store inst, try to extract type via TBAA
412	const auto* const start_node = llvm::dyn_cast_or_null<llvm::StoreInst>(*call_path.path.start_value());	872!
413	if (start_node) {	872✔
414	auto type_tbaa = tbaa::resolve_tbaa(type.value(), *llvm::dyn_cast<llvm::Instruction>(start_node));	708!
415	if (type_tbaa) {	708✔
416	dipath.emplace_back(start_node, type_tbaa.value(), "TBAA");	696!
417	}	696✔
418	}	708✔
419	}	872✔
420	#endif
421
422	LOG_DEBUG("Final mapping\n" << dipath)
423
424	return dipath.final_type();	2,576!
425	}	2,602✔
426	} // namespace dimeta::type

ahueck / llvm-dimeta / 19337095106

Source File Press 'n' to go to next uncovered line, 'b' for previous

Source File
Press 'n' to go to next uncovered line, 'b' for previous