13955121336

Committed 19 Mar 2025 07:30PM UTC coverage: 88.842%. First build

Build # 13955121336

Build Type

push

github

Committed by

web-flow

Commit Message

Merge PR #167 from tudasc/devel

Run Details

1174 of 1361 new or added lines in 49 files covered. (86.26%)

4212 of 4741 relevant lines covered (88.84%)

261950.04 hits per line

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

73.95

/lib/passes/filter/CGInterface.cpp

// TypeART library
//
// Copyright (c) 2017-2025 TypeART Authors
// Distributed under the BSD 3-Clause license.
// (See accompanying file LICENSE.txt or copy at
// https://opensource.org/licenses/BSD-3-Clause)
//
// Project home: https://github.com/tudasc/TypeART
//
// SPDX-License-Identifier: BSD-3-Clause
//

#include "CGInterface.h"

#include "support/Logger.h"
#include "support/Util.h"

#include "llvm/Support/Error.h"
#include "llvm/Support/ErrorOr.h"
#include "llvm/Support/JSON.h"
#include "llvm/Support/MemoryBuffer.h"
#include "llvm/Support/raw_ostream.h"

#include <algorithm>
#include <cassert>
#include <cstdlib>

namespace typeart::filter {

CGInterface::ReachabilityResult JSONCG::reachable(const std::string& source, const std::string& target,
                                                  bool case_sensitive, bool short_circuit) {
  const auto reachables = get_reachable_functions(source);
  bool matches          = false;
  bool allBodies        = true;

  // bool what{false};
  if (!hasBodyMap[source]) {
    ++no_call_chain;
  }

  if (!reachables.empty()) {
    ++call_chain;
    f.push_back(source);
  }

  for (const auto& function_n : reachables) {
    matches |= util::regex_matches(target, function_n, case_sensitive);
    allBodies = allBodies && hasBodyMap[target];

    if (matches && short_circuit) {
      // we have found a match -> whether all functions have bodies is irrelevant
      return ReachabilityResult::reaches;
    }
  }

  if (matches) {
    // matches but no short circuit
    return ReachabilityResult::reaches;
  }

  if (!matches && (!allBodies || !hasBodyMap[source])) {
    // We did not find a match, but not all functions had bodies, we don't know
    return ReachabilityResult::maybe_reaches;
  }

  // No match and all functions had bodies -> never reaches (all call targets found)
  return ReachabilityResult::never_reaches;
}

std::vector<std::string> JSONCG::get_decl_only() {
  std::vector<std::string> list;
  list.reserve(hasBodyMap.size());
  for (const auto& [func, has_body] : hasBodyMap) {
    if (!has_body) {
      list.push_back(func);
    }
  }
  return list;
}

std::unordered_set<std::string> JSONCG::get_reachable_functions(const std::string& caller,
                                                                bool considerOverrides) const {
  std::unordered_set<std::string> ret;
  std::unordered_set<std::string> worklist;

  worklist = get_directly_called_function_names(caller, considerOverrides);
  while (!worklist.empty()) {
    const std::string func_name = *worklist.begin();
    // Check if we did not already handled it
    if (ret.find(func_name) == ret.end()) {
      worklist.merge(get_directly_called_function_names(func_name));
      ret.insert(func_name);
    }
    worklist.erase(worklist.find(func_name));  // Iterators get invalidated by merge, so we need to search again
  }
  return ret;
}

std::unordered_set<std::string> JSONCG::get_directly_called_function_names(const std::string& caller,
                                                                           bool considerOverrides) const {
  auto ref = directly_called_functions.find(caller);
  if (ref != std::end(directly_called_functions)) {
    // If the caller is virtual and overridden, add the overriding functions
    if (considerOverrides && (virtualTargets.find(caller) != std::end(virtualTargets))) {
      auto targets  = ref->second;
      auto vTargets = virtualTargets.find(caller)->second;
      targets.merge(vTargets);
      return targets;
    }
    return ref->second;
  }
  return {};
}

JSONCG::JSONCG(const llvm::json::Value& cg) {
  // Expected json format is the following:
  // A top level object/map Key is the function name, value is an object/map with information
  // We only care about "callees"
  // callees itself is an array with function names (as strings)
  assert(cg.kind() == llvm::json::Value::Kind::Object && "Top level json must be an Object");
  const llvm::json::Object* tlobj = cg.getAsObject();
  if (tlobj != nullptr) {
    for (const auto& entry : *tlobj) {
      // std::cout << "Building call site info for " << entry.first.str() << std::endl;
      construct_call_information(entry.first.str(), *tlobj);
    }
  }
}

void JSONCG::construct_call_information(const std::string& entry_caller, const llvm::json::Object& j) {
  if (directly_called_functions.find(entry_caller) == directly_called_functions.end()) {
    // We did not handle this function yet
    directly_called_functions[entry_caller] = std::unordered_set<std::string>();
    const auto caller                       = j.getObject(entry_caller);
    if (caller != nullptr) {
      const auto hasBody = caller->get("hasBody");
      if (hasBody != nullptr) {
        assert(hasBody->kind() == llvm::json::Value::Kind::Boolean && "hasBody must be boolean");
        hasBodyMap[entry_caller] = *hasBody->getAsBoolean();
      }
      const auto calls = caller->getArray("callees");
      assert(calls != nullptr && "Json callee information is missing");
      if (calls != nullptr) {
        // Now iterate over them
        for (const auto& callee : *calls) {
          assert(callee.kind() == llvm::json::Value::Kind::String && "Callees must be strings");
          const auto callee_json_string = callee.getAsString();
          assert(callee_json_string && "Could not get callee as string");
          if (callee_json_string) {
            const std::string callee_string = std::string{*callee_json_string};
            directly_called_functions[entry_caller].insert(callee_string);
          }
        }
      }
      // if the function is virtual, overriding functions are _potential_ call targets
      const auto overridingFunctions = caller->getArray("overriddenBy");
      if (overridingFunctions != nullptr) {
        for (const auto& function : *overridingFunctions) {
          assert(function.kind() == llvm::json::Value::Kind::String && "Function names are always strings");
          const auto functionStr = function.getAsString();
          assert(functionStr && "Retrieving overriding function as String failed");
          if (functionStr) {
            const std::string functionName = std::string{*functionStr};
            virtualTargets[entry_caller].insert(functionName);
          }
        }
      }
    }
  }
}

std::unique_ptr<JSONCG> JSONCG::getJSON(const std::string& fileName) {
  using namespace llvm;
  auto memBuffer = MemoryBuffer::getFile(fileName);

  if (memBuffer) {
    auto json = llvm::json::parse(memBuffer.get()->getBuffer());

    if (!json) {
      std::string str;
      llvm::raw_string_ostream ostr(str);
      ostr << json.takeError();
      LOG_FATAL(ostr.str());
      exit(-1);
    }

    return std::make_unique<JSONCG>(json.get());
  }
  LOG_FATAL("No CG file provided / file cannot be found: " << fileName);
  return nullptr;
}

}  // namespace typeart::filter

1	// TypeART library
2	//
3	// Copyright (c) 2017-2025 TypeART Authors
4	// Distributed under the BSD 3-Clause license.
5	// (See accompanying file LICENSE.txt or copy at
6	// https://opensource.org/licenses/BSD-3-Clause)
7	//
8	// Project home: https://github.com/tudasc/TypeART
9	//
10	// SPDX-License-Identifier: BSD-3-Clause
11	//
12
13	#include "CGInterface.h"
14
15	#include "support/Logger.h"
16	#include "support/Util.h"
17
18	#include "llvm/Support/Error.h"
19	#include "llvm/Support/ErrorOr.h"
20	#include "llvm/Support/JSON.h"
21	#include "llvm/Support/MemoryBuffer.h"
22	#include "llvm/Support/raw_ostream.h"
23
24	#include <algorithm>
25	#include <cassert>
26	#include <cstdlib>
27
28	namespace typeart::filter {
29
30	CGInterface::ReachabilityResult JSONCG::reachable(const std::string& source, const std::string& target,	24✔
31	bool case_sensitive, bool short_circuit) {
32	const auto reachables = get_reachable_functions(source);	24✔
33	bool matches = false;	24✔
34	bool allBodies = true;	24✔
35
36	// bool what{false};
37	if (!hasBodyMap[source]) {	24✔
38	++no_call_chain;	24✔
39	}	24✔
40
41	if (!reachables.empty()) {	24✔
42	++call_chain;	12✔
43	f.push_back(source);	12✔
44	}	12✔
45
46	for (const auto& function_n : reachables) {	24✔
47	matches \|= util::regex_matches(target, function_n, case_sensitive);	12✔
48	allBodies = allBodies && hasBodyMap[target];	12✔
49
50	if (matches && short_circuit) {	12✔
51	// we have found a match -> whether all functions have bodies is irrelevant
52	return ReachabilityResult::reaches;	12✔
53	}
54	}
55
56	if (matches) {	12✔
57	// matches but no short circuit
58	return ReachabilityResult::reaches;	×
59	}	28✔
60		28✔
61	if (!matches && (!allBodies \|\| !hasBodyMap[source])) {	12✔
62	// We did not find a match, but not all functions had bodies, we don't know
63	return ReachabilityResult::maybe_reaches;	12✔
64	}
65
66	// No match and all functions had bodies -> never reaches (all call targets found)
67	return ReachabilityResult::never_reaches;	×
68	}	24✔
69
70	std::vector<std::string> JSONCG::get_decl_only() {	×
71	std::vector<std::string> list;	×
72	list.reserve(hasBodyMap.size());	×
73	for (const auto& [func, has_body] : hasBodyMap) {	×
74	if (!has_body) {	×
75	list.push_back(func);	×
76	}	×
77	}
78	return list;	×
79	}	×
80
81	std::unordered_set<std::string> JSONCG::get_reachable_functions(const std::string& caller,	24✔
82	bool considerOverrides) const {
83	std::unordered_set<std::string> ret;	24✔
84	std::unordered_set<std::string> worklist;	24✔
85
86	worklist = get_directly_called_function_names(caller, considerOverrides);	24✔
87	while (!worklist.empty()) {	36✔
88	const std::string func_name = *worklist.begin();	12✔
89	// Check if we did not already handled it
90	if (ret.find(func_name) == ret.end()) {	12✔
91	worklist.merge(get_directly_called_function_names(func_name));	12✔
92	ret.insert(func_name);	12✔
93	}	12✔
94	worklist.erase(worklist.find(func_name)); // Iterators get invalidated by merge, so we need to search again	12✔
95	}	12✔
96	return ret;	24✔
97	}	24✔
98
99	std::unordered_set<std::string> JSONCG::get_directly_called_function_names(const std::string& caller,	36✔
100	bool considerOverrides) const {
101	auto ref = directly_called_functions.find(caller);	36✔
102	if (ref != std::end(directly_called_functions)) {	36✔
103	// If the caller is virtual and overridden, add the overriding functions
104	if (considerOverrides && (virtualTargets.find(caller) != std::end(virtualTargets))) {	24✔
105	auto targets = ref->second;	×
106	auto vTargets = virtualTargets.find(caller)->second;	×
107	targets.merge(vTargets);	×
108	return targets;	×
109	}	×
110	return ref->second;	24✔
111	}
112	return {};	12✔
113	}	36✔
114
115	JSONCG::JSONCG(const llvm::json::Value& cg) {	68✔
116	// Expected json format is the following:
117	// A top level object/map Key is the function name, value is an object/map with information
118	// We only care about "callees"
119	// callees itself is an array with function names (as strings)
120	assert(cg.kind() == llvm::json::Value::Kind::Object && "Top level json must be an Object");	80✔
121	const llvm::json::Object* tlobj = cg.getAsObject();	40✔
122	if (tlobj != nullptr) {	40✔
123	for (const auto& entry : *tlobj) {	104✔
124	// std::cout << "Building call site info for " << entry.first.str() << std::endl;
125	construct_call_information(entry.first.str(), *tlobj);	64✔
126	}
127	}	40✔
128	}	28✔
129
130	void JSONCG::construct_call_information(const std::string& entry_caller, const llvm::json::Object& j) {	64✔
131	if (directly_called_functions.find(entry_caller) == directly_called_functions.end()) {	64✔
132	// We did not handle this function yet
133	directly_called_functions[entry_caller] = std::unordered_set<std::string>();	64✔
134	const auto caller = j.getObject(entry_caller);	64✔
135	if (caller != nullptr) {	64✔
136	const auto hasBody = caller->get("hasBody");	64✔
137	if (hasBody != nullptr) {	64✔
138	assert(hasBody->kind() == llvm::json::Value::Kind::Boolean && "hasBody must be boolean");	128✔
139	hasBodyMap[entry_caller] = *hasBody->getAsBoolean();	64✔
140	}	64✔
141	const auto calls = caller->getArray("callees");	64✔
142	assert(calls != nullptr && "Json callee information is missing");	128✔
143	if (calls != nullptr) {	64✔
144	// Now iterate over them
145	for (const auto& callee : *calls) {	144✔
146	assert(callee.kind() == llvm::json::Value::Kind::String && "Callees must be strings");	160✔
147	const auto callee_json_string = callee.getAsString();	80✔
148	assert(callee_json_string && "Could not get callee as string");	160✔
149	if (callee_json_string) {	80✔
150	const std::string callee_string = std::string{*callee_json_string};	80✔
151	directly_called_functions[entry_caller].insert(callee_string);	80✔
152	}	80✔
153	}
154	}	64✔
155	// if the function is virtual, overriding functions are _potential_ call targets
156	const auto overridingFunctions = caller->getArray("overriddenBy");	64✔
157	if (overridingFunctions != nullptr) {	64✔
158	for (const auto& function : *overridingFunctions) {	64✔
159	assert(function.kind() == llvm::json::Value::Kind::String && "Function names are always strings");	×
160	const auto functionStr = function.getAsString();	×
NEW 161	assert(functionStr && "Retrieving overriding function as String failed");	×
NEW 162	if (functionStr) {	×
NEW 163	const std::string functionName = std::string{*functionStr};	×
164	virtualTargets[entry_caller].insert(functionName);	×
165	}	×
166	}
167	}	64✔
168	}	64✔
169	}	64✔
170	}	64✔
171
172	std::unique_ptr<JSONCG> JSONCG::getJSON(const std::string& fileName) {	40✔
173	using namespace llvm;
174	auto memBuffer = MemoryBuffer::getFile(fileName);	40✔
175
176	if (memBuffer) {	40✔
177	auto json = llvm::json::parse(memBuffer.get()->getBuffer());	40✔
178
179	if (!json) {	40✔
180	std::string str;	×
181	llvm::raw_string_ostream ostr(str);	×
182	ostr << json.takeError();	×
183	LOG_FATAL(ostr.str());	×
184	exit(-1);	×
185	}	×
186
187	return std::make_unique<JSONCG>(json.get());	40✔
188	}	40✔
189	LOG_FATAL("No CG file provided / file cannot be found: " << fileName);	×
190	return nullptr;	×
191	}	40✔
192
193	} // namespace typeart::filter

tudasc / TypeART / 13955121336

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