1 //===-- InstrumentationRuntimeTSan.cpp ------------------------------------===//
2 //
3 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
4 // See https://llvm.org/LICENSE.txt for license information.
5 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
6 //
7 //===----------------------------------------------------------------------===//
8 
9 #include "InstrumentationRuntimeTSan.h"
10 
11 #include "Plugins/Process/Utility/HistoryThread.h"
12 #include "lldb/Breakpoint/StoppointCallbackContext.h"
13 #include "lldb/Core/Debugger.h"
14 #include "lldb/Core/Module.h"
15 #include "lldb/Core/PluginInterface.h"
16 #include "lldb/Core/PluginManager.h"
17 #include "lldb/Core/ValueObject.h"
18 #include "lldb/Expression/UserExpression.h"
19 #include "lldb/Host/StreamFile.h"
20 #include "lldb/Interpreter/CommandReturnObject.h"
21 #include "lldb/Symbol/Symbol.h"
22 #include "lldb/Symbol/SymbolContext.h"
23 #include "lldb/Symbol/Variable.h"
24 #include "lldb/Symbol/VariableList.h"
25 #include "lldb/Target/InstrumentationRuntimeStopInfo.h"
26 #include "lldb/Target/SectionLoadList.h"
27 #include "lldb/Target/StopInfo.h"
28 #include "lldb/Target/Target.h"
29 #include "lldb/Target/Thread.h"
30 #include "lldb/Utility/LLDBLog.h"
31 #include "lldb/Utility/Log.h"
32 #include "lldb/Utility/RegularExpression.h"
33 #include "lldb/Utility/Stream.h"
34 
35 #include <memory>
36 
37 using namespace lldb;
38 using namespace lldb_private;
39 
40 LLDB_PLUGIN_DEFINE(InstrumentationRuntimeTSan)
41 
42 lldb::InstrumentationRuntimeSP
43 InstrumentationRuntimeTSan::CreateInstance(const lldb::ProcessSP &process_sp) {
44   return InstrumentationRuntimeSP(new InstrumentationRuntimeTSan(process_sp));
45 }
46 
47 void InstrumentationRuntimeTSan::Initialize() {
48   PluginManager::RegisterPlugin(
49       GetPluginNameStatic(), "ThreadSanitizer instrumentation runtime plugin.",
50       CreateInstance, GetTypeStatic);
51 }
52 
53 void InstrumentationRuntimeTSan::Terminate() {
54   PluginManager::UnregisterPlugin(CreateInstance);
55 }
56 
57 lldb::InstrumentationRuntimeType InstrumentationRuntimeTSan::GetTypeStatic() {
58   return eInstrumentationRuntimeTypeThreadSanitizer;
59 }
60 
61 InstrumentationRuntimeTSan::~InstrumentationRuntimeTSan() { Deactivate(); }
62 
63 const char *thread_sanitizer_retrieve_report_data_prefix = R"(
64 extern "C"
65 {
66     void *__tsan_get_current_report();
67     int __tsan_get_report_data(void *report, const char **description, int *count,
68                                int *stack_count, int *mop_count, int *loc_count,
69                                int *mutex_count, int *thread_count,
70                                int *unique_tid_count, void **sleep_trace,
71                                unsigned long trace_size);
72     int __tsan_get_report_stack(void *report, unsigned long idx, void **trace,
73                                 unsigned long trace_size);
74     int __tsan_get_report_mop(void *report, unsigned long idx, int *tid, void **addr,
75                               int *size, int *write, int *atomic, void **trace,
76                               unsigned long trace_size);
77     int __tsan_get_report_loc(void *report, unsigned long idx, const char **type,
78                               void **addr, unsigned long *start, unsigned long *size, int *tid,
79                               int *fd, int *suppressable, void **trace,
80                               unsigned long trace_size);
81     int __tsan_get_report_mutex(void *report, unsigned long idx, unsigned long *mutex_id, void **addr,
82                                 int *destroyed, void **trace, unsigned long trace_size);
83     int __tsan_get_report_thread(void *report, unsigned long idx, int *tid, unsigned long *os_id,
84                                  int *running, const char **name, int *parent_tid,
85                                  void **trace, unsigned long trace_size);
86     int __tsan_get_report_unique_tid(void *report, unsigned long idx, int *tid);
87 
88     // TODO: dlsym won't work on Windows.
89     void *dlsym(void* handle, const char* symbol);
90     int (*ptr__tsan_get_report_loc_object_type)(void *report, unsigned long idx, const char **object_type);
91 }
92 )";
93 
94 const char *thread_sanitizer_retrieve_report_data_command = R"(
95 
96 const int REPORT_TRACE_SIZE = 128;
97 const int REPORT_ARRAY_SIZE = 4;
98 
99 struct {
100     void *report;
101     const char *description;
102     int report_count;
103 
104     void *sleep_trace[REPORT_TRACE_SIZE];
105 
106     int stack_count;
107     struct {
108         int idx;
109         void *trace[REPORT_TRACE_SIZE];
110     } stacks[REPORT_ARRAY_SIZE];
111 
112     int mop_count;
113     struct {
114         int idx;
115         int tid;
116         int size;
117         int write;
118         int atomic;
119         void *addr;
120         void *trace[REPORT_TRACE_SIZE];
121     } mops[REPORT_ARRAY_SIZE];
122 
123     int loc_count;
124     struct {
125         int idx;
126         const char *type;
127         void *addr;
128         unsigned long start;
129         unsigned long size;
130         int tid;
131         int fd;
132         int suppressable;
133         void *trace[REPORT_TRACE_SIZE];
134         const char *object_type;
135     } locs[REPORT_ARRAY_SIZE];
136 
137     int mutex_count;
138     struct {
139         int idx;
140         unsigned long mutex_id;
141         void *addr;
142         int destroyed;
143         void *trace[REPORT_TRACE_SIZE];
144     } mutexes[REPORT_ARRAY_SIZE];
145 
146     int thread_count;
147     struct {
148         int idx;
149         int tid;
150         unsigned long os_id;
151         int running;
152         const char *name;
153         int parent_tid;
154         void *trace[REPORT_TRACE_SIZE];
155     } threads[REPORT_ARRAY_SIZE];
156 
157     int unique_tid_count;
158     struct {
159         int idx;
160         int tid;
161     } unique_tids[REPORT_ARRAY_SIZE];
162 } t = {0};
163 
164 ptr__tsan_get_report_loc_object_type = (typeof(ptr__tsan_get_report_loc_object_type))(void *)dlsym((void*)-2 /*RTLD_DEFAULT*/, "__tsan_get_report_loc_object_type");
165 
166 t.report = __tsan_get_current_report();
167 __tsan_get_report_data(t.report, &t.description, &t.report_count, &t.stack_count, &t.mop_count, &t.loc_count, &t.mutex_count, &t.thread_count, &t.unique_tid_count, t.sleep_trace, REPORT_TRACE_SIZE);
168 
169 if (t.stack_count > REPORT_ARRAY_SIZE) t.stack_count = REPORT_ARRAY_SIZE;
170 for (int i = 0; i < t.stack_count; i++) {
171     t.stacks[i].idx = i;
172     __tsan_get_report_stack(t.report, i, t.stacks[i].trace, REPORT_TRACE_SIZE);
173 }
174 
175 if (t.mop_count > REPORT_ARRAY_SIZE) t.mop_count = REPORT_ARRAY_SIZE;
176 for (int i = 0; i < t.mop_count; i++) {
177     t.mops[i].idx = i;
178     __tsan_get_report_mop(t.report, i, &t.mops[i].tid, &t.mops[i].addr, &t.mops[i].size, &t.mops[i].write, &t.mops[i].atomic, t.mops[i].trace, REPORT_TRACE_SIZE);
179 }
180 
181 if (t.loc_count > REPORT_ARRAY_SIZE) t.loc_count = REPORT_ARRAY_SIZE;
182 for (int i = 0; i < t.loc_count; i++) {
183     t.locs[i].idx = i;
184     __tsan_get_report_loc(t.report, i, &t.locs[i].type, &t.locs[i].addr, &t.locs[i].start, &t.locs[i].size, &t.locs[i].tid, &t.locs[i].fd, &t.locs[i].suppressable, t.locs[i].trace, REPORT_TRACE_SIZE);
185     if (ptr__tsan_get_report_loc_object_type)
186         ptr__tsan_get_report_loc_object_type(t.report, i, &t.locs[i].object_type);
187 }
188 
189 if (t.mutex_count > REPORT_ARRAY_SIZE) t.mutex_count = REPORT_ARRAY_SIZE;
190 for (int i = 0; i < t.mutex_count; i++) {
191     t.mutexes[i].idx = i;
192     __tsan_get_report_mutex(t.report, i, &t.mutexes[i].mutex_id, &t.mutexes[i].addr, &t.mutexes[i].destroyed, t.mutexes[i].trace, REPORT_TRACE_SIZE);
193 }
194 
195 if (t.thread_count > REPORT_ARRAY_SIZE) t.thread_count = REPORT_ARRAY_SIZE;
196 for (int i = 0; i < t.thread_count; i++) {
197     t.threads[i].idx = i;
198     __tsan_get_report_thread(t.report, i, &t.threads[i].tid, &t.threads[i].os_id, &t.threads[i].running, &t.threads[i].name, &t.threads[i].parent_tid, t.threads[i].trace, REPORT_TRACE_SIZE);
199 }
200 
201 if (t.unique_tid_count > REPORT_ARRAY_SIZE) t.unique_tid_count = REPORT_ARRAY_SIZE;
202 for (int i = 0; i < t.unique_tid_count; i++) {
203     t.unique_tids[i].idx = i;
204     __tsan_get_report_unique_tid(t.report, i, &t.unique_tids[i].tid);
205 }
206 
207 t;
208 )";
209 
210 static StructuredData::ArraySP
211 CreateStackTrace(ValueObjectSP o,
212                  const std::string &trace_item_name = ".trace") {
213   auto trace_sp = std::make_shared<StructuredData::Array>();
214   ValueObjectSP trace_value_object =
215       o->GetValueForExpressionPath(trace_item_name.c_str());
216   size_t count = trace_value_object->GetNumChildrenIgnoringErrors();
217   for (size_t j = 0; j < count; j++) {
218     addr_t trace_addr =
219         trace_value_object->GetChildAtIndex(j)->GetValueAsUnsigned(0);
220     if (trace_addr == 0)
221       break;
222     trace_sp->AddIntegerItem(trace_addr);
223   }
224   return trace_sp;
225 }
226 
227 static StructuredData::ArraySP ConvertToStructuredArray(
228     ValueObjectSP return_value_sp, const std::string &items_name,
229     const std::string &count_name,
230     std::function<void(const ValueObjectSP &o,
231                        const StructuredData::DictionarySP &dict)> const
232         &callback) {
233   auto array_sp = std::make_shared<StructuredData::Array>();
234   unsigned int count =
235       return_value_sp->GetValueForExpressionPath(count_name.c_str())
236           ->GetValueAsUnsigned(0);
237   ValueObjectSP objects =
238       return_value_sp->GetValueForExpressionPath(items_name.c_str());
239   for (unsigned int i = 0; i < count; i++) {
240     ValueObjectSP o = objects->GetChildAtIndex(i);
241     auto dict_sp = std::make_shared<StructuredData::Dictionary>();
242 
243     callback(o, dict_sp);
244 
245     array_sp->AddItem(dict_sp);
246   }
247   return array_sp;
248 }
249 
250 static std::string RetrieveString(ValueObjectSP return_value_sp,
251                                   ProcessSP process_sp,
252                                   const std::string &expression_path) {
253   addr_t ptr =
254       return_value_sp->GetValueForExpressionPath(expression_path.c_str())
255           ->GetValueAsUnsigned(0);
256   std::string str;
257   Status error;
258   process_sp->ReadCStringFromMemory(ptr, str, error);
259   return str;
260 }
261 
262 static void
263 GetRenumberedThreadIds(ProcessSP process_sp, ValueObjectSP data,
264                        std::map<uint64_t, user_id_t> &thread_id_map) {
265   ConvertToStructuredArray(
266       data, ".threads", ".thread_count",
267       [process_sp, &thread_id_map](const ValueObjectSP &o,
268                                    const StructuredData::DictionarySP &dict) {
269         uint64_t thread_id =
270             o->GetValueForExpressionPath(".tid")->GetValueAsUnsigned(0);
271         uint64_t thread_os_id =
272             o->GetValueForExpressionPath(".os_id")->GetValueAsUnsigned(0);
273         user_id_t lldb_user_id = 0;
274 
275         bool can_update = true;
276         ThreadSP lldb_thread = process_sp->GetThreadList().FindThreadByID(
277             thread_os_id, can_update);
278         if (lldb_thread) {
279           lldb_user_id = lldb_thread->GetIndexID();
280         } else {
281           // This isn't a live thread anymore.  Ask process to assign a new
282           // Index ID (or return an old one if we've already seen this
283           // thread_os_id). It will also make sure that no new threads are
284           // assigned this Index ID.
285           lldb_user_id = process_sp->AssignIndexIDToThread(thread_os_id);
286         }
287 
288         thread_id_map[thread_id] = lldb_user_id;
289       });
290 }
291 
292 static user_id_t Renumber(uint64_t id,
293                           std::map<uint64_t, user_id_t> &thread_id_map) {
294   auto IT = thread_id_map.find(id);
295   if (IT == thread_id_map.end())
296     return 0;
297 
298   return IT->second;
299 }
300 
301 StructuredData::ObjectSP InstrumentationRuntimeTSan::RetrieveReportData(
302     ExecutionContextRef exe_ctx_ref) {
303   ProcessSP process_sp = GetProcessSP();
304   if (!process_sp)
305     return StructuredData::ObjectSP();
306 
307   ThreadSP thread_sp = exe_ctx_ref.GetThreadSP();
308   StackFrameSP frame_sp =
309       thread_sp->GetSelectedFrame(DoNoSelectMostRelevantFrame);
310 
311   if (!frame_sp)
312     return StructuredData::ObjectSP();
313 
314   EvaluateExpressionOptions options;
315   options.SetUnwindOnError(true);
316   options.SetTryAllThreads(true);
317   options.SetStopOthers(true);
318   options.SetIgnoreBreakpoints(true);
319   options.SetTimeout(process_sp->GetUtilityExpressionTimeout());
320   options.SetPrefix(thread_sanitizer_retrieve_report_data_prefix);
321   options.SetAutoApplyFixIts(false);
322   options.SetLanguage(eLanguageTypeObjC_plus_plus);
323 
324   ValueObjectSP main_value;
325   ExecutionContext exe_ctx;
326   Status eval_error;
327   frame_sp->CalculateExecutionContext(exe_ctx);
328   ExpressionResults result = UserExpression::Evaluate(
329       exe_ctx, options, thread_sanitizer_retrieve_report_data_command, "",
330       main_value, eval_error);
331   if (result != eExpressionCompleted) {
332     StreamString ss;
333     ss << "cannot evaluate ThreadSanitizer expression:\n";
334     ss << eval_error.AsCString();
335     Debugger::ReportWarning(ss.GetString().str(),
336                             process_sp->GetTarget().GetDebugger().GetID());
337     return StructuredData::ObjectSP();
338   }
339 
340   std::map<uint64_t, user_id_t> thread_id_map;
341   GetRenumberedThreadIds(process_sp, main_value, thread_id_map);
342 
343   auto dict = std::make_shared<StructuredData::Dictionary>();
344   dict->AddStringItem("instrumentation_class", "ThreadSanitizer");
345   dict->AddStringItem("issue_type",
346                       RetrieveString(main_value, process_sp, ".description"));
347   dict->AddIntegerItem("report_count",
348                        main_value->GetValueForExpressionPath(".report_count")
349                            ->GetValueAsUnsigned(0));
350   dict->AddItem("sleep_trace", CreateStackTrace(
351                                    main_value, ".sleep_trace"));
352 
353   StructuredData::ArraySP stacks = ConvertToStructuredArray(
354       main_value, ".stacks", ".stack_count",
355       [thread_sp](const ValueObjectSP &o,
356                   const StructuredData::DictionarySP &dict) {
357         dict->AddIntegerItem(
358             "index",
359             o->GetValueForExpressionPath(".idx")->GetValueAsUnsigned(0));
360         dict->AddItem("trace", CreateStackTrace(o));
361         // "stacks" happen on the current thread
362         dict->AddIntegerItem("thread_id", thread_sp->GetIndexID());
363       });
364   dict->AddItem("stacks", stacks);
365 
366   StructuredData::ArraySP mops = ConvertToStructuredArray(
367       main_value, ".mops", ".mop_count",
368       [&thread_id_map](const ValueObjectSP &o,
369                        const StructuredData::DictionarySP &dict) {
370         dict->AddIntegerItem(
371             "index",
372             o->GetValueForExpressionPath(".idx")->GetValueAsUnsigned(0));
373         dict->AddIntegerItem(
374             "thread_id",
375             Renumber(
376                 o->GetValueForExpressionPath(".tid")->GetValueAsUnsigned(0),
377                 thread_id_map));
378         dict->AddIntegerItem(
379             "size",
380             o->GetValueForExpressionPath(".size")->GetValueAsUnsigned(0));
381         dict->AddBooleanItem(
382             "is_write",
383             o->GetValueForExpressionPath(".write")->GetValueAsUnsigned(0));
384         dict->AddBooleanItem(
385             "is_atomic",
386             o->GetValueForExpressionPath(".atomic")->GetValueAsUnsigned(0));
387         dict->AddIntegerItem(
388             "address",
389             o->GetValueForExpressionPath(".addr")->GetValueAsUnsigned(0));
390         dict->AddItem("trace", CreateStackTrace(o));
391       });
392   dict->AddItem("mops", mops);
393 
394   StructuredData::ArraySP locs = ConvertToStructuredArray(
395       main_value, ".locs", ".loc_count",
396       [process_sp, &thread_id_map](const ValueObjectSP &o,
397                                    const StructuredData::DictionarySP &dict) {
398         dict->AddIntegerItem(
399             "index",
400             o->GetValueForExpressionPath(".idx")->GetValueAsUnsigned(0));
401         dict->AddStringItem("type", RetrieveString(o, process_sp, ".type"));
402         dict->AddIntegerItem(
403             "address",
404             o->GetValueForExpressionPath(".addr")->GetValueAsUnsigned(0));
405         dict->AddIntegerItem(
406             "start",
407             o->GetValueForExpressionPath(".start")->GetValueAsUnsigned(0));
408         dict->AddIntegerItem(
409             "size",
410             o->GetValueForExpressionPath(".size")->GetValueAsUnsigned(0));
411         dict->AddIntegerItem(
412             "thread_id",
413             Renumber(
414                 o->GetValueForExpressionPath(".tid")->GetValueAsUnsigned(0),
415                 thread_id_map));
416         dict->AddIntegerItem(
417             "file_descriptor",
418             o->GetValueForExpressionPath(".fd")->GetValueAsUnsigned(0));
419         dict->AddIntegerItem("suppressable",
420                              o->GetValueForExpressionPath(".suppressable")
421                                  ->GetValueAsUnsigned(0));
422         dict->AddItem("trace", CreateStackTrace(o));
423         dict->AddStringItem("object_type",
424                             RetrieveString(o, process_sp, ".object_type"));
425       });
426   dict->AddItem("locs", locs);
427 
428   StructuredData::ArraySP mutexes = ConvertToStructuredArray(
429       main_value, ".mutexes", ".mutex_count",
430       [](const ValueObjectSP &o, const StructuredData::DictionarySP &dict) {
431         dict->AddIntegerItem(
432             "index",
433             o->GetValueForExpressionPath(".idx")->GetValueAsUnsigned(0));
434         dict->AddIntegerItem(
435             "mutex_id",
436             o->GetValueForExpressionPath(".mutex_id")->GetValueAsUnsigned(0));
437         dict->AddIntegerItem(
438             "address",
439             o->GetValueForExpressionPath(".addr")->GetValueAsUnsigned(0));
440         dict->AddIntegerItem(
441             "destroyed",
442             o->GetValueForExpressionPath(".destroyed")->GetValueAsUnsigned(0));
443         dict->AddItem("trace", CreateStackTrace(o));
444       });
445   dict->AddItem("mutexes", mutexes);
446 
447   StructuredData::ArraySP threads = ConvertToStructuredArray(
448       main_value, ".threads", ".thread_count",
449       [process_sp, &thread_id_map](const ValueObjectSP &o,
450                                    const StructuredData::DictionarySP &dict) {
451         dict->AddIntegerItem(
452             "index",
453             o->GetValueForExpressionPath(".idx")->GetValueAsUnsigned(0));
454         dict->AddIntegerItem(
455             "thread_id",
456             Renumber(
457                 o->GetValueForExpressionPath(".tid")->GetValueAsUnsigned(0),
458                 thread_id_map));
459         dict->AddIntegerItem(
460             "thread_os_id",
461             o->GetValueForExpressionPath(".os_id")->GetValueAsUnsigned(0));
462         dict->AddIntegerItem(
463             "running",
464             o->GetValueForExpressionPath(".running")->GetValueAsUnsigned(0));
465         dict->AddStringItem("name", RetrieveString(o, process_sp, ".name"));
466         dict->AddIntegerItem(
467             "parent_thread_id",
468             Renumber(o->GetValueForExpressionPath(".parent_tid")
469                          ->GetValueAsUnsigned(0),
470                      thread_id_map));
471         dict->AddItem("trace", CreateStackTrace(o));
472       });
473   dict->AddItem("threads", threads);
474 
475   StructuredData::ArraySP unique_tids = ConvertToStructuredArray(
476       main_value, ".unique_tids", ".unique_tid_count",
477       [&thread_id_map](const ValueObjectSP &o,
478                        const StructuredData::DictionarySP &dict) {
479         dict->AddIntegerItem(
480             "index",
481             o->GetValueForExpressionPath(".idx")->GetValueAsUnsigned(0));
482         dict->AddIntegerItem(
483             "tid",
484             Renumber(
485                 o->GetValueForExpressionPath(".tid")->GetValueAsUnsigned(0),
486                 thread_id_map));
487       });
488   dict->AddItem("unique_tids", unique_tids);
489 
490   return dict;
491 }
492 
493 std::string
494 InstrumentationRuntimeTSan::FormatDescription(StructuredData::ObjectSP report) {
495   std::string description = std::string(report->GetAsDictionary()
496                                             ->GetValueForKey("issue_type")
497                                             ->GetAsString()
498                                             ->GetValue());
499 
500   if (description == "data-race") {
501     return "Data race";
502   } else if (description == "data-race-vptr") {
503     return "Data race on C++ virtual pointer";
504   } else if (description == "heap-use-after-free") {
505     return "Use of deallocated memory";
506   } else if (description == "heap-use-after-free-vptr") {
507     return "Use of deallocated C++ virtual pointer";
508   } else if (description == "thread-leak") {
509     return "Thread leak";
510   } else if (description == "locked-mutex-destroy") {
511     return "Destruction of a locked mutex";
512   } else if (description == "mutex-double-lock") {
513     return "Double lock of a mutex";
514   } else if (description == "mutex-invalid-access") {
515     return "Use of an uninitialized or destroyed mutex";
516   } else if (description == "mutex-bad-unlock") {
517     return "Unlock of an unlocked mutex (or by a wrong thread)";
518   } else if (description == "mutex-bad-read-lock") {
519     return "Read lock of a write locked mutex";
520   } else if (description == "mutex-bad-read-unlock") {
521     return "Read unlock of a write locked mutex";
522   } else if (description == "signal-unsafe-call") {
523     return "Signal-unsafe call inside a signal handler";
524   } else if (description == "errno-in-signal-handler") {
525     return "Overwrite of errno in a signal handler";
526   } else if (description == "lock-order-inversion") {
527     return "Lock order inversion (potential deadlock)";
528   } else if (description == "external-race") {
529     return "Race on a library object";
530   } else if (description == "swift-access-race") {
531     return "Swift access race";
532   }
533 
534   // for unknown report codes just show the code
535   return description;
536 }
537 
538 static std::string Sprintf(const char *format, ...) {
539   StreamString s;
540   va_list args;
541   va_start(args, format);
542   s.PrintfVarArg(format, args);
543   va_end(args);
544   return std::string(s.GetString());
545 }
546 
547 static std::string GetSymbolNameFromAddress(ProcessSP process_sp, addr_t addr) {
548   lldb_private::Address so_addr;
549   if (!process_sp->GetTarget().GetSectionLoadList().ResolveLoadAddress(addr,
550                                                                        so_addr))
551     return "";
552 
553   lldb_private::Symbol *symbol = so_addr.CalculateSymbolContextSymbol();
554   if (!symbol)
555     return "";
556 
557   std::string sym_name = symbol->GetName().GetCString();
558   return sym_name;
559 }
560 
561 static void GetSymbolDeclarationFromAddress(ProcessSP process_sp, addr_t addr,
562                                             Declaration &decl) {
563   lldb_private::Address so_addr;
564   if (!process_sp->GetTarget().GetSectionLoadList().ResolveLoadAddress(addr,
565                                                                        so_addr))
566     return;
567 
568   lldb_private::Symbol *symbol = so_addr.CalculateSymbolContextSymbol();
569   if (!symbol)
570     return;
571 
572   ConstString sym_name = symbol->GetMangled().GetName(Mangled::ePreferMangled);
573 
574   ModuleSP module = symbol->CalculateSymbolContextModule();
575   if (!module)
576     return;
577 
578   VariableList var_list;
579   module->FindGlobalVariables(sym_name, CompilerDeclContext(), 1U, var_list);
580   if (var_list.GetSize() < 1)
581     return;
582 
583   VariableSP var = var_list.GetVariableAtIndex(0);
584   decl = var->GetDeclaration();
585 }
586 
587 addr_t InstrumentationRuntimeTSan::GetFirstNonInternalFramePc(
588     StructuredData::ObjectSP trace, bool skip_one_frame) {
589   ProcessSP process_sp = GetProcessSP();
590   ModuleSP runtime_module_sp = GetRuntimeModuleSP();
591 
592   StructuredData::Array *trace_array = trace->GetAsArray();
593   for (size_t i = 0; i < trace_array->GetSize(); i++) {
594     if (skip_one_frame && i == 0)
595       continue;
596 
597     auto maybe_addr = trace_array->GetItemAtIndexAsInteger<addr_t>(i);
598     if (!maybe_addr)
599       continue;
600     addr_t addr = *maybe_addr;
601 
602     lldb_private::Address so_addr;
603     if (!process_sp->GetTarget().GetSectionLoadList().ResolveLoadAddress(
604             addr, so_addr))
605       continue;
606 
607     if (so_addr.GetModule() == runtime_module_sp)
608       continue;
609 
610     return addr;
611   }
612 
613   return 0;
614 }
615 
616 std::string
617 InstrumentationRuntimeTSan::GenerateSummary(StructuredData::ObjectSP report) {
618   ProcessSP process_sp = GetProcessSP();
619 
620   std::string summary = std::string(report->GetAsDictionary()
621                                         ->GetValueForKey("description")
622                                         ->GetAsString()
623                                         ->GetValue());
624   bool skip_one_frame =
625       report->GetObjectForDotSeparatedPath("issue_type")->GetStringValue() ==
626       "external-race";
627 
628   addr_t pc = 0;
629   if (report->GetAsDictionary()
630           ->GetValueForKey("mops")
631           ->GetAsArray()
632           ->GetSize() > 0)
633     pc = GetFirstNonInternalFramePc(report->GetAsDictionary()
634                                         ->GetValueForKey("mops")
635                                         ->GetAsArray()
636                                         ->GetItemAtIndex(0)
637                                         ->GetAsDictionary()
638                                         ->GetValueForKey("trace"),
639                                     skip_one_frame);
640 
641   if (report->GetAsDictionary()
642           ->GetValueForKey("stacks")
643           ->GetAsArray()
644           ->GetSize() > 0)
645     pc = GetFirstNonInternalFramePc(report->GetAsDictionary()
646                                         ->GetValueForKey("stacks")
647                                         ->GetAsArray()
648                                         ->GetItemAtIndex(0)
649                                         ->GetAsDictionary()
650                                         ->GetValueForKey("trace"),
651                                     skip_one_frame);
652 
653   if (pc != 0) {
654     summary = summary + " in " + GetSymbolNameFromAddress(process_sp, pc);
655   }
656 
657   if (report->GetAsDictionary()
658           ->GetValueForKey("locs")
659           ->GetAsArray()
660           ->GetSize() > 0) {
661     StructuredData::ObjectSP loc = report->GetAsDictionary()
662                                        ->GetValueForKey("locs")
663                                        ->GetAsArray()
664                                        ->GetItemAtIndex(0);
665     std::string object_type = std::string(loc->GetAsDictionary()
666                                               ->GetValueForKey("object_type")
667                                               ->GetAsString()
668                                               ->GetValue());
669     if (!object_type.empty()) {
670       summary = "Race on " + object_type + " object";
671     }
672     addr_t addr = loc->GetAsDictionary()
673                       ->GetValueForKey("address")
674                       ->GetUnsignedIntegerValue();
675     if (addr == 0)
676       addr = loc->GetAsDictionary()
677                  ->GetValueForKey("start")
678                  ->GetUnsignedIntegerValue();
679 
680     if (addr != 0) {
681       std::string global_name = GetSymbolNameFromAddress(process_sp, addr);
682       if (!global_name.empty()) {
683         summary = summary + " at " + global_name;
684       } else {
685         summary = summary + " at " + Sprintf("0x%llx", addr);
686       }
687     } else {
688       int fd = loc->GetAsDictionary()
689                    ->GetValueForKey("file_descriptor")
690                    ->GetSignedIntegerValue();
691       if (fd != 0) {
692         summary = summary + " on file descriptor " + Sprintf("%d", fd);
693       }
694     }
695   }
696 
697   return summary;
698 }
699 
700 addr_t InstrumentationRuntimeTSan::GetMainRacyAddress(
701     StructuredData::ObjectSP report) {
702   addr_t result = (addr_t)-1;
703 
704   report->GetObjectForDotSeparatedPath("mops")->GetAsArray()->ForEach(
705       [&result](StructuredData::Object *o) -> bool {
706         addr_t addr = o->GetObjectForDotSeparatedPath("address")
707                           ->GetUnsignedIntegerValue();
708         if (addr < result)
709           result = addr;
710         return true;
711       });
712 
713   return (result == (addr_t)-1) ? 0 : result;
714 }
715 
716 std::string InstrumentationRuntimeTSan::GetLocationDescription(
717     StructuredData::ObjectSP report, addr_t &global_addr,
718     std::string &global_name, std::string &filename, uint32_t &line) {
719   std::string result;
720 
721   ProcessSP process_sp = GetProcessSP();
722 
723   if (report->GetAsDictionary()
724           ->GetValueForKey("locs")
725           ->GetAsArray()
726           ->GetSize() > 0) {
727     StructuredData::ObjectSP loc = report->GetAsDictionary()
728                                        ->GetValueForKey("locs")
729                                        ->GetAsArray()
730                                        ->GetItemAtIndex(0);
731     std::string type = std::string(
732         loc->GetAsDictionary()->GetValueForKey("type")->GetStringValue());
733     if (type == "global") {
734       global_addr = loc->GetAsDictionary()
735                         ->GetValueForKey("address")
736                         ->GetUnsignedIntegerValue();
737 
738       global_name = GetSymbolNameFromAddress(process_sp, global_addr);
739       if (!global_name.empty()) {
740         result = Sprintf("'%s' is a global variable (0x%llx)",
741                          global_name.c_str(), global_addr);
742       } else {
743         result = Sprintf("0x%llx is a global variable", global_addr);
744       }
745 
746       Declaration decl;
747       GetSymbolDeclarationFromAddress(process_sp, global_addr, decl);
748       if (decl.GetFile()) {
749         filename = decl.GetFile().GetPath();
750         line = decl.GetLine();
751       }
752     } else if (type == "heap") {
753       addr_t addr = loc->GetAsDictionary()
754                         ->GetValueForKey("start")
755                         ->GetUnsignedIntegerValue();
756 
757       size_t size = loc->GetAsDictionary()
758                         ->GetValueForKey("size")
759                         ->GetUnsignedIntegerValue();
760 
761       std::string object_type = std::string(loc->GetAsDictionary()
762                                                 ->GetValueForKey("object_type")
763                                                 ->GetAsString()
764                                                 ->GetValue());
765       if (!object_type.empty()) {
766         result = Sprintf("Location is a %ld-byte %s object at 0x%llx", size,
767                          object_type.c_str(), addr);
768       } else {
769         result =
770             Sprintf("Location is a %ld-byte heap object at 0x%llx", size, addr);
771       }
772     } else if (type == "stack") {
773       tid_t tid = loc->GetAsDictionary()
774                       ->GetValueForKey("thread_id")
775                       ->GetUnsignedIntegerValue();
776 
777       result = Sprintf("Location is stack of thread %d", tid);
778     } else if (type == "tls") {
779       tid_t tid = loc->GetAsDictionary()
780                       ->GetValueForKey("thread_id")
781                       ->GetUnsignedIntegerValue();
782 
783       result = Sprintf("Location is TLS of thread %d", tid);
784     } else if (type == "fd") {
785       int fd = loc->GetAsDictionary()
786                    ->GetValueForKey("file_descriptor")
787                    ->GetSignedIntegerValue();
788 
789       result = Sprintf("Location is file descriptor %d", fd);
790     }
791   }
792 
793   return result;
794 }
795 
796 bool InstrumentationRuntimeTSan::NotifyBreakpointHit(
797     void *baton, StoppointCallbackContext *context, user_id_t break_id,
798     user_id_t break_loc_id) {
799   assert(baton && "null baton");
800   if (!baton)
801     return false;
802 
803   InstrumentationRuntimeTSan *const instance =
804       static_cast<InstrumentationRuntimeTSan *>(baton);
805 
806   ProcessSP process_sp = instance->GetProcessSP();
807 
808   if (process_sp->GetModIDRef().IsLastResumeForUserExpression())
809     return false;
810 
811   StructuredData::ObjectSP report =
812       instance->RetrieveReportData(context->exe_ctx_ref);
813   std::string stop_reason_description =
814       "unknown thread sanitizer fault (unable to extract thread sanitizer "
815       "report)";
816   if (report) {
817     std::string issue_description = instance->FormatDescription(report);
818     report->GetAsDictionary()->AddStringItem("description", issue_description);
819     stop_reason_description = issue_description + " detected";
820     report->GetAsDictionary()->AddStringItem("stop_description",
821                                              stop_reason_description);
822     std::string summary = instance->GenerateSummary(report);
823     report->GetAsDictionary()->AddStringItem("summary", summary);
824     addr_t main_address = instance->GetMainRacyAddress(report);
825     report->GetAsDictionary()->AddIntegerItem("memory_address", main_address);
826 
827     addr_t global_addr = 0;
828     std::string global_name;
829     std::string location_filename;
830     uint32_t location_line = 0;
831     std::string location_description = instance->GetLocationDescription(
832         report, global_addr, global_name, location_filename, location_line);
833     report->GetAsDictionary()->AddStringItem("location_description",
834                                              location_description);
835     if (global_addr != 0) {
836       report->GetAsDictionary()->AddIntegerItem("global_address", global_addr);
837     }
838     if (!global_name.empty()) {
839       report->GetAsDictionary()->AddStringItem("global_name", global_name);
840     }
841     if (location_filename != "") {
842       report->GetAsDictionary()->AddStringItem("location_filename",
843                                                location_filename);
844       report->GetAsDictionary()->AddIntegerItem("location_line", location_line);
845     }
846 
847     bool all_addresses_are_same = true;
848     report->GetObjectForDotSeparatedPath("mops")->GetAsArray()->ForEach(
849         [&all_addresses_are_same,
850          main_address](StructuredData::Object *o) -> bool {
851           addr_t addr = o->GetObjectForDotSeparatedPath("address")
852                             ->GetUnsignedIntegerValue();
853           if (main_address != addr)
854             all_addresses_are_same = false;
855           return true;
856         });
857     report->GetAsDictionary()->AddBooleanItem("all_addresses_are_same",
858                                               all_addresses_are_same);
859   }
860 
861   // Make sure this is the right process
862   if (process_sp && process_sp == context->exe_ctx_ref.GetProcessSP()) {
863     ThreadSP thread_sp = context->exe_ctx_ref.GetThreadSP();
864     if (thread_sp)
865       thread_sp->SetStopInfo(
866           InstrumentationRuntimeStopInfo::
867               CreateStopReasonWithInstrumentationData(
868                   *thread_sp, stop_reason_description, report));
869 
870     StreamFile &s = process_sp->GetTarget().GetDebugger().GetOutputStream();
871     s.Printf("ThreadSanitizer report breakpoint hit. Use 'thread "
872              "info -s' to get extended information about the "
873              "report.\n");
874 
875     return true; // Return true to stop the target
876   } else
877     return false; // Let target run
878 }
879 
880 const RegularExpression &
881 InstrumentationRuntimeTSan::GetPatternForRuntimeLibrary() {
882   static RegularExpression regex(llvm::StringRef("libclang_rt.tsan_"));
883   return regex;
884 }
885 
886 bool InstrumentationRuntimeTSan::CheckIfRuntimeIsValid(
887     const lldb::ModuleSP module_sp) {
888   static ConstString g_tsan_get_current_report("__tsan_get_current_report");
889   const Symbol *symbol = module_sp->FindFirstSymbolWithNameAndType(
890       g_tsan_get_current_report, lldb::eSymbolTypeAny);
891   return symbol != nullptr;
892 }
893 
894 void InstrumentationRuntimeTSan::Activate() {
895   if (IsActive())
896     return;
897 
898   ProcessSP process_sp = GetProcessSP();
899   if (!process_sp)
900     return;
901 
902   ConstString symbol_name("__tsan_on_report");
903   const Symbol *symbol = GetRuntimeModuleSP()->FindFirstSymbolWithNameAndType(
904       symbol_name, eSymbolTypeCode);
905 
906   if (symbol == nullptr)
907     return;
908 
909   if (!symbol->ValueIsAddress() || !symbol->GetAddressRef().IsValid())
910     return;
911 
912   Target &target = process_sp->GetTarget();
913   addr_t symbol_address = symbol->GetAddressRef().GetOpcodeLoadAddress(&target);
914 
915   if (symbol_address == LLDB_INVALID_ADDRESS)
916     return;
917 
918   const bool internal = true;
919   const bool hardware = false;
920   const bool sync = false;
921   Breakpoint *breakpoint =
922       process_sp->GetTarget()
923           .CreateBreakpoint(symbol_address, internal, hardware)
924           .get();
925   breakpoint->SetCallback(InstrumentationRuntimeTSan::NotifyBreakpointHit, this,
926                           sync);
927   breakpoint->SetBreakpointKind("thread-sanitizer-report");
928   SetBreakpointID(breakpoint->GetID());
929 
930   SetActive(true);
931 }
932 
933 void InstrumentationRuntimeTSan::Deactivate() {
934   if (GetBreakpointID() != LLDB_INVALID_BREAK_ID) {
935     ProcessSP process_sp = GetProcessSP();
936     if (process_sp) {
937       process_sp->GetTarget().RemoveBreakpointByID(GetBreakpointID());
938       SetBreakpointID(LLDB_INVALID_BREAK_ID);
939     }
940   }
941   SetActive(false);
942 }
943 static std::string GenerateThreadName(const std::string &path,
944                                       StructuredData::Object *o,
945                                       StructuredData::ObjectSP main_info) {
946   std::string result = "additional information";
947 
948   if (path == "mops") {
949     size_t size =
950         o->GetObjectForDotSeparatedPath("size")->GetUnsignedIntegerValue();
951     tid_t thread_id =
952         o->GetObjectForDotSeparatedPath("thread_id")->GetUnsignedIntegerValue();
953     bool is_write =
954         o->GetObjectForDotSeparatedPath("is_write")->GetBooleanValue();
955     bool is_atomic =
956         o->GetObjectForDotSeparatedPath("is_atomic")->GetBooleanValue();
957     addr_t addr =
958         o->GetObjectForDotSeparatedPath("address")->GetUnsignedIntegerValue();
959 
960     std::string addr_string = Sprintf(" at 0x%llx", addr);
961 
962     if (main_info->GetObjectForDotSeparatedPath("all_addresses_are_same")
963             ->GetBooleanValue()) {
964       addr_string = "";
965     }
966 
967     if (main_info->GetObjectForDotSeparatedPath("issue_type")
968             ->GetStringValue() == "external-race") {
969       result = Sprintf("%s access by thread %d",
970                        is_write ? "mutating" : "read-only", thread_id);
971     } else if (main_info->GetObjectForDotSeparatedPath("issue_type")
972                    ->GetStringValue() == "swift-access-race") {
973       result = Sprintf("modifying access by thread %d", thread_id);
974     } else {
975       result = Sprintf("%s%s of size %zu%s by thread %" PRIu64,
976                        is_atomic ? "atomic " : "", is_write ? "write" : "read",
977                        size, addr_string.c_str(), thread_id);
978     }
979   }
980 
981   if (path == "threads") {
982     tid_t thread_id =
983         o->GetObjectForDotSeparatedPath("thread_id")->GetUnsignedIntegerValue();
984     result = Sprintf("Thread %zu created", thread_id);
985   }
986 
987   if (path == "locs") {
988     std::string type = std::string(
989         o->GetAsDictionary()->GetValueForKey("type")->GetStringValue());
990     tid_t thread_id =
991         o->GetObjectForDotSeparatedPath("thread_id")->GetUnsignedIntegerValue();
992     int fd = o->GetObjectForDotSeparatedPath("file_descriptor")
993                  ->GetSignedIntegerValue();
994     if (type == "heap") {
995       result = Sprintf("Heap block allocated by thread %" PRIu64, thread_id);
996     } else if (type == "fd") {
997       result = Sprintf("File descriptor %d created by thread %" PRIu64, fd,
998                        thread_id);
999     }
1000   }
1001 
1002   if (path == "mutexes") {
1003     int mutex_id =
1004         o->GetObjectForDotSeparatedPath("mutex_id")->GetSignedIntegerValue();
1005 
1006     result = Sprintf("Mutex M%d created", mutex_id);
1007   }
1008 
1009   if (path == "stacks") {
1010     tid_t thread_id =
1011         o->GetObjectForDotSeparatedPath("thread_id")->GetUnsignedIntegerValue();
1012     result = Sprintf("Thread %" PRIu64, thread_id);
1013   }
1014 
1015   result[0] = toupper(result[0]);
1016 
1017   return result;
1018 }
1019 
1020 static void AddThreadsForPath(const std::string &path,
1021                               ThreadCollectionSP threads, ProcessSP process_sp,
1022                               StructuredData::ObjectSP info) {
1023   info->GetObjectForDotSeparatedPath(path)->GetAsArray()->ForEach(
1024       [process_sp, threads, path, info](StructuredData::Object *o) -> bool {
1025         std::vector<lldb::addr_t> pcs;
1026         o->GetObjectForDotSeparatedPath("trace")->GetAsArray()->ForEach(
1027             [&pcs](StructuredData::Object *pc) -> bool {
1028               pcs.push_back(pc->GetUnsignedIntegerValue());
1029               return true;
1030             });
1031 
1032         if (pcs.size() == 0)
1033           return true;
1034 
1035         StructuredData::ObjectSP thread_id_obj =
1036             o->GetObjectForDotSeparatedPath("thread_os_id");
1037         tid_t tid =
1038             thread_id_obj ? thread_id_obj->GetUnsignedIntegerValue() : 0;
1039 
1040         ThreadSP new_thread_sp =
1041             std::make_shared<HistoryThread>(*process_sp, tid, pcs);
1042         new_thread_sp->SetName(GenerateThreadName(path, o, info).c_str());
1043 
1044         // Save this in the Process' ExtendedThreadList so a strong pointer
1045         // retains the object
1046         process_sp->GetExtendedThreadList().AddThread(new_thread_sp);
1047         threads->AddThread(new_thread_sp);
1048 
1049         return true;
1050       });
1051 }
1052 
1053 lldb::ThreadCollectionSP
1054 InstrumentationRuntimeTSan::GetBacktracesFromExtendedStopInfo(
1055     StructuredData::ObjectSP info) {
1056 
1057   ThreadCollectionSP threads = std::make_shared<ThreadCollection>();
1058 
1059   if (info->GetObjectForDotSeparatedPath("instrumentation_class")
1060           ->GetStringValue() != "ThreadSanitizer")
1061     return threads;
1062 
1063   ProcessSP process_sp = GetProcessSP();
1064 
1065   AddThreadsForPath("stacks", threads, process_sp, info);
1066   AddThreadsForPath("mops", threads, process_sp, info);
1067   AddThreadsForPath("locs", threads, process_sp, info);
1068   AddThreadsForPath("mutexes", threads, process_sp, info);
1069   AddThreadsForPath("threads", threads, process_sp, info);
1070 
1071   return threads;
1072 }
1073