1 //===-- Symbol.cpp ----------------------------------------------*- C++ -*-===//
2 //
3 //                     The LLVM Compiler Infrastructure
4 //
5 // This file is distributed under the University of Illinois Open Source
6 // License. See LICENSE.TXT for details.
7 //
8 //===----------------------------------------------------------------------===//
9 
10 #include "lldb/Symbol/Symbol.h"
11 
12 #include "lldb/Core/Module.h"
13 #include "lldb/Core/ModuleSpec.h"
14 #include "lldb/Core/Section.h"
15 #include "lldb/Core/Stream.h"
16 #include "lldb/Symbol/ObjectFile.h"
17 #include "lldb/Symbol/Symtab.h"
18 #include "lldb/Symbol/Function.h"
19 #include "lldb/Target/Process.h"
20 #include "lldb/Target/Target.h"
21 #include "lldb/Symbol/SymbolVendor.h"
22 
23 using namespace lldb;
24 using namespace lldb_private;
25 
26 
27 Symbol::Symbol() :
28     SymbolContextScope (),
29     m_uid (UINT32_MAX),
30     m_type_data (0),
31     m_type_data_resolved (false),
32     m_is_synthetic (false),
33     m_is_debug (false),
34     m_is_external (false),
35     m_size_is_sibling (false),
36     m_size_is_synthesized (false),
37     m_size_is_valid (false),
38     m_demangled_is_synthesized (false),
39     m_type (eSymbolTypeInvalid),
40     m_mangled (),
41     m_addr_range (),
42     m_flags ()
43 {
44 }
45 
46 Symbol::Symbol
47 (
48     uint32_t symID,
49     const char *name,
50     bool name_is_mangled,
51     SymbolType type,
52     bool external,
53     bool is_debug,
54     bool is_trampoline,
55     bool is_artificial,
56     const lldb::SectionSP &section_sp,
57     addr_t offset,
58     addr_t size,
59     bool size_is_valid,
60     uint32_t flags
61 ) :
62     SymbolContextScope (),
63     m_uid (symID),
64     m_type_data (0),
65     m_type_data_resolved (false),
66     m_is_synthetic (is_artificial),
67     m_is_debug (is_debug),
68     m_is_external (external),
69     m_size_is_sibling (false),
70     m_size_is_synthesized (false),
71     m_size_is_valid (size_is_valid || size > 0),
72     m_demangled_is_synthesized (false),
73     m_type (type),
74     m_mangled (ConstString(name), name_is_mangled),
75     m_addr_range (section_sp, offset, size),
76     m_flags (flags)
77 {
78 }
79 
80 Symbol::Symbol
81 (
82     uint32_t symID,
83     const char *name,
84     bool name_is_mangled,
85     SymbolType type,
86     bool external,
87     bool is_debug,
88     bool is_trampoline,
89     bool is_artificial,
90     const AddressRange &range,
91     bool size_is_valid,
92     uint32_t flags
93 ) :
94     SymbolContextScope (),
95     m_uid (symID),
96     m_type_data (0),
97     m_type_data_resolved (false),
98     m_is_synthetic (is_artificial),
99     m_is_debug (is_debug),
100     m_is_external (external),
101     m_size_is_sibling (false),
102     m_size_is_synthesized (false),
103     m_size_is_valid (size_is_valid || range.GetByteSize() > 0),
104     m_demangled_is_synthesized (false),
105     m_type (type),
106     m_mangled (ConstString(name), name_is_mangled),
107     m_addr_range (range),
108     m_flags (flags)
109 {
110 }
111 
112 Symbol::Symbol(const Symbol& rhs):
113     SymbolContextScope (rhs),
114     m_uid (rhs.m_uid),
115     m_type_data (rhs.m_type_data),
116     m_type_data_resolved (rhs.m_type_data_resolved),
117     m_is_synthetic (rhs.m_is_synthetic),
118     m_is_debug (rhs.m_is_debug),
119     m_is_external (rhs.m_is_external),
120     m_size_is_sibling (rhs.m_size_is_sibling),
121     m_size_is_synthesized (false),
122     m_size_is_valid (rhs.m_size_is_valid),
123     m_demangled_is_synthesized (rhs.m_demangled_is_synthesized),
124     m_type (rhs.m_type),
125     m_mangled (rhs.m_mangled),
126     m_addr_range (rhs.m_addr_range),
127     m_flags (rhs.m_flags)
128 {
129 }
130 
131 const Symbol&
132 Symbol::operator= (const Symbol& rhs)
133 {
134     if (this != &rhs)
135     {
136         SymbolContextScope::operator= (rhs);
137         m_uid = rhs.m_uid;
138         m_type_data = rhs.m_type_data;
139         m_type_data_resolved = rhs.m_type_data_resolved;
140         m_is_synthetic = rhs.m_is_synthetic;
141         m_is_debug = rhs.m_is_debug;
142         m_is_external = rhs.m_is_external;
143         m_size_is_sibling = rhs.m_size_is_sibling;
144         m_size_is_synthesized = rhs.m_size_is_sibling;
145         m_size_is_valid = rhs.m_size_is_valid;
146         m_demangled_is_synthesized = rhs.m_demangled_is_synthesized;
147         m_type = rhs.m_type;
148         m_mangled = rhs.m_mangled;
149         m_addr_range = rhs.m_addr_range;
150         m_flags = rhs.m_flags;
151     }
152     return *this;
153 }
154 
155 void
156 Symbol::Clear()
157 {
158     m_uid = UINT32_MAX;
159     m_mangled.Clear();
160     m_type_data = 0;
161     m_type_data_resolved = false;
162     m_is_synthetic = false;
163     m_is_debug = false;
164     m_is_external = false;
165     m_size_is_sibling = false;
166     m_size_is_synthesized = false;
167     m_size_is_valid = false;
168     m_demangled_is_synthesized = false;
169     m_type = eSymbolTypeInvalid;
170     m_flags = 0;
171     m_addr_range.Clear();
172 }
173 
174 bool
175 Symbol::ValueIsAddress() const
176 {
177     return m_addr_range.GetBaseAddress().GetSection().get() != nullptr;
178 }
179 
180 ConstString
181 Symbol::GetReExportedSymbolName() const
182 {
183     if (m_type == eSymbolTypeReExported)
184     {
185         // For eSymbolTypeReExported, the "const char *" from a ConstString
186         // is used as the offset in the address range base address. We can
187         // then make this back into a string that is the re-exported name.
188         intptr_t str_ptr = m_addr_range.GetBaseAddress().GetOffset();
189         if (str_ptr != 0)
190             return ConstString((const char *)str_ptr);
191         else
192             return GetName();
193     }
194     return ConstString();
195 }
196 
197 FileSpec
198 Symbol::GetReExportedSymbolSharedLibrary() const
199 {
200     if (m_type == eSymbolTypeReExported)
201     {
202         // For eSymbolTypeReExported, the "const char *" from a ConstString
203         // is used as the offset in the address range base address. We can
204         // then make this back into a string that is the re-exported name.
205         intptr_t str_ptr = m_addr_range.GetByteSize();
206         if (str_ptr != 0)
207             return FileSpec((const char *)str_ptr, false);
208     }
209     return FileSpec();
210 }
211 
212 void
213 Symbol::SetReExportedSymbolName(const ConstString &name)
214 {
215     SetType (eSymbolTypeReExported);
216     // For eSymbolTypeReExported, the "const char *" from a ConstString
217     // is used as the offset in the address range base address.
218     m_addr_range.GetBaseAddress().SetOffset((uintptr_t)name.GetCString());
219 }
220 
221 bool
222 Symbol::SetReExportedSymbolSharedLibrary(const FileSpec &fspec)
223 {
224     if (m_type == eSymbolTypeReExported)
225     {
226         // For eSymbolTypeReExported, the "const char *" from a ConstString
227         // is used as the offset in the address range base address.
228         m_addr_range.SetByteSize((uintptr_t)ConstString(fspec.GetPath().c_str()).GetCString());
229         return true;
230     }
231     return false;
232 
233 }
234 
235 uint32_t
236 Symbol::GetSiblingIndex() const
237 {
238     return m_size_is_sibling ? m_addr_range.GetByteSize() : 0;
239 }
240 
241 bool
242 Symbol::IsTrampoline () const
243 {
244     return m_type == eSymbolTypeTrampoline;
245 }
246 
247 bool
248 Symbol::IsIndirect () const
249 {
250     return m_type == eSymbolTypeResolver;
251 }
252 
253 void
254 Symbol::GetDescription (Stream *s, lldb::DescriptionLevel level, Target *target) const
255 {
256     s->Printf("id = {0x%8.8x}", m_uid);
257 
258     if (m_addr_range.GetBaseAddress().GetSection())
259     {
260         if (ValueIsAddress())
261         {
262             const lldb::addr_t byte_size = GetByteSize();
263             if (byte_size > 0)
264             {
265                 s->PutCString (", range = ");
266                 m_addr_range.Dump(s, target, Address::DumpStyleLoadAddress, Address::DumpStyleFileAddress);
267             }
268             else
269             {
270                 s->PutCString (", address = ");
271                 m_addr_range.GetBaseAddress().Dump(s, target, Address::DumpStyleLoadAddress, Address::DumpStyleFileAddress);
272             }
273         }
274         else
275             s->Printf (", value = 0x%16.16" PRIx64, m_addr_range.GetBaseAddress().GetOffset());
276     }
277     else
278     {
279         if (m_size_is_sibling)
280             s->Printf (", sibling = %5" PRIu64, m_addr_range.GetBaseAddress().GetOffset());
281         else
282             s->Printf (", value = 0x%16.16" PRIx64, m_addr_range.GetBaseAddress().GetOffset());
283     }
284     if (m_mangled.GetDemangledName())
285         s->Printf(", name=\"%s\"", m_mangled.GetDemangledName().AsCString());
286     if (m_mangled.GetMangledName())
287         s->Printf(", mangled=\"%s\"", m_mangled.GetMangledName().AsCString());
288 
289 }
290 
291 void
292 Symbol::Dump(Stream *s, Target *target, uint32_t index) const
293 {
294 //  s->Printf("%.*p: ", (int)sizeof(void*) * 2, this);
295 //  s->Indent();
296 //  s->Printf("Symbol[%5u] %6u %c%c %-12s ",
297     s->Printf("[%5u] %6u %c%c%c %-12s ",
298               index,
299               GetID(),
300               m_is_debug ? 'D' : ' ',
301               m_is_synthetic ? 'S' : ' ',
302               m_is_external ? 'X' : ' ',
303               GetTypeAsString());
304 
305     // Make sure the size of the symbol is up to date before dumping
306     GetByteSize();
307 
308     if (ValueIsAddress())
309     {
310         if (!m_addr_range.GetBaseAddress().Dump(s, nullptr, Address::DumpStyleFileAddress))
311             s->Printf("%*s", 18, "");
312 
313         s->PutChar(' ');
314 
315         if (!m_addr_range.GetBaseAddress().Dump(s, target, Address::DumpStyleLoadAddress))
316             s->Printf("%*s", 18, "");
317 
318         const char *format = m_size_is_sibling ?
319                             " Sibling -> [%5llu] 0x%8.8x %s\n":
320                             " 0x%16.16" PRIx64 " 0x%8.8x %s\n";
321         s->Printf(  format,
322                     GetByteSize(),
323                     m_flags,
324                     m_mangled.GetName().AsCString(""));
325     }
326     else if (m_type == eSymbolTypeReExported)
327     {
328         s->Printf ("                                                         0x%8.8x %s",
329                    m_flags,
330                    m_mangled.GetName().AsCString(""));
331 
332         ConstString reexport_name = GetReExportedSymbolName();
333         intptr_t shlib = m_addr_range.GetByteSize();
334         if (shlib)
335             s->Printf(" -> %s`%s\n", (const char *)shlib, reexport_name.GetCString());
336         else
337             s->Printf(" -> %s\n", reexport_name.GetCString());
338     }
339     else
340     {
341         const char *format = m_size_is_sibling ?
342                             "0x%16.16" PRIx64 "                    Sibling -> [%5llu] 0x%8.8x %s\n":
343                             "0x%16.16" PRIx64 "                    0x%16.16" PRIx64 " 0x%8.8x %s\n";
344         s->Printf(  format,
345                     m_addr_range.GetBaseAddress().GetOffset(),
346                     GetByteSize(),
347                     m_flags,
348                     m_mangled.GetName().AsCString(""));
349     }
350 }
351 
352 uint32_t
353 Symbol::GetPrologueByteSize ()
354 {
355     if (m_type == eSymbolTypeCode || m_type == eSymbolTypeResolver)
356     {
357         if (!m_type_data_resolved)
358         {
359             m_type_data_resolved = true;
360 
361             const Address &base_address = m_addr_range.GetBaseAddress();
362             Function *function = base_address.CalculateSymbolContextFunction();
363             if (function)
364             {
365                 // Functions have line entries which can also potentially have end of prologue information.
366                 // So if this symbol points to a function, use the prologue information from there.
367                 m_type_data = function->GetPrologueByteSize();
368             }
369             else
370             {
371                 ModuleSP module_sp (base_address.GetModule());
372                 SymbolContext sc;
373                 if (module_sp)
374                 {
375                     uint32_t resolved_flags = module_sp->ResolveSymbolContextForAddress (base_address,
376                                                                                          eSymbolContextLineEntry,
377                                                                                          sc);
378                     if (resolved_flags & eSymbolContextLineEntry)
379                     {
380                         // Default to the end of the first line entry.
381                         m_type_data = sc.line_entry.range.GetByteSize();
382 
383                         // Set address for next line.
384                         Address addr (base_address);
385                         addr.Slide (m_type_data);
386 
387                         // Check the first few instructions and look for one that has a line number that is
388                         // different than the first entry. This is also done in Function::GetPrologueByteSize().
389                         uint16_t total_offset = m_type_data;
390                         for (int idx = 0; idx < 6; ++idx)
391                         {
392                             SymbolContext sc_temp;
393                             resolved_flags = module_sp->ResolveSymbolContextForAddress (addr, eSymbolContextLineEntry, sc_temp);
394                             // Make sure we got line number information...
395                             if (!(resolved_flags & eSymbolContextLineEntry))
396                                 break;
397 
398                             // If this line number is different than our first one, use it and we're done.
399                             if (sc_temp.line_entry.line != sc.line_entry.line)
400                             {
401                                 m_type_data = total_offset;
402                                 break;
403                             }
404 
405                             // Slide addr up to the next line address.
406                             addr.Slide (sc_temp.line_entry.range.GetByteSize());
407                             total_offset += sc_temp.line_entry.range.GetByteSize();
408                             // If we've gone too far, bail out.
409                             if (total_offset >= m_addr_range.GetByteSize())
410                                 break;
411                         }
412 
413                         // Sanity check - this may be a function in the middle of code that has debug information, but
414                         // not for this symbol.  So the line entries surrounding us won't lie inside our function.
415                         // In that case, the line entry will be bigger than we are, so we do that quick check and
416                         // if that is true, we just return 0.
417                         if (m_type_data >= m_addr_range.GetByteSize())
418                             m_type_data = 0;
419                     }
420                     else
421                     {
422                         // TODO: expose something in Process to figure out the
423                         // size of a function prologue.
424                         m_type_data = 0;
425                     }
426                 }
427             }
428         }
429         return m_type_data;
430     }
431     return 0;
432 }
433 
434 bool
435 Symbol::Compare(const ConstString& name, SymbolType type) const
436 {
437     if (type == eSymbolTypeAny || m_type == type)
438         return m_mangled.GetMangledName() == name || m_mangled.GetDemangledName() == name;
439     return false;
440 }
441 
442 #define ENUM_TO_CSTRING(x)  case eSymbolType##x: return #x;
443 
444 const char *
445 Symbol::GetTypeAsString() const
446 {
447     switch (m_type)
448     {
449     ENUM_TO_CSTRING(Invalid);
450     ENUM_TO_CSTRING(Absolute);
451     ENUM_TO_CSTRING(Code);
452     ENUM_TO_CSTRING(Resolver);
453     ENUM_TO_CSTRING(Data);
454     ENUM_TO_CSTRING(Trampoline);
455     ENUM_TO_CSTRING(Runtime);
456     ENUM_TO_CSTRING(Exception);
457     ENUM_TO_CSTRING(SourceFile);
458     ENUM_TO_CSTRING(HeaderFile);
459     ENUM_TO_CSTRING(ObjectFile);
460     ENUM_TO_CSTRING(CommonBlock);
461     ENUM_TO_CSTRING(Block);
462     ENUM_TO_CSTRING(Local);
463     ENUM_TO_CSTRING(Param);
464     ENUM_TO_CSTRING(Variable);
465     ENUM_TO_CSTRING(VariableType);
466     ENUM_TO_CSTRING(LineEntry);
467     ENUM_TO_CSTRING(LineHeader);
468     ENUM_TO_CSTRING(ScopeBegin);
469     ENUM_TO_CSTRING(ScopeEnd);
470     ENUM_TO_CSTRING(Additional);
471     ENUM_TO_CSTRING(Compiler);
472     ENUM_TO_CSTRING(Instrumentation);
473     ENUM_TO_CSTRING(Undefined);
474     ENUM_TO_CSTRING(ObjCClass);
475     ENUM_TO_CSTRING(ObjCMetaClass);
476     ENUM_TO_CSTRING(ObjCIVar);
477     ENUM_TO_CSTRING(ReExported);
478     default:
479         break;
480     }
481     return "<unknown SymbolType>";
482 }
483 
484 void
485 Symbol::CalculateSymbolContext (SymbolContext *sc)
486 {
487     // Symbols can reconstruct the symbol and the module in the symbol context
488     sc->symbol = this;
489     if (ValueIsAddress())
490         sc->module_sp = GetAddress().GetModule();
491     else
492         sc->module_sp.reset();
493 }
494 
495 ModuleSP
496 Symbol::CalculateSymbolContextModule ()
497 {
498     if (ValueIsAddress())
499         return GetAddress().GetModule();
500     return ModuleSP();
501 }
502 
503 Symbol *
504 Symbol::CalculateSymbolContextSymbol ()
505 {
506     return this;
507 }
508 
509 void
510 Symbol::DumpSymbolContext (Stream *s)
511 {
512     bool dumped_module = false;
513     if (ValueIsAddress())
514     {
515         ModuleSP module_sp (GetAddress().GetModule());
516         if (module_sp)
517         {
518             dumped_module = true;
519             module_sp->DumpSymbolContext(s);
520         }
521     }
522     if (dumped_module)
523         s->PutCString(", ");
524 
525     s->Printf("Symbol{0x%8.8x}", GetID());
526 }
527 
528 lldb::addr_t
529 Symbol::GetByteSize () const
530 {
531     return m_addr_range.GetByteSize();
532 }
533 
534 
535 Symbol *
536 Symbol::ResolveReExportedSymbolInModuleSpec (Target &target,
537                                              ConstString &reexport_name,
538                                              ModuleSpec &module_spec,
539                                              ModuleList &seen_modules) const
540 {
541     ModuleSP module_sp;
542     if (module_spec.GetFileSpec())
543     {
544         // Try searching for the module file spec first using the full path
545         module_sp = target.GetImages().FindFirstModule(module_spec);
546         if (!module_sp)
547         {
548             // Next try and find the module by basename in case environment
549             // variables or other runtime trickery causes shared libraries
550             // to be loaded from alternate paths
551             module_spec.GetFileSpec().GetDirectory().Clear();
552             module_sp = target.GetImages().FindFirstModule(module_spec);
553         }
554     }
555 
556     if (module_sp)
557     {
558         // There should not be cycles in the reexport list, but we don't want to crash if there are so make sure
559         // we haven't seen this before:
560         if (!seen_modules.AppendIfNeeded(module_sp))
561             return nullptr;
562 
563         lldb_private::SymbolContextList sc_list;
564         module_sp->FindSymbolsWithNameAndType(reexport_name, eSymbolTypeAny, sc_list);
565         const size_t num_scs = sc_list.GetSize();
566         if (num_scs > 0)
567         {
568             for (size_t i=0; i<num_scs; ++i)
569             {
570                 lldb_private::SymbolContext sc;
571                 if (sc_list.GetContextAtIndex(i, sc))
572                 {
573                     if (sc.symbol->IsExternal())
574                         return sc.symbol;
575                 }
576             }
577         }
578         // If we didn't find the symbol in this module, it may be because this module re-exports some
579         // whole other library.  We have to search those as well:
580         seen_modules.Append(module_sp);
581 
582         FileSpecList reexported_libraries = module_sp->GetObjectFile()->GetReExportedLibraries();
583         size_t num_reexported_libraries = reexported_libraries.GetSize();
584         for (size_t idx = 0; idx < num_reexported_libraries; idx++)
585         {
586             ModuleSpec reexported_module_spec;
587             reexported_module_spec.GetFileSpec() = reexported_libraries.GetFileSpecAtIndex(idx);
588             Symbol *result_symbol = ResolveReExportedSymbolInModuleSpec(target,
589                                                                         reexport_name,
590                                                                         reexported_module_spec,
591                                                                         seen_modules);
592             if (result_symbol)
593                 return result_symbol;
594         }
595     }
596     return nullptr;
597 }
598 
599 Symbol *
600 Symbol::ResolveReExportedSymbol (Target &target) const
601 {
602     ConstString reexport_name (GetReExportedSymbolName());
603     if (reexport_name)
604     {
605         ModuleSpec module_spec;
606         ModuleList seen_modules;
607         module_spec.GetFileSpec() = GetReExportedSymbolSharedLibrary();
608         if (module_spec.GetFileSpec())
609         {
610             return ResolveReExportedSymbolInModuleSpec(target, reexport_name, module_spec, seen_modules);
611         }
612     }
613     return nullptr;
614 }
615 
616 lldb::addr_t
617 Symbol::ResolveCallableAddress(Target &target) const
618 {
619     if (GetType() == lldb::eSymbolTypeUndefined)
620         return LLDB_INVALID_ADDRESS;
621 
622     Address func_so_addr;
623 
624     bool is_indirect = IsIndirect();
625     if (GetType() == eSymbolTypeReExported)
626     {
627         Symbol *reexported_symbol = ResolveReExportedSymbol(target);
628         if (reexported_symbol)
629         {
630             func_so_addr = reexported_symbol->GetAddress();
631             is_indirect = reexported_symbol->IsIndirect();
632         }
633     }
634     else
635     {
636         func_so_addr = GetAddress();
637         is_indirect = IsIndirect();
638     }
639 
640     if (func_so_addr.IsValid())
641     {
642         if (!target.GetProcessSP() && is_indirect)
643         {
644             // can't resolve indirect symbols without calling a function...
645             return LLDB_INVALID_ADDRESS;
646         }
647 
648         lldb::addr_t load_addr = func_so_addr.GetCallableLoadAddress (&target, is_indirect);
649 
650         if (load_addr != LLDB_INVALID_ADDRESS)
651         {
652             return load_addr;
653         }
654     }
655 
656     return LLDB_INVALID_ADDRESS;
657 }
658 
659 
660 lldb::DisassemblerSP
661 Symbol::GetInstructions (const ExecutionContext &exe_ctx,
662                          const char *flavor,
663                          bool prefer_file_cache)
664 {
665     ModuleSP module_sp (m_addr_range.GetBaseAddress().GetModule());
666     if (module_sp)
667     {
668         const bool prefer_file_cache = false;
669         return Disassembler::DisassembleRange (module_sp->GetArchitecture(),
670                                                nullptr,
671                                                flavor,
672                                                exe_ctx,
673                                                m_addr_range,
674                                                prefer_file_cache);
675     }
676     return lldb::DisassemblerSP();
677 }
678 
679 bool
680 Symbol::GetDisassembly (const ExecutionContext &exe_ctx,
681                         const char *flavor,
682                         bool prefer_file_cache,
683                         Stream &strm)
684 {
685     lldb::DisassemblerSP disassembler_sp = GetInstructions (exe_ctx, flavor, prefer_file_cache);
686     if (disassembler_sp)
687     {
688         const bool show_address = true;
689         const bool show_bytes = false;
690         disassembler_sp->GetInstructionList().Dump (&strm, show_address, show_bytes, &exe_ctx);
691         return true;
692     }
693     return false;
694 }
695