ObjectFileMachO.cpp revision dd29b97f71187509df339596c3397dea0e429754 
1//===-- ObjectFileMachO.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 "llvm/ADT/StringRef.h"
11#include "llvm/Support/MachO.h"
12
13#include "ObjectFileMachO.h"
14
15#include "lldb/Core/ArchSpec.h"
16#include "lldb/Core/DataBuffer.h"
17#include "lldb/Core/FileSpecList.h"
18#include "lldb/Core/Module.h"
19#include "lldb/Core/PluginManager.h"
20#include "lldb/Core/RangeMap.h"
21#include "lldb/Core/Section.h"
22#include "lldb/Core/StreamFile.h"
23#include "lldb/Core/StreamString.h"
24#include "lldb/Core/Timer.h"
25#include "lldb/Core/UUID.h"
26#include "lldb/Host/Host.h"
27#include "lldb/Host/FileSpec.h"
28#include "lldb/Symbol/ClangNamespaceDecl.h"
29#include "lldb/Symbol/ObjectFile.h"
30#include "lldb/Target/Platform.h"
31#include "lldb/Target/Process.h"
32#include "lldb/Target/Target.h"
33#include "Plugins/Process/Utility/RegisterContextDarwin_arm.h"
34#include "Plugins/Process/Utility/RegisterContextDarwin_i386.h"
35#include "Plugins/Process/Utility/RegisterContextDarwin_x86_64.h"
36
37
38using namespace lldb;
39using namespace lldb_private;
40using namespace llvm::MachO;
41
42class RegisterContextDarwin_x86_64_Mach : public RegisterContextDarwin_x86_64
43{
44public:
45    RegisterContextDarwin_x86_64_Mach (lldb_private::Thread &thread, const DataExtractor &data) :
46        RegisterContextDarwin_x86_64 (thread, 0)
47    {
48        SetRegisterDataFrom_LC_THREAD (data);
49    }
50
51    virtual void
52    InvalidateAllRegisters ()
53    {
54        // Do nothing... registers are always valid...
55    }
56
57    void
58    SetRegisterDataFrom_LC_THREAD (const DataExtractor &data)
59    {
60        uint32_t offset = 0;
61        SetError (GPRRegSet, Read, -1);
62        SetError (FPURegSet, Read, -1);
63        SetError (EXCRegSet, Read, -1);
64        bool done = false;
65
66        while (!done)
67        {
68            int flavor = data.GetU32 (&offset);
69            if (flavor == 0)
70                done = true;
71            else
72            {
73                uint32_t i;
74                uint32_t count = data.GetU32 (&offset);
75                switch (flavor)
76                {
77                    case GPRRegSet:
78                        for (i=0; i<count; ++i)
79                            (&gpr.rax)[i] = data.GetU64(&offset);
80                        SetError (GPRRegSet, Read, 0);
81                        done = true;
82
83                        break;
84                    case FPURegSet:
85                        // TODO: fill in FPU regs....
86                        //SetError (FPURegSet, Read, -1);
87                        done = true;
88
89                        break;
90                    case EXCRegSet:
91                        exc.trapno = data.GetU32(&offset);
92                        exc.err = data.GetU32(&offset);
93                        exc.faultvaddr = data.GetU64(&offset);
94                        SetError (EXCRegSet, Read, 0);
95                        done = true;
96                        break;
97                    case 7:
98                    case 8:
99                    case 9:
100                        // fancy flavors that encapsulate of the the above
101                        // falvors...
102                        break;
103
104                    default:
105                        done = true;
106                        break;
107                }
108            }
109        }
110    }
111protected:
112    virtual int
113    DoReadGPR (lldb::tid_t tid, int flavor, GPR &gpr)
114    {
115        return 0;
116    }
117
118    virtual int
119    DoReadFPU (lldb::tid_t tid, int flavor, FPU &fpu)
120    {
121        return 0;
122    }
123
124    virtual int
125    DoReadEXC (lldb::tid_t tid, int flavor, EXC &exc)
126    {
127        return 0;
128    }
129
130    virtual int
131    DoWriteGPR (lldb::tid_t tid, int flavor, const GPR &gpr)
132    {
133        return 0;
134    }
135
136    virtual int
137    DoWriteFPU (lldb::tid_t tid, int flavor, const FPU &fpu)
138    {
139        return 0;
140    }
141
142    virtual int
143    DoWriteEXC (lldb::tid_t tid, int flavor, const EXC &exc)
144    {
145        return 0;
146    }
147};
148
149
150class RegisterContextDarwin_i386_Mach : public RegisterContextDarwin_i386
151{
152public:
153    RegisterContextDarwin_i386_Mach (lldb_private::Thread &thread, const DataExtractor &data) :
154    RegisterContextDarwin_i386 (thread, 0)
155    {
156        SetRegisterDataFrom_LC_THREAD (data);
157    }
158
159    virtual void
160    InvalidateAllRegisters ()
161    {
162        // Do nothing... registers are always valid...
163    }
164
165    void
166    SetRegisterDataFrom_LC_THREAD (const DataExtractor &data)
167    {
168        uint32_t offset = 0;
169        SetError (GPRRegSet, Read, -1);
170        SetError (FPURegSet, Read, -1);
171        SetError (EXCRegSet, Read, -1);
172        bool done = false;
173
174        while (!done)
175        {
176            int flavor = data.GetU32 (&offset);
177            if (flavor == 0)
178                done = true;
179            else
180            {
181                uint32_t i;
182                uint32_t count = data.GetU32 (&offset);
183                switch (flavor)
184                {
185                    case GPRRegSet:
186                        for (i=0; i<count; ++i)
187                            (&gpr.eax)[i] = data.GetU32(&offset);
188                        SetError (GPRRegSet, Read, 0);
189                        done = true;
190
191                        break;
192                    case FPURegSet:
193                        // TODO: fill in FPU regs....
194                        //SetError (FPURegSet, Read, -1);
195                        done = true;
196
197                        break;
198                    case EXCRegSet:
199                        exc.trapno = data.GetU32(&offset);
200                        exc.err = data.GetU32(&offset);
201                        exc.faultvaddr = data.GetU32(&offset);
202                        SetError (EXCRegSet, Read, 0);
203                        done = true;
204                        break;
205                    case 7:
206                    case 8:
207                    case 9:
208                        // fancy flavors that encapsulate of the the above
209                        // falvors...
210                        break;
211
212                    default:
213                        done = true;
214                        break;
215                }
216            }
217        }
218    }
219protected:
220    virtual int
221    DoReadGPR (lldb::tid_t tid, int flavor, GPR &gpr)
222    {
223        return 0;
224    }
225
226    virtual int
227    DoReadFPU (lldb::tid_t tid, int flavor, FPU &fpu)
228    {
229        return 0;
230    }
231
232    virtual int
233    DoReadEXC (lldb::tid_t tid, int flavor, EXC &exc)
234    {
235        return 0;
236    }
237
238    virtual int
239    DoWriteGPR (lldb::tid_t tid, int flavor, const GPR &gpr)
240    {
241        return 0;
242    }
243
244    virtual int
245    DoWriteFPU (lldb::tid_t tid, int flavor, const FPU &fpu)
246    {
247        return 0;
248    }
249
250    virtual int
251    DoWriteEXC (lldb::tid_t tid, int flavor, const EXC &exc)
252    {
253        return 0;
254    }
255};
256
257class RegisterContextDarwin_arm_Mach : public RegisterContextDarwin_arm
258{
259public:
260    RegisterContextDarwin_arm_Mach (lldb_private::Thread &thread, const DataExtractor &data) :
261    RegisterContextDarwin_arm (thread, 0)
262    {
263        SetRegisterDataFrom_LC_THREAD (data);
264    }
265
266    virtual void
267    InvalidateAllRegisters ()
268    {
269        // Do nothing... registers are always valid...
270    }
271
272    void
273    SetRegisterDataFrom_LC_THREAD (const DataExtractor &data)
274    {
275        uint32_t offset = 0;
276        SetError (GPRRegSet, Read, -1);
277        SetError (FPURegSet, Read, -1);
278        SetError (EXCRegSet, Read, -1);
279        int flavor = data.GetU32 (&offset);
280        uint32_t count = data.GetU32 (&offset);
281        switch (flavor)
282        {
283            case GPRRegSet:
284                for (uint32_t i=0; i<count; ++i)
285                    gpr.r[i] = data.GetU32(&offset);
286                SetError (GPRRegSet, Read, 0);
287                break;
288            case FPURegSet:
289                // TODO: fill in FPU regs....
290                //SetError (FPURegSet, Read, -1);
291                break;
292            case EXCRegSet:
293                exc.exception = data.GetU32(&offset);
294                exc.fsr = data.GetU32(&offset);
295                exc.far = data.GetU32(&offset);
296                SetError (EXCRegSet, Read, 0);
297                break;
298        }
299    }
300protected:
301    virtual int
302    DoReadGPR (lldb::tid_t tid, int flavor, GPR &gpr)
303    {
304        return 0;
305    }
306
307    virtual int
308    DoReadFPU (lldb::tid_t tid, int flavor, FPU &fpu)
309    {
310        return 0;
311    }
312
313    virtual int
314    DoReadEXC (lldb::tid_t tid, int flavor, EXC &exc)
315    {
316        return 0;
317    }
318
319    virtual int
320    DoWriteGPR (lldb::tid_t tid, int flavor, const GPR &gpr)
321    {
322        return 0;
323    }
324
325    virtual int
326    DoWriteFPU (lldb::tid_t tid, int flavor, const FPU &fpu)
327    {
328        return 0;
329    }
330
331    virtual int
332    DoWriteEXC (lldb::tid_t tid, int flavor, const EXC &exc)
333    {
334        return 0;
335    }
336};
337
338#define MACHO_NLIST_ARM_SYMBOL_IS_THUMB 0x0008
339
340void
341ObjectFileMachO::Initialize()
342{
343    PluginManager::RegisterPlugin (GetPluginNameStatic(),
344                                   GetPluginDescriptionStatic(),
345                                   CreateInstance,
346                                   CreateMemoryInstance);
347}
348
349void
350ObjectFileMachO::Terminate()
351{
352    PluginManager::UnregisterPlugin (CreateInstance);
353}
354
355
356const char *
357ObjectFileMachO::GetPluginNameStatic()
358{
359    return "object-file.mach-o";
360}
361
362const char *
363ObjectFileMachO::GetPluginDescriptionStatic()
364{
365    return "Mach-o object file reader (32 and 64 bit)";
366}
367
368
369ObjectFile *
370ObjectFileMachO::CreateInstance (const lldb::ModuleSP &module_sp, DataBufferSP& data_sp, const FileSpec* file, addr_t offset, addr_t length)
371{
372    if (ObjectFileMachO::MagicBytesMatch(data_sp, offset, length))
373    {
374        std::auto_ptr<ObjectFile> objfile_ap(new ObjectFileMachO (module_sp, data_sp, file, offset, length));
375        if (objfile_ap.get() && objfile_ap->ParseHeader())
376            return objfile_ap.release();
377    }
378    return NULL;
379}
380
381ObjectFile *
382ObjectFileMachO::CreateMemoryInstance (const lldb::ModuleSP &module_sp,
383                                       DataBufferSP& data_sp,
384                                       const ProcessSP &process_sp,
385                                       lldb::addr_t header_addr)
386{
387    if (ObjectFileMachO::MagicBytesMatch(data_sp, 0, data_sp->GetByteSize()))
388    {
389        std::auto_ptr<ObjectFile> objfile_ap(new ObjectFileMachO (module_sp, data_sp, process_sp, header_addr));
390        if (objfile_ap.get() && objfile_ap->ParseHeader())
391            return objfile_ap.release();
392    }
393    return NULL;
394}
395
396
397const ConstString &
398ObjectFileMachO::GetSegmentNameTEXT()
399{
400    static ConstString g_segment_name_TEXT ("__TEXT");
401    return g_segment_name_TEXT;
402}
403
404const ConstString &
405ObjectFileMachO::GetSegmentNameDATA()
406{
407    static ConstString g_segment_name_DATA ("__DATA");
408    return g_segment_name_DATA;
409}
410
411const ConstString &
412ObjectFileMachO::GetSegmentNameOBJC()
413{
414    static ConstString g_segment_name_OBJC ("__OBJC");
415    return g_segment_name_OBJC;
416}
417
418const ConstString &
419ObjectFileMachO::GetSegmentNameLINKEDIT()
420{
421    static ConstString g_section_name_LINKEDIT ("__LINKEDIT");
422    return g_section_name_LINKEDIT;
423}
424
425const ConstString &
426ObjectFileMachO::GetSectionNameEHFrame()
427{
428    static ConstString g_section_name_eh_frame ("__eh_frame");
429    return g_section_name_eh_frame;
430}
431
432
433
434static uint32_t
435MachHeaderSizeFromMagic(uint32_t magic)
436{
437    switch (magic)
438    {
439    case HeaderMagic32:
440    case HeaderMagic32Swapped:
441        return sizeof(struct mach_header);
442
443    case HeaderMagic64:
444    case HeaderMagic64Swapped:
445        return sizeof(struct mach_header_64);
446        break;
447
448    default:
449        break;
450    }
451    return 0;
452}
453
454
455bool
456ObjectFileMachO::MagicBytesMatch (DataBufferSP& data_sp,
457                                  lldb::addr_t data_offset,
458                                  lldb::addr_t data_length)
459{
460    DataExtractor data;
461    data.SetData (data_sp, data_offset, data_length);
462    uint32_t offset = 0;
463    uint32_t magic = data.GetU32(&offset);
464    return MachHeaderSizeFromMagic(magic) != 0;
465}
466
467
468ObjectFileMachO::ObjectFileMachO(const lldb::ModuleSP &module_sp, DataBufferSP& data_sp, const FileSpec* file, addr_t offset, addr_t length) :
469    ObjectFile(module_sp, file, offset, length, data_sp),
470    m_sections_ap(),
471    m_symtab_ap(),
472    m_mach_segments(),
473    m_mach_sections(),
474    m_entry_point_address(),
475    m_thread_context_offsets(),
476    m_thread_context_offsets_valid(false)
477{
478    ::memset (&m_header, 0, sizeof(m_header));
479    ::memset (&m_dysymtab, 0, sizeof(m_dysymtab));
480}
481
482ObjectFileMachO::ObjectFileMachO (const lldb::ModuleSP &module_sp,
483                                  lldb::DataBufferSP& header_data_sp,
484                                  const lldb::ProcessSP &process_sp,
485                                  lldb::addr_t header_addr) :
486    ObjectFile(module_sp, process_sp, header_addr, header_data_sp),
487    m_sections_ap(),
488    m_symtab_ap(),
489    m_mach_segments(),
490    m_mach_sections(),
491    m_entry_point_address(),
492    m_thread_context_offsets(),
493    m_thread_context_offsets_valid(false)
494{
495    ::memset (&m_header, 0, sizeof(m_header));
496    ::memset (&m_dysymtab, 0, sizeof(m_dysymtab));
497}
498
499ObjectFileMachO::~ObjectFileMachO()
500{
501}
502
503
504bool
505ObjectFileMachO::ParseHeader ()
506{
507    ModuleSP module_sp(GetModule());
508    if (module_sp)
509    {
510        lldb_private::Mutex::Locker locker(module_sp->GetMutex());
511        bool can_parse = false;
512        uint32_t offset = 0;
513        m_data.SetByteOrder (lldb::endian::InlHostByteOrder());
514        // Leave magic in the original byte order
515        m_header.magic = m_data.GetU32(&offset);
516        switch (m_header.magic)
517        {
518        case HeaderMagic32:
519            m_data.SetByteOrder (lldb::endian::InlHostByteOrder());
520            m_data.SetAddressByteSize(4);
521            can_parse = true;
522            break;
523
524        case HeaderMagic64:
525            m_data.SetByteOrder (lldb::endian::InlHostByteOrder());
526            m_data.SetAddressByteSize(8);
527            can_parse = true;
528            break;
529
530        case HeaderMagic32Swapped:
531            m_data.SetByteOrder(lldb::endian::InlHostByteOrder() == eByteOrderBig ? eByteOrderLittle : eByteOrderBig);
532            m_data.SetAddressByteSize(4);
533            can_parse = true;
534            break;
535
536        case HeaderMagic64Swapped:
537            m_data.SetByteOrder(lldb::endian::InlHostByteOrder() == eByteOrderBig ? eByteOrderLittle : eByteOrderBig);
538            m_data.SetAddressByteSize(8);
539            can_parse = true;
540            break;
541
542        default:
543            break;
544        }
545
546        if (can_parse)
547        {
548            m_data.GetU32(&offset, &m_header.cputype, 6);
549
550            ArchSpec mach_arch(eArchTypeMachO, m_header.cputype, m_header.cpusubtype);
551
552            if (SetModulesArchitecture (mach_arch))
553            {
554                const size_t header_and_lc_size = m_header.sizeofcmds + MachHeaderSizeFromMagic(m_header.magic);
555                if (m_data.GetByteSize() < header_and_lc_size)
556                {
557                    DataBufferSP data_sp;
558                    ProcessSP process_sp (m_process_wp.lock());
559                    if (process_sp)
560                    {
561                        data_sp = ReadMemory (process_sp, m_offset, header_and_lc_size);
562                    }
563                    else
564                    {
565                        // Read in all only the load command data from the file on disk
566                        data_sp = m_file.ReadFileContents(m_offset, header_and_lc_size);
567                        if (data_sp->GetByteSize() != header_and_lc_size)
568                            return false;
569                    }
570                    if (data_sp)
571                        m_data.SetData (data_sp);
572                }
573            }
574            return true;
575        }
576        else
577        {
578            memset(&m_header, 0, sizeof(struct mach_header));
579        }
580    }
581    return false;
582}
583
584
585ByteOrder
586ObjectFileMachO::GetByteOrder () const
587{
588    return m_data.GetByteOrder ();
589}
590
591bool
592ObjectFileMachO::IsExecutable() const
593{
594    return m_header.filetype == HeaderFileTypeExecutable;
595}
596
597size_t
598ObjectFileMachO::GetAddressByteSize () const
599{
600    return m_data.GetAddressByteSize ();
601}
602
603AddressClass
604ObjectFileMachO::GetAddressClass (lldb::addr_t file_addr)
605{
606    Symtab *symtab = GetSymtab();
607    if (symtab)
608    {
609        Symbol *symbol = symtab->FindSymbolContainingFileAddress(file_addr);
610        if (symbol)
611        {
612            if (symbol->ValueIsAddress())
613            {
614                SectionSP section_sp (symbol->GetAddress().GetSection());
615                if (section_sp)
616                {
617                    const SectionType section_type = section_sp->GetType();
618                    switch (section_type)
619                    {
620                    case eSectionTypeInvalid:               return eAddressClassUnknown;
621                    case eSectionTypeCode:
622                        if (m_header.cputype == llvm::MachO::CPUTypeARM)
623                        {
624                            // For ARM we have a bit in the n_desc field of the symbol
625                            // that tells us ARM/Thumb which is bit 0x0008.
626                            if (symbol->GetFlags() & MACHO_NLIST_ARM_SYMBOL_IS_THUMB)
627                                return eAddressClassCodeAlternateISA;
628                        }
629                        return eAddressClassCode;
630
631                    case eSectionTypeContainer:             return eAddressClassUnknown;
632                    case eSectionTypeData:
633                    case eSectionTypeDataCString:
634                    case eSectionTypeDataCStringPointers:
635                    case eSectionTypeDataSymbolAddress:
636                    case eSectionTypeData4:
637                    case eSectionTypeData8:
638                    case eSectionTypeData16:
639                    case eSectionTypeDataPointers:
640                    case eSectionTypeZeroFill:
641                    case eSectionTypeDataObjCMessageRefs:
642                    case eSectionTypeDataObjCCFStrings:
643                        return eAddressClassData;
644                    case eSectionTypeDebug:
645                    case eSectionTypeDWARFDebugAbbrev:
646                    case eSectionTypeDWARFDebugAranges:
647                    case eSectionTypeDWARFDebugFrame:
648                    case eSectionTypeDWARFDebugInfo:
649                    case eSectionTypeDWARFDebugLine:
650                    case eSectionTypeDWARFDebugLoc:
651                    case eSectionTypeDWARFDebugMacInfo:
652                    case eSectionTypeDWARFDebugPubNames:
653                    case eSectionTypeDWARFDebugPubTypes:
654                    case eSectionTypeDWARFDebugRanges:
655                    case eSectionTypeDWARFDebugStr:
656                    case eSectionTypeDWARFAppleNames:
657                    case eSectionTypeDWARFAppleTypes:
658                    case eSectionTypeDWARFAppleNamespaces:
659                    case eSectionTypeDWARFAppleObjC:
660                        return eAddressClassDebug;
661                    case eSectionTypeEHFrame:               return eAddressClassRuntime;
662                    case eSectionTypeOther:                 return eAddressClassUnknown;
663                    }
664                }
665            }
666
667            const SymbolType symbol_type = symbol->GetType();
668            switch (symbol_type)
669            {
670            case eSymbolTypeAny:            return eAddressClassUnknown;
671            case eSymbolTypeAbsolute:       return eAddressClassUnknown;
672
673            case eSymbolTypeCode:
674            case eSymbolTypeTrampoline:
675                if (m_header.cputype == llvm::MachO::CPUTypeARM)
676                {
677                    // For ARM we have a bit in the n_desc field of the symbol
678                    // that tells us ARM/Thumb which is bit 0x0008.
679                    if (symbol->GetFlags() & MACHO_NLIST_ARM_SYMBOL_IS_THUMB)
680                        return eAddressClassCodeAlternateISA;
681                }
682                return eAddressClassCode;
683
684            case eSymbolTypeData:           return eAddressClassData;
685            case eSymbolTypeRuntime:        return eAddressClassRuntime;
686            case eSymbolTypeException:      return eAddressClassRuntime;
687            case eSymbolTypeSourceFile:     return eAddressClassDebug;
688            case eSymbolTypeHeaderFile:     return eAddressClassDebug;
689            case eSymbolTypeObjectFile:     return eAddressClassDebug;
690            case eSymbolTypeCommonBlock:    return eAddressClassDebug;
691            case eSymbolTypeBlock:          return eAddressClassDebug;
692            case eSymbolTypeLocal:          return eAddressClassData;
693            case eSymbolTypeParam:          return eAddressClassData;
694            case eSymbolTypeVariable:       return eAddressClassData;
695            case eSymbolTypeVariableType:   return eAddressClassDebug;
696            case eSymbolTypeLineEntry:      return eAddressClassDebug;
697            case eSymbolTypeLineHeader:     return eAddressClassDebug;
698            case eSymbolTypeScopeBegin:     return eAddressClassDebug;
699            case eSymbolTypeScopeEnd:       return eAddressClassDebug;
700            case eSymbolTypeAdditional:     return eAddressClassUnknown;
701            case eSymbolTypeCompiler:       return eAddressClassDebug;
702            case eSymbolTypeInstrumentation:return eAddressClassDebug;
703            case eSymbolTypeUndefined:      return eAddressClassUnknown;
704            case eSymbolTypeObjCClass:      return eAddressClassRuntime;
705            case eSymbolTypeObjCMetaClass:  return eAddressClassRuntime;
706            case eSymbolTypeObjCIVar:       return eAddressClassRuntime;
707            }
708        }
709    }
710    return eAddressClassUnknown;
711}
712
713Symtab *
714ObjectFileMachO::GetSymtab()
715{
716    ModuleSP module_sp(GetModule());
717    if (module_sp)
718    {
719        lldb_private::Mutex::Locker locker(module_sp->GetMutex());
720        if (m_symtab_ap.get() == NULL)
721        {
722            m_symtab_ap.reset(new Symtab(this));
723            Mutex::Locker symtab_locker (m_symtab_ap->GetMutex());
724            ParseSymtab (true);
725            m_symtab_ap->Finalize ();
726        }
727    }
728    return m_symtab_ap.get();
729}
730
731
732SectionList *
733ObjectFileMachO::GetSectionList()
734{
735    ModuleSP module_sp(GetModule());
736    if (module_sp)
737    {
738        lldb_private::Mutex::Locker locker(module_sp->GetMutex());
739        if (m_sections_ap.get() == NULL)
740        {
741            m_sections_ap.reset(new SectionList());
742            ParseSections();
743        }
744    }
745    return m_sections_ap.get();
746}
747
748
749size_t
750ObjectFileMachO::ParseSections ()
751{
752    lldb::user_id_t segID = 0;
753    lldb::user_id_t sectID = 0;
754    struct segment_command_64 load_cmd;
755    uint32_t offset = MachHeaderSizeFromMagic(m_header.magic);
756    uint32_t i;
757    const bool is_core = GetType() == eTypeCoreFile;
758    //bool dump_sections = false;
759    ModuleSP module_sp (GetModule());
760    // First look up any LC_ENCRYPTION_INFO load commands
761    typedef RangeArray<uint32_t, uint32_t, 8> EncryptedFileRanges;
762    EncryptedFileRanges encrypted_file_ranges;
763    for (i=0; i<m_header.ncmds; ++i)
764    {
765        const uint32_t load_cmd_offset = offset;
766        if (m_data.GetU32(&offset, &load_cmd, 2) == NULL)
767            break;
768
769        if (load_cmd.cmd == LoadCommandEncryptionInfo)
770        {
771            EncryptedFileRanges::Entry entry;
772            entry.SetRangeBase(m_data.GetU32(&offset));
773            entry.SetByteSize(m_data.GetU32(&offset));
774            encrypted_file_ranges.Append(entry);
775        }
776        offset = load_cmd_offset + load_cmd.cmdsize;
777    }
778
779    offset = MachHeaderSizeFromMagic(m_header.magic);
780
781    for (i=0; i<m_header.ncmds; ++i)
782    {
783        const uint32_t load_cmd_offset = offset;
784        if (m_data.GetU32(&offset, &load_cmd, 2) == NULL)
785            break;
786
787        if (load_cmd.cmd == LoadCommandSegment32 || load_cmd.cmd == LoadCommandSegment64)
788        {
789            if (m_data.GetU8(&offset, (uint8_t*)load_cmd.segname, 16))
790            {
791                load_cmd.vmaddr = m_data.GetAddress(&offset);
792                load_cmd.vmsize = m_data.GetAddress(&offset);
793                load_cmd.fileoff = m_data.GetAddress(&offset);
794                load_cmd.filesize = m_data.GetAddress(&offset);
795                if (m_data.GetU32(&offset, &load_cmd.maxprot, 4))
796                {
797
798                    const bool segment_is_encrypted = (load_cmd.flags & SegmentCommandFlagBitProtectedVersion1) != 0;
799
800                    // Keep a list of mach segments around in case we need to
801                    // get at data that isn't stored in the abstracted Sections.
802                    m_mach_segments.push_back (load_cmd);
803
804                    ConstString segment_name (load_cmd.segname, std::min<int>(strlen(load_cmd.segname), sizeof(load_cmd.segname)));
805                    // Use a segment ID of the segment index shifted left by 8 so they
806                    // never conflict with any of the sections.
807                    SectionSP segment_sp;
808                    if (segment_name || is_core)
809                    {
810                        segment_sp.reset(new Section (module_sp,              // Module to which this section belongs
811                                                      ++segID << 8,           // Section ID is the 1 based segment index shifted right by 8 bits as not to collide with any of the 256 section IDs that are possible
812                                                      segment_name,           // Name of this section
813                                                      eSectionTypeContainer,  // This section is a container of other sections.
814                                                      load_cmd.vmaddr,        // File VM address == addresses as they are found in the object file
815                                                      load_cmd.vmsize,        // VM size in bytes of this section
816                                                      load_cmd.fileoff,       // Offset to the data for this section in the file
817                                                      load_cmd.filesize,      // Size in bytes of this section as found in the the file
818                                                      load_cmd.flags));       // Flags for this section
819
820                        segment_sp->SetIsEncrypted (segment_is_encrypted);
821                        m_sections_ap->AddSection(segment_sp);
822                    }
823
824                    struct section_64 sect64;
825                    ::memset (&sect64, 0, sizeof(sect64));
826                    // Push a section into our mach sections for the section at
827                    // index zero (NListSectionNoSection) if we don't have any
828                    // mach sections yet...
829                    if (m_mach_sections.empty())
830                        m_mach_sections.push_back(sect64);
831                    uint32_t segment_sect_idx;
832                    const lldb::user_id_t first_segment_sectID = sectID + 1;
833
834
835                    const uint32_t num_u32s = load_cmd.cmd == LoadCommandSegment32 ? 7 : 8;
836                    for (segment_sect_idx=0; segment_sect_idx<load_cmd.nsects; ++segment_sect_idx)
837                    {
838                        if (m_data.GetU8(&offset, (uint8_t*)sect64.sectname, sizeof(sect64.sectname)) == NULL)
839                            break;
840                        if (m_data.GetU8(&offset, (uint8_t*)sect64.segname, sizeof(sect64.segname)) == NULL)
841                            break;
842                        sect64.addr = m_data.GetAddress(&offset);
843                        sect64.size = m_data.GetAddress(&offset);
844
845                        if (m_data.GetU32(&offset, &sect64.offset, num_u32s) == NULL)
846                            break;
847
848                        // Keep a list of mach sections around in case we need to
849                        // get at data that isn't stored in the abstracted Sections.
850                        m_mach_sections.push_back (sect64);
851
852                        ConstString section_name (sect64.sectname, std::min<size_t>(strlen(sect64.sectname), sizeof(sect64.sectname)));
853                        if (!segment_name)
854                        {
855                            // We have a segment with no name so we need to conjure up
856                            // segments that correspond to the section's segname if there
857                            // isn't already such a section. If there is such a section,
858                            // we resize the section so that it spans all sections.
859                            // We also mark these sections as fake so address matches don't
860                            // hit if they land in the gaps between the child sections.
861                            segment_name.SetTrimmedCStringWithLength(sect64.segname, sizeof(sect64.segname));
862                            segment_sp = m_sections_ap->FindSectionByName (segment_name);
863                            if (segment_sp.get())
864                            {
865                                Section *segment = segment_sp.get();
866                                // Grow the section size as needed.
867                                const lldb::addr_t sect64_min_addr = sect64.addr;
868                                const lldb::addr_t sect64_max_addr = sect64_min_addr + sect64.size;
869                                const lldb::addr_t curr_seg_byte_size = segment->GetByteSize();
870                                const lldb::addr_t curr_seg_min_addr = segment->GetFileAddress();
871                                const lldb::addr_t curr_seg_max_addr = curr_seg_min_addr + curr_seg_byte_size;
872                                if (sect64_min_addr >= curr_seg_min_addr)
873                                {
874                                    const lldb::addr_t new_seg_byte_size = sect64_max_addr - curr_seg_min_addr;
875                                    // Only grow the section size if needed
876                                    if (new_seg_byte_size > curr_seg_byte_size)
877                                        segment->SetByteSize (new_seg_byte_size);
878                                }
879                                else
880                                {
881                                    // We need to change the base address of the segment and
882                                    // adjust the child section offsets for all existing children.
883                                    const lldb::addr_t slide_amount = sect64_min_addr - curr_seg_min_addr;
884                                    segment->Slide(slide_amount, false);
885                                    segment->GetChildren().Slide (-slide_amount, false);
886                                    segment->SetByteSize (curr_seg_max_addr - sect64_min_addr);
887                                }
888
889                                // Grow the section size as needed.
890                                if (sect64.offset)
891                                {
892                                    const lldb::addr_t segment_min_file_offset = segment->GetFileOffset();
893                                    const lldb::addr_t segment_max_file_offset = segment_min_file_offset + segment->GetFileSize();
894
895                                    const lldb::addr_t section_min_file_offset = sect64.offset;
896                                    const lldb::addr_t section_max_file_offset = section_min_file_offset + sect64.size;
897                                    const lldb::addr_t new_file_offset = std::min (section_min_file_offset, segment_min_file_offset);
898                                    const lldb::addr_t new_file_size = std::max (section_max_file_offset, segment_max_file_offset) - new_file_offset;
899                                    segment->SetFileOffset (new_file_offset);
900                                    segment->SetFileSize (new_file_size);
901                                }
902                            }
903                            else
904                            {
905                                // Create a fake section for the section's named segment
906                                segment_sp.reset(new Section (segment_sp,            // Parent section
907                                                              module_sp,           // Module to which this section belongs
908                                                              ++segID << 8,          // Section ID is the 1 based segment index shifted right by 8 bits as not to collide with any of the 256 section IDs that are possible
909                                                              segment_name,          // Name of this section
910                                                              eSectionTypeContainer, // This section is a container of other sections.
911                                                              sect64.addr,           // File VM address == addresses as they are found in the object file
912                                                              sect64.size,           // VM size in bytes of this section
913                                                              sect64.offset,         // Offset to the data for this section in the file
914                                                              sect64.offset ? sect64.size : 0,        // Size in bytes of this section as found in the the file
915                                                              load_cmd.flags));      // Flags for this section
916                                segment_sp->SetIsFake(true);
917                                m_sections_ap->AddSection(segment_sp);
918                                segment_sp->SetIsEncrypted (segment_is_encrypted);
919                            }
920                        }
921                        assert (segment_sp.get());
922
923                        uint32_t mach_sect_type = sect64.flags & SectionFlagMaskSectionType;
924                        static ConstString g_sect_name_objc_data ("__objc_data");
925                        static ConstString g_sect_name_objc_msgrefs ("__objc_msgrefs");
926                        static ConstString g_sect_name_objc_selrefs ("__objc_selrefs");
927                        static ConstString g_sect_name_objc_classrefs ("__objc_classrefs");
928                        static ConstString g_sect_name_objc_superrefs ("__objc_superrefs");
929                        static ConstString g_sect_name_objc_const ("__objc_const");
930                        static ConstString g_sect_name_objc_classlist ("__objc_classlist");
931                        static ConstString g_sect_name_cfstring ("__cfstring");
932
933                        static ConstString g_sect_name_dwarf_debug_abbrev ("__debug_abbrev");
934                        static ConstString g_sect_name_dwarf_debug_aranges ("__debug_aranges");
935                        static ConstString g_sect_name_dwarf_debug_frame ("__debug_frame");
936                        static ConstString g_sect_name_dwarf_debug_info ("__debug_info");
937                        static ConstString g_sect_name_dwarf_debug_line ("__debug_line");
938                        static ConstString g_sect_name_dwarf_debug_loc ("__debug_loc");
939                        static ConstString g_sect_name_dwarf_debug_macinfo ("__debug_macinfo");
940                        static ConstString g_sect_name_dwarf_debug_pubnames ("__debug_pubnames");
941                        static ConstString g_sect_name_dwarf_debug_pubtypes ("__debug_pubtypes");
942                        static ConstString g_sect_name_dwarf_debug_ranges ("__debug_ranges");
943                        static ConstString g_sect_name_dwarf_debug_str ("__debug_str");
944                        static ConstString g_sect_name_dwarf_apple_names ("__apple_names");
945                        static ConstString g_sect_name_dwarf_apple_types ("__apple_types");
946                        static ConstString g_sect_name_dwarf_apple_namespaces ("__apple_namespac");
947                        static ConstString g_sect_name_dwarf_apple_objc ("__apple_objc");
948                        static ConstString g_sect_name_eh_frame ("__eh_frame");
949                        static ConstString g_sect_name_DATA ("__DATA");
950                        static ConstString g_sect_name_TEXT ("__TEXT");
951
952                        SectionType sect_type = eSectionTypeOther;
953
954                        if (section_name == g_sect_name_dwarf_debug_abbrev)
955                            sect_type = eSectionTypeDWARFDebugAbbrev;
956                        else if (section_name == g_sect_name_dwarf_debug_aranges)
957                            sect_type = eSectionTypeDWARFDebugAranges;
958                        else if (section_name == g_sect_name_dwarf_debug_frame)
959                            sect_type = eSectionTypeDWARFDebugFrame;
960                        else if (section_name == g_sect_name_dwarf_debug_info)
961                            sect_type = eSectionTypeDWARFDebugInfo;
962                        else if (section_name == g_sect_name_dwarf_debug_line)
963                            sect_type = eSectionTypeDWARFDebugLine;
964                        else if (section_name == g_sect_name_dwarf_debug_loc)
965                            sect_type = eSectionTypeDWARFDebugLoc;
966                        else if (section_name == g_sect_name_dwarf_debug_macinfo)
967                            sect_type = eSectionTypeDWARFDebugMacInfo;
968                        else if (section_name == g_sect_name_dwarf_debug_pubnames)
969                            sect_type = eSectionTypeDWARFDebugPubNames;
970                        else if (section_name == g_sect_name_dwarf_debug_pubtypes)
971                            sect_type = eSectionTypeDWARFDebugPubTypes;
972                        else if (section_name == g_sect_name_dwarf_debug_ranges)
973                            sect_type = eSectionTypeDWARFDebugRanges;
974                        else if (section_name == g_sect_name_dwarf_debug_str)
975                            sect_type = eSectionTypeDWARFDebugStr;
976                        else if (section_name == g_sect_name_dwarf_apple_names)
977                            sect_type = eSectionTypeDWARFAppleNames;
978                        else if (section_name == g_sect_name_dwarf_apple_types)
979                            sect_type = eSectionTypeDWARFAppleTypes;
980                        else if (section_name == g_sect_name_dwarf_apple_namespaces)
981                            sect_type = eSectionTypeDWARFAppleNamespaces;
982                        else if (section_name == g_sect_name_dwarf_apple_objc)
983                            sect_type = eSectionTypeDWARFAppleObjC;
984                        else if (section_name == g_sect_name_objc_selrefs)
985                            sect_type = eSectionTypeDataCStringPointers;
986                        else if (section_name == g_sect_name_objc_msgrefs)
987                            sect_type = eSectionTypeDataObjCMessageRefs;
988                        else if (section_name == g_sect_name_eh_frame)
989                            sect_type = eSectionTypeEHFrame;
990                        else if (section_name == g_sect_name_cfstring)
991                            sect_type = eSectionTypeDataObjCCFStrings;
992                        else if (section_name == g_sect_name_objc_data ||
993                                 section_name == g_sect_name_objc_classrefs ||
994                                 section_name == g_sect_name_objc_superrefs ||
995                                 section_name == g_sect_name_objc_const ||
996                                 section_name == g_sect_name_objc_classlist)
997                        {
998                            sect_type = eSectionTypeDataPointers;
999                        }
1000
1001                        if (sect_type == eSectionTypeOther)
1002                        {
1003                            switch (mach_sect_type)
1004                            {
1005                            // TODO: categorize sections by other flags for regular sections
1006                            case SectionTypeRegular:
1007                                if (segment_sp->GetName() == g_sect_name_TEXT)
1008                                    sect_type = eSectionTypeCode;
1009                                else if (segment_sp->GetName() == g_sect_name_DATA)
1010                                    sect_type = eSectionTypeData;
1011                                else
1012                                    sect_type = eSectionTypeOther;
1013                                break;
1014                            case SectionTypeZeroFill:                   sect_type = eSectionTypeZeroFill; break;
1015                            case SectionTypeCStringLiterals:            sect_type = eSectionTypeDataCString;    break; // section with only literal C strings
1016                            case SectionType4ByteLiterals:              sect_type = eSectionTypeData4;    break; // section with only 4 byte literals
1017                            case SectionType8ByteLiterals:              sect_type = eSectionTypeData8;    break; // section with only 8 byte literals
1018                            case SectionTypeLiteralPointers:            sect_type = eSectionTypeDataPointers;  break; // section with only pointers to literals
1019                            case SectionTypeNonLazySymbolPointers:      sect_type = eSectionTypeDataPointers;  break; // section with only non-lazy symbol pointers
1020                            case SectionTypeLazySymbolPointers:         sect_type = eSectionTypeDataPointers;  break; // section with only lazy symbol pointers
1021                            case SectionTypeSymbolStubs:                sect_type = eSectionTypeCode;  break; // section with only symbol stubs, byte size of stub in the reserved2 field
1022                            case SectionTypeModuleInitFunctionPointers: sect_type = eSectionTypeDataPointers;    break; // section with only function pointers for initialization
1023                            case SectionTypeModuleTermFunctionPointers: sect_type = eSectionTypeDataPointers; break; // section with only function pointers for termination
1024                            case SectionTypeCoalesced:                  sect_type = eSectionTypeOther; break;
1025                            case SectionTypeZeroFillLarge:              sect_type = eSectionTypeZeroFill; break;
1026                            case SectionTypeInterposing:                sect_type = eSectionTypeCode;  break; // section with only pairs of function pointers for interposing
1027                            case SectionType16ByteLiterals:             sect_type = eSectionTypeData16; break; // section with only 16 byte literals
1028                            case SectionTypeDTraceObjectFormat:         sect_type = eSectionTypeDebug; break;
1029                            case SectionTypeLazyDylibSymbolPointers:    sect_type = eSectionTypeDataPointers;  break;
1030                            default: break;
1031                            }
1032                        }
1033
1034                        SectionSP section_sp(new Section (segment_sp,
1035                                                          module_sp,
1036                                                          ++sectID,
1037                                                          section_name,
1038                                                          sect_type,
1039                                                          sect64.addr - segment_sp->GetFileAddress(),
1040                                                          sect64.size,
1041                                                          sect64.offset,
1042                                                          sect64.offset == 0 ? 0 : sect64.size,
1043                                                          sect64.flags));
1044                        // Set the section to be encrypted to match the segment
1045
1046                        bool section_is_encrypted = false;
1047                        if (!segment_is_encrypted && load_cmd.filesize != 0)
1048                            section_is_encrypted = encrypted_file_ranges.FindEntryThatContains(sect64.offset) != NULL;
1049
1050                        section_sp->SetIsEncrypted (segment_is_encrypted || section_is_encrypted);
1051                        segment_sp->GetChildren().AddSection(section_sp);
1052
1053                        if (segment_sp->IsFake())
1054                        {
1055                            segment_sp.reset();
1056                            segment_name.Clear();
1057                        }
1058                    }
1059                    if (segment_sp && m_header.filetype == HeaderFileTypeDSYM)
1060                    {
1061                        if (first_segment_sectID <= sectID)
1062                        {
1063                            lldb::user_id_t sect_uid;
1064                            for (sect_uid = first_segment_sectID; sect_uid <= sectID; ++sect_uid)
1065                            {
1066                                SectionSP curr_section_sp(segment_sp->GetChildren().FindSectionByID (sect_uid));
1067                                SectionSP next_section_sp;
1068                                if (sect_uid + 1 <= sectID)
1069                                    next_section_sp = segment_sp->GetChildren().FindSectionByID (sect_uid+1);
1070
1071                                if (curr_section_sp.get())
1072                                {
1073                                    if (curr_section_sp->GetByteSize() == 0)
1074                                    {
1075                                        if (next_section_sp.get() != NULL)
1076                                            curr_section_sp->SetByteSize ( next_section_sp->GetFileAddress() - curr_section_sp->GetFileAddress() );
1077                                        else
1078                                            curr_section_sp->SetByteSize ( load_cmd.vmsize );
1079                                    }
1080                                }
1081                            }
1082                        }
1083                    }
1084                }
1085            }
1086        }
1087        else if (load_cmd.cmd == LoadCommandDynamicSymtabInfo)
1088        {
1089            m_dysymtab.cmd = load_cmd.cmd;
1090            m_dysymtab.cmdsize = load_cmd.cmdsize;
1091            m_data.GetU32 (&offset, &m_dysymtab.ilocalsym, (sizeof(m_dysymtab) / sizeof(uint32_t)) - 2);
1092        }
1093
1094        offset = load_cmd_offset + load_cmd.cmdsize;
1095    }
1096//    if (dump_sections)
1097//    {
1098//        StreamFile s(stdout);
1099//        m_sections_ap->Dump(&s, true);
1100//    }
1101    return sectID;  // Return the number of sections we registered with the module
1102}
1103
1104class MachSymtabSectionInfo
1105{
1106public:
1107
1108    MachSymtabSectionInfo (SectionList *section_list) :
1109        m_section_list (section_list),
1110        m_section_infos()
1111    {
1112        // Get the number of sections down to a depth of 1 to include
1113        // all segments and their sections, but no other sections that
1114        // may be added for debug map or
1115        m_section_infos.resize(section_list->GetNumSections(1));
1116    }
1117
1118
1119    SectionSP
1120    GetSection (uint8_t n_sect, addr_t file_addr)
1121    {
1122        if (n_sect == 0)
1123            return SectionSP();
1124        if (n_sect < m_section_infos.size())
1125        {
1126            if (!m_section_infos[n_sect].section_sp)
1127            {
1128                SectionSP section_sp (m_section_list->FindSectionByID (n_sect));
1129                m_section_infos[n_sect].section_sp = section_sp;
1130                if (section_sp != NULL)
1131                {
1132                    m_section_infos[n_sect].vm_range.SetBaseAddress (section_sp->GetFileAddress());
1133                    m_section_infos[n_sect].vm_range.SetByteSize (section_sp->GetByteSize());
1134                }
1135                else
1136                {
1137                    Host::SystemLog (Host::eSystemLogError, "error: unable to find section for section %u\n", n_sect);
1138                }
1139            }
1140            if (m_section_infos[n_sect].vm_range.Contains(file_addr))
1141            {
1142                // Symbol is in section.
1143                return m_section_infos[n_sect].section_sp;
1144            }
1145            else if (m_section_infos[n_sect].vm_range.GetByteSize () == 0 &&
1146                     m_section_infos[n_sect].vm_range.GetBaseAddress() == file_addr)
1147            {
1148                // Symbol is in section with zero size, but has the same start
1149                // address as the section. This can happen with linker symbols
1150                // (symbols that start with the letter 'l' or 'L'.
1151                return m_section_infos[n_sect].section_sp;
1152            }
1153        }
1154        return m_section_list->FindSectionContainingFileAddress(file_addr);
1155    }
1156
1157protected:
1158    struct SectionInfo
1159    {
1160        SectionInfo () :
1161            vm_range(),
1162            section_sp ()
1163        {
1164        }
1165
1166        VMRange vm_range;
1167        SectionSP section_sp;
1168    };
1169    SectionList *m_section_list;
1170    std::vector<SectionInfo> m_section_infos;
1171};
1172
1173
1174
1175size_t
1176ObjectFileMachO::ParseSymtab (bool minimize)
1177{
1178    Timer scoped_timer(__PRETTY_FUNCTION__,
1179                       "ObjectFileMachO::ParseSymtab () module = %s",
1180                       m_file.GetFilename().AsCString(""));
1181    ModuleSP module_sp (GetModule());
1182    if (!module_sp)
1183        return 0;
1184
1185    struct symtab_command symtab_load_command = { 0, 0, 0, 0, 0, 0 };
1186    struct linkedit_data_command function_starts_load_command = { 0, 0, 0, 0 };
1187    typedef AddressDataArray<lldb::addr_t, bool, 100> FunctionStarts;
1188    FunctionStarts function_starts;
1189    uint32_t offset = MachHeaderSizeFromMagic(m_header.magic);
1190    uint32_t i;
1191
1192    LogSP log(lldb_private::GetLogIfAllCategoriesSet (LIBLLDB_LOG_SYMBOLS));
1193
1194    for (i=0; i<m_header.ncmds; ++i)
1195    {
1196        const uint32_t cmd_offset = offset;
1197        // Read in the load command and load command size
1198        struct load_command lc;
1199        if (m_data.GetU32(&offset, &lc, 2) == NULL)
1200            break;
1201        // Watch for the symbol table load command
1202        switch (lc.cmd)
1203        {
1204        case LoadCommandSymtab:
1205            symtab_load_command.cmd = lc.cmd;
1206            symtab_load_command.cmdsize = lc.cmdsize;
1207            // Read in the rest of the symtab load command
1208            if (m_data.GetU32(&offset, &symtab_load_command.symoff, 4) == 0) // fill in symoff, nsyms, stroff, strsize fields
1209                return 0;
1210            if (symtab_load_command.symoff == 0)
1211            {
1212                if (log)
1213                    module_sp->LogMessage(log.get(), "LC_SYMTAB.symoff == 0");
1214                return 0;
1215            }
1216
1217            if (symtab_load_command.stroff == 0)
1218            {
1219                if (log)
1220                    module_sp->LogMessage(log.get(), "LC_SYMTAB.stroff == 0");
1221                return 0;
1222            }
1223
1224            if (symtab_load_command.nsyms == 0)
1225            {
1226                if (log)
1227                    module_sp->LogMessage(log.get(), "LC_SYMTAB.nsyms == 0");
1228                return 0;
1229            }
1230
1231            if (symtab_load_command.strsize == 0)
1232            {
1233                if (log)
1234                    module_sp->LogMessage(log.get(), "LC_SYMTAB.strsize == 0");
1235                return 0;
1236            }
1237            break;
1238
1239        case LoadCommandFunctionStarts:
1240            function_starts_load_command.cmd = lc.cmd;
1241            function_starts_load_command.cmdsize = lc.cmdsize;
1242            if (m_data.GetU32(&offset, &function_starts_load_command.dataoff, 2) == NULL) // fill in symoff, nsyms, stroff, strsize fields
1243                bzero (&function_starts_load_command, sizeof(function_starts_load_command));
1244            break;
1245
1246        default:
1247            break;
1248        }
1249        offset = cmd_offset + lc.cmdsize;
1250    }
1251
1252    if (symtab_load_command.cmd)
1253    {
1254        Symtab *symtab = m_symtab_ap.get();
1255        SectionList *section_list = GetSectionList();
1256        if (section_list == NULL)
1257            return 0;
1258
1259        ProcessSP process_sp (m_process_wp.lock());
1260        Process *process = process_sp.get();
1261
1262        const size_t addr_byte_size = m_data.GetAddressByteSize();
1263        bool bit_width_32 = addr_byte_size == 4;
1264        const size_t nlist_byte_size = bit_width_32 ? sizeof(struct nlist) : sizeof(struct nlist_64);
1265
1266        DataExtractor nlist_data (NULL, 0, m_data.GetByteOrder(), m_data.GetAddressByteSize());
1267        DataExtractor strtab_data (NULL, 0, m_data.GetByteOrder(), m_data.GetAddressByteSize());
1268        DataExtractor function_starts_data (NULL, 0, m_data.GetByteOrder(), m_data.GetAddressByteSize());
1269
1270        const addr_t nlist_data_byte_size = symtab_load_command.nsyms * nlist_byte_size;
1271        const addr_t strtab_data_byte_size = symtab_load_command.strsize;
1272        addr_t strtab_addr = LLDB_INVALID_ADDRESS;
1273        if (process)
1274        {
1275            Target &target = process->GetTarget();
1276            SectionSP linkedit_section_sp(section_list->FindSectionByName(GetSegmentNameLINKEDIT()));
1277            // Reading mach file from memory in a process or core file...
1278
1279            if (linkedit_section_sp)
1280            {
1281                const addr_t linkedit_load_addr = linkedit_section_sp->GetLoadBaseAddress(&target);
1282                const addr_t linkedit_file_offset = linkedit_section_sp->GetFileOffset();
1283                const addr_t symoff_addr = linkedit_load_addr + symtab_load_command.symoff - linkedit_file_offset;
1284                strtab_addr = linkedit_load_addr + symtab_load_command.stroff - linkedit_file_offset;
1285
1286                bool data_was_read = false;
1287
1288#if defined (__APPLE__) && defined (__arm__)
1289                if (m_header.flags & 0x80000000u)
1290                {
1291                    // This mach-o memory file is in the dyld shared cache. If this
1292                    // program is not remote and this is iOS, then this process will
1293                    // share the same shared cache as the process we are debugging and
1294                    // we can read the entire __LINKEDIT from the address space in this
1295                    // process. This is a needed optimization that is used for local iOS
1296                    // debugging only since all shared libraries in the shared cache do
1297                    // not have corresponding files that exist in the file system of the
1298                    // device. They have been combined into a single file. This means we
1299                    // always have to load these files from memory. All of the symbol and
1300                    // string tables from all of the __LINKEDIT sections from the shared
1301                    // libraries in the shared cache have been merged into a single large
1302                    // symbol and string table. Reading all of this symbol and string table
1303                    // data across can slow down debug launch times, so we optimize this by
1304                    // reading the memory for the __LINKEDIT section from this process.
1305                    PlatformSP platform_sp (target.GetPlatform());
1306                    if (platform_sp && platform_sp->IsHost())
1307                    {
1308                        data_was_read = true;
1309                        nlist_data.SetData((void *)symoff_addr, nlist_data_byte_size, eByteOrderLittle);
1310                        strtab_data.SetData((void *)strtab_addr, strtab_data_byte_size, eByteOrderLittle);
1311                        if (function_starts_load_command.cmd)
1312                        {
1313                            const addr_t func_start_addr = linkedit_load_addr + function_starts_load_command.dataoff - linkedit_file_offset;
1314                            function_starts_data.SetData ((void *)func_start_addr, function_starts_load_command.datasize, eByteOrderLittle);
1315                        }
1316                    }
1317                }
1318#endif
1319
1320                if (!data_was_read)
1321                {
1322                    DataBufferSP nlist_data_sp (ReadMemory (process_sp, symoff_addr, nlist_data_byte_size));
1323                    if (nlist_data_sp)
1324                        nlist_data.SetData (nlist_data_sp, 0, nlist_data_sp->GetByteSize());
1325                    //DataBufferSP strtab_data_sp (ReadMemory (process_sp, strtab_addr, strtab_data_byte_size));
1326                    //if (strtab_data_sp)
1327                    //    strtab_data.SetData (strtab_data_sp, 0, strtab_data_sp->GetByteSize());
1328                    if (function_starts_load_command.cmd)
1329                    {
1330                        const addr_t func_start_addr = linkedit_load_addr + function_starts_load_command.dataoff - linkedit_file_offset;
1331                        DataBufferSP func_start_data_sp (ReadMemory (process_sp, func_start_addr, function_starts_load_command.datasize));
1332                        if (func_start_data_sp)
1333                            function_starts_data.SetData (func_start_data_sp, 0, func_start_data_sp->GetByteSize());
1334                    }
1335                }
1336            }
1337        }
1338        else
1339        {
1340            nlist_data.SetData (m_data,
1341                                symtab_load_command.symoff,
1342                                nlist_data_byte_size);
1343            strtab_data.SetData (m_data,
1344                                 symtab_load_command.stroff,
1345                                 strtab_data_byte_size);
1346            if (function_starts_load_command.cmd)
1347            {
1348                function_starts_data.SetData (m_data,
1349                                              function_starts_load_command.dataoff,
1350                                              function_starts_load_command.datasize);
1351            }
1352        }
1353
1354        if (nlist_data.GetByteSize() == 0)
1355        {
1356            if (log)
1357                module_sp->LogMessage(log.get(), "failed to read nlist data");
1358            return 0;
1359        }
1360
1361
1362        if (process)
1363        {
1364            if (strtab_addr == LLDB_INVALID_ADDRESS)
1365            {
1366                if (log)
1367                    module_sp->LogMessage(log.get(), "failed to locate the strtab in memory");
1368                return 0;
1369            }
1370        }
1371        else if (strtab_data.GetByteSize() == 0)
1372        {
1373            if (log)
1374                module_sp->LogMessage(log.get(), "failed to read strtab data");
1375            return 0;
1376        }
1377
1378        const ConstString &g_segment_name_TEXT = GetSegmentNameTEXT();
1379        const ConstString &g_segment_name_DATA = GetSegmentNameDATA();
1380        const ConstString &g_segment_name_OBJC = GetSegmentNameOBJC();
1381        const ConstString &g_section_name_eh_frame = GetSectionNameEHFrame();
1382        SectionSP text_section_sp(section_list->FindSectionByName(g_segment_name_TEXT));
1383        SectionSP data_section_sp(section_list->FindSectionByName(g_segment_name_DATA));
1384        SectionSP objc_section_sp(section_list->FindSectionByName(g_segment_name_OBJC));
1385        SectionSP eh_frame_section_sp;
1386        if (text_section_sp.get())
1387            eh_frame_section_sp = text_section_sp->GetChildren().FindSectionByName (g_section_name_eh_frame);
1388        else
1389            eh_frame_section_sp = section_list->FindSectionByName (g_section_name_eh_frame);
1390
1391        const bool is_arm = (m_header.cputype == llvm::MachO::CPUTypeARM);
1392        if (text_section_sp && function_starts_data.GetByteSize())
1393        {
1394            FunctionStarts::Entry function_start_entry;
1395            function_start_entry.data = false;
1396            uint32_t function_start_offset = 0;
1397            function_start_entry.addr = text_section_sp->GetFileAddress();
1398            uint64_t delta;
1399            while ((delta = function_starts_data.GetULEB128(&function_start_offset)) > 0)
1400            {
1401                // Now append the current entry
1402                function_start_entry.addr += delta;
1403                function_starts.Append(function_start_entry);
1404            }
1405        }
1406
1407        const uint32_t function_starts_count = function_starts.GetSize();
1408
1409        uint8_t TEXT_eh_frame_sectID = eh_frame_section_sp.get() ? eh_frame_section_sp->GetID() : NListSectionNoSection;
1410
1411        uint32_t nlist_data_offset = 0;
1412
1413        uint32_t N_SO_index = UINT32_MAX;
1414
1415        MachSymtabSectionInfo section_info (section_list);
1416        std::vector<uint32_t> N_FUN_indexes;
1417        std::vector<uint32_t> N_NSYM_indexes;
1418        std::vector<uint32_t> N_INCL_indexes;
1419        std::vector<uint32_t> N_BRAC_indexes;
1420        std::vector<uint32_t> N_COMM_indexes;
1421        typedef std::map <uint64_t, uint32_t> ValueToSymbolIndexMap;
1422        typedef std::map <uint32_t, uint32_t> NListIndexToSymbolIndexMap;
1423        ValueToSymbolIndexMap N_FUN_addr_to_sym_idx;
1424        ValueToSymbolIndexMap N_STSYM_addr_to_sym_idx;
1425        // Any symbols that get merged into another will get an entry
1426        // in this map so we know
1427        NListIndexToSymbolIndexMap m_nlist_idx_to_sym_idx;
1428        uint32_t nlist_idx = 0;
1429        Symbol *symbol_ptr = NULL;
1430
1431        uint32_t sym_idx = 0;
1432        Symbol *sym = symtab->Resize (symtab_load_command.nsyms + m_dysymtab.nindirectsyms);
1433        uint32_t num_syms = symtab->GetNumSymbols();
1434        std::string memory_symbol_name;
1435
1436        //symtab->Reserve (symtab_load_command.nsyms + m_dysymtab.nindirectsyms);
1437        for (nlist_idx = 0; nlist_idx < symtab_load_command.nsyms; ++nlist_idx)
1438        {
1439            struct nlist_64 nlist;
1440            if (!nlist_data.ValidOffsetForDataOfSize(nlist_data_offset, nlist_byte_size))
1441                break;
1442
1443            nlist.n_strx  = nlist_data.GetU32_unchecked(&nlist_data_offset);
1444            nlist.n_type  = nlist_data.GetU8_unchecked (&nlist_data_offset);
1445            nlist.n_sect  = nlist_data.GetU8_unchecked (&nlist_data_offset);
1446            nlist.n_desc  = nlist_data.GetU16_unchecked (&nlist_data_offset);
1447            nlist.n_value = nlist_data.GetAddress_unchecked (&nlist_data_offset);
1448
1449            SymbolType type = eSymbolTypeInvalid;
1450            const char *symbol_name = NULL;
1451
1452            if (process)
1453            {
1454                const addr_t str_addr = strtab_addr + nlist.n_strx;
1455                Error str_error;
1456                if (process->ReadCStringFromMemory(str_addr, memory_symbol_name, str_error))
1457                    symbol_name = memory_symbol_name.c_str();
1458            }
1459            else
1460            {
1461                symbol_name = strtab_data.PeekCStr(nlist.n_strx);
1462
1463                if (symbol_name == NULL)
1464                {
1465                    // No symbol should be NULL, even the symbols with no
1466                    // string values should have an offset zero which points
1467                    // to an empty C-string
1468                    Host::SystemLog (Host::eSystemLogError,
1469                                     "error: symbol[%u] has invalid string table offset 0x%x in %s/%s, ignoring symbol\n",
1470                                     nlist_idx,
1471                                     nlist.n_strx,
1472                                     module_sp->GetFileSpec().GetDirectory().GetCString(),
1473                                     module_sp->GetFileSpec().GetFilename().GetCString());
1474                    continue;
1475                }
1476                if (symbol_name[0] == '\0')
1477                    symbol_name = NULL;
1478            }
1479            const char *symbol_name_non_abi_mangled = NULL;
1480
1481            SectionSP symbol_section;
1482            uint32_t symbol_byte_size = 0;
1483            bool add_nlist = true;
1484            bool is_debug = ((nlist.n_type & NlistMaskStab) != 0);
1485
1486            assert (sym_idx < num_syms);
1487
1488            sym[sym_idx].SetDebug (is_debug);
1489
1490            if (is_debug)
1491            {
1492                switch (nlist.n_type)
1493                {
1494                case StabGlobalSymbol:
1495                    // N_GSYM -- global symbol: name,,NO_SECT,type,0
1496                    // Sometimes the N_GSYM value contains the address.
1497
1498                    // FIXME: In the .o files, we have a GSYM and a debug symbol for all the ObjC data.  They
1499                    // have the same address, but we want to ensure that we always find only the real symbol,
1500                    // 'cause we don't currently correctly attribute the GSYM one to the ObjCClass/Ivar/MetaClass
1501                    // symbol type.  This is a temporary hack to make sure the ObjectiveC symbols get treated
1502                    // correctly.  To do this right, we should coalesce all the GSYM & global symbols that have the
1503                    // same address.
1504
1505                    if (symbol_name && symbol_name[0] == '_' && symbol_name[1] ==  'O'
1506                        && (strncmp (symbol_name, "_OBJC_IVAR_$_", strlen ("_OBJC_IVAR_$_")) == 0
1507                            || strncmp (symbol_name, "_OBJC_CLASS_$_", strlen ("_OBJC_CLASS_$_")) == 0
1508                            || strncmp (symbol_name, "_OBJC_METACLASS_$_", strlen ("_OBJC_METACLASS_$_")) == 0))
1509                        add_nlist = false;
1510                    else
1511                    {
1512                        sym[sym_idx].SetExternal(true);
1513                        if (nlist.n_value != 0)
1514                            symbol_section = section_info.GetSection (nlist.n_sect, nlist.n_value);
1515                        type = eSymbolTypeData;
1516                    }
1517                    break;
1518
1519                case StabFunctionName:
1520                    // N_FNAME -- procedure name (f77 kludge): name,,NO_SECT,0,0
1521                    type = eSymbolTypeCompiler;
1522                    break;
1523
1524                case StabFunction:
1525                    // N_FUN -- procedure: name,,n_sect,linenumber,address
1526                    if (symbol_name)
1527                    {
1528                        type = eSymbolTypeCode;
1529                        symbol_section = section_info.GetSection (nlist.n_sect, nlist.n_value);
1530
1531                        N_FUN_addr_to_sym_idx[nlist.n_value] = sym_idx;
1532                        // We use the current number of symbols in the symbol table in lieu of
1533                        // using nlist_idx in case we ever start trimming entries out
1534                        N_FUN_indexes.push_back(sym_idx);
1535                    }
1536                    else
1537                    {
1538                        type = eSymbolTypeCompiler;
1539
1540                        if ( !N_FUN_indexes.empty() )
1541                        {
1542                            // Copy the size of the function into the original STAB entry so we don't have
1543                            // to hunt for it later
1544                            symtab->SymbolAtIndex(N_FUN_indexes.back())->SetByteSize(nlist.n_value);
1545                            N_FUN_indexes.pop_back();
1546                            // We don't really need the end function STAB as it contains the size which
1547                            // we already placed with the original symbol, so don't add it if we want a
1548                            // minimal symbol table
1549                            if (minimize)
1550                                add_nlist = false;
1551                        }
1552                    }
1553                    break;
1554
1555                case StabStaticSymbol:
1556                    // N_STSYM -- static symbol: name,,n_sect,type,address
1557                    N_STSYM_addr_to_sym_idx[nlist.n_value] = sym_idx;
1558                    symbol_section = section_info.GetSection (nlist.n_sect, nlist.n_value);
1559                    type = eSymbolTypeData;
1560                    break;
1561
1562                case StabLocalCommon:
1563                    // N_LCSYM -- .lcomm symbol: name,,n_sect,type,address
1564                    symbol_section = section_info.GetSection (nlist.n_sect, nlist.n_value);
1565                    type = eSymbolTypeCommonBlock;
1566                    break;
1567
1568                case StabBeginSymbol:
1569                    // N_BNSYM
1570                    // We use the current number of symbols in the symbol table in lieu of
1571                    // using nlist_idx in case we ever start trimming entries out
1572                    if (minimize)
1573                    {
1574                        // Skip these if we want minimal symbol tables
1575                        add_nlist = false;
1576                    }
1577                    else
1578                    {
1579                        symbol_section = section_info.GetSection (nlist.n_sect, nlist.n_value);
1580                        N_NSYM_indexes.push_back(sym_idx);
1581                        type = eSymbolTypeScopeBegin;
1582                    }
1583                    break;
1584
1585                case StabEndSymbol:
1586                    // N_ENSYM
1587                    // Set the size of the N_BNSYM to the terminating index of this N_ENSYM
1588                    // so that we can always skip the entire symbol if we need to navigate
1589                    // more quickly at the source level when parsing STABS
1590                    if (minimize)
1591                    {
1592                        // Skip these if we want minimal symbol tables
1593                        add_nlist = false;
1594                    }
1595                    else
1596                    {
1597                        if ( !N_NSYM_indexes.empty() )
1598                        {
1599                            symbol_ptr = symtab->SymbolAtIndex(N_NSYM_indexes.back());
1600                            symbol_ptr->SetByteSize(sym_idx + 1);
1601                            symbol_ptr->SetSizeIsSibling(true);
1602                            N_NSYM_indexes.pop_back();
1603                        }
1604                        type = eSymbolTypeScopeEnd;
1605                    }
1606                    break;
1607
1608
1609                case StabSourceFileOptions:
1610                    // N_OPT - emitted with gcc2_compiled and in gcc source
1611                    type = eSymbolTypeCompiler;
1612                    break;
1613
1614                case StabRegisterSymbol:
1615                    // N_RSYM - register sym: name,,NO_SECT,type,register
1616                    type = eSymbolTypeVariable;
1617                    break;
1618
1619                case StabSourceLine:
1620                    // N_SLINE - src line: 0,,n_sect,linenumber,address
1621                    symbol_section = section_info.GetSection (nlist.n_sect, nlist.n_value);
1622                    type = eSymbolTypeLineEntry;
1623                    break;
1624
1625                case StabStructureType:
1626                    // N_SSYM - structure elt: name,,NO_SECT,type,struct_offset
1627                    type = eSymbolTypeVariableType;
1628                    break;
1629
1630                case StabSourceFileName:
1631                    // N_SO - source file name
1632                    type = eSymbolTypeSourceFile;
1633                    if (symbol_name == NULL)
1634                    {
1635                        if (minimize)
1636                            add_nlist = false;
1637                        if (N_SO_index != UINT32_MAX)
1638                        {
1639                            // Set the size of the N_SO to the terminating index of this N_SO
1640                            // so that we can always skip the entire N_SO if we need to navigate
1641                            // more quickly at the source level when parsing STABS
1642                            symbol_ptr = symtab->SymbolAtIndex(N_SO_index);
1643                            symbol_ptr->SetByteSize(sym_idx + (minimize ? 0 : 1));
1644                            symbol_ptr->SetSizeIsSibling(true);
1645                        }
1646                        N_NSYM_indexes.clear();
1647                        N_INCL_indexes.clear();
1648                        N_BRAC_indexes.clear();
1649                        N_COMM_indexes.clear();
1650                        N_FUN_indexes.clear();
1651                        N_SO_index = UINT32_MAX;
1652                    }
1653                    else
1654                    {
1655                        // We use the current number of symbols in the symbol table in lieu of
1656                        // using nlist_idx in case we ever start trimming entries out
1657                        if (symbol_name[0] == '/')
1658                            N_SO_index = sym_idx;
1659                        else if (minimize && (N_SO_index == sym_idx - 1) && ((sym_idx - 1) < num_syms))
1660                        {
1661                            const char *so_path = sym[sym_idx - 1].GetMangled().GetDemangledName().AsCString();
1662                            if (so_path && so_path[0])
1663                            {
1664                                std::string full_so_path (so_path);
1665                                if (*full_so_path.rbegin() != '/')
1666                                    full_so_path += '/';
1667                                full_so_path += symbol_name;
1668                                sym[sym_idx - 1].GetMangled().SetValue(full_so_path.c_str(), false);
1669                                add_nlist = false;
1670                                m_nlist_idx_to_sym_idx[nlist_idx] = sym_idx - 1;
1671                            }
1672                        }
1673                    }
1674
1675                    break;
1676
1677                case StabObjectFileName:
1678                    // N_OSO - object file name: name,,0,0,st_mtime
1679                    type = eSymbolTypeObjectFile;
1680                    break;
1681
1682                case StabLocalSymbol:
1683                    // N_LSYM - local sym: name,,NO_SECT,type,offset
1684                    type = eSymbolTypeLocal;
1685                    break;
1686
1687                //----------------------------------------------------------------------
1688                // INCL scopes
1689                //----------------------------------------------------------------------
1690                case StabBeginIncludeFileName:
1691                    // N_BINCL - include file beginning: name,,NO_SECT,0,sum
1692                    // We use the current number of symbols in the symbol table in lieu of
1693                    // using nlist_idx in case we ever start trimming entries out
1694                    N_INCL_indexes.push_back(sym_idx);
1695                    type = eSymbolTypeScopeBegin;
1696                    break;
1697
1698                case StabEndIncludeFile:
1699                    // N_EINCL - include file end: name,,NO_SECT,0,0
1700                    // Set the size of the N_BINCL to the terminating index of this N_EINCL
1701                    // so that we can always skip the entire symbol if we need to navigate
1702                    // more quickly at the source level when parsing STABS
1703                    if ( !N_INCL_indexes.empty() )
1704                    {
1705                        symbol_ptr = symtab->SymbolAtIndex(N_INCL_indexes.back());
1706                        symbol_ptr->SetByteSize(sym_idx + 1);
1707                        symbol_ptr->SetSizeIsSibling(true);
1708                        N_INCL_indexes.pop_back();
1709                    }
1710                    type = eSymbolTypeScopeEnd;
1711                    break;
1712
1713                case StabIncludeFileName:
1714                    // N_SOL - #included file name: name,,n_sect,0,address
1715                    type = eSymbolTypeHeaderFile;
1716
1717                    // We currently don't use the header files on darwin
1718                    if (minimize)
1719                        add_nlist = false;
1720                    break;
1721
1722                case StabCompilerParameters:
1723                    // N_PARAMS - compiler parameters: name,,NO_SECT,0,0
1724                    type = eSymbolTypeCompiler;
1725                    break;
1726
1727                case StabCompilerVersion:
1728                    // N_VERSION - compiler version: name,,NO_SECT,0,0
1729                    type = eSymbolTypeCompiler;
1730                    break;
1731
1732                case StabCompilerOptLevel:
1733                    // N_OLEVEL - compiler -O level: name,,NO_SECT,0,0
1734                    type = eSymbolTypeCompiler;
1735                    break;
1736
1737                case StabParameter:
1738                    // N_PSYM - parameter: name,,NO_SECT,type,offset
1739                    type = eSymbolTypeVariable;
1740                    break;
1741
1742                case StabAlternateEntry:
1743                    // N_ENTRY - alternate entry: name,,n_sect,linenumber,address
1744                    symbol_section = section_info.GetSection (nlist.n_sect, nlist.n_value);
1745                    type = eSymbolTypeLineEntry;
1746                    break;
1747
1748                //----------------------------------------------------------------------
1749                // Left and Right Braces
1750                //----------------------------------------------------------------------
1751                case StabLeftBracket:
1752                    // N_LBRAC - left bracket: 0,,NO_SECT,nesting level,address
1753                    // We use the current number of symbols in the symbol table in lieu of
1754                    // using nlist_idx in case we ever start trimming entries out
1755                    symbol_section = section_info.GetSection (nlist.n_sect, nlist.n_value);
1756                    N_BRAC_indexes.push_back(sym_idx);
1757                    type = eSymbolTypeScopeBegin;
1758                    break;
1759
1760                case StabRightBracket:
1761                    // N_RBRAC - right bracket: 0,,NO_SECT,nesting level,address
1762                    // Set the size of the N_LBRAC to the terminating index of this N_RBRAC
1763                    // so that we can always skip the entire symbol if we need to navigate
1764                    // more quickly at the source level when parsing STABS
1765                    symbol_section = section_info.GetSection (nlist.n_sect, nlist.n_value);
1766                    if ( !N_BRAC_indexes.empty() )
1767                    {
1768                        symbol_ptr = symtab->SymbolAtIndex(N_BRAC_indexes.back());
1769                        symbol_ptr->SetByteSize(sym_idx + 1);
1770                        symbol_ptr->SetSizeIsSibling(true);
1771                        N_BRAC_indexes.pop_back();
1772                    }
1773                    type = eSymbolTypeScopeEnd;
1774                    break;
1775
1776                case StabDeletedIncludeFile:
1777                    // N_EXCL - deleted include file: name,,NO_SECT,0,sum
1778                    type = eSymbolTypeHeaderFile;
1779                    break;
1780
1781                //----------------------------------------------------------------------
1782                // COMM scopes
1783                //----------------------------------------------------------------------
1784                case StabBeginCommon:
1785                    // N_BCOMM - begin common: name,,NO_SECT,0,0
1786                    // We use the current number of symbols in the symbol table in lieu of
1787                    // using nlist_idx in case we ever start trimming entries out
1788                    type = eSymbolTypeScopeBegin;
1789                    N_COMM_indexes.push_back(sym_idx);
1790                    break;
1791
1792                case StabEndCommonLocal:
1793                    // N_ECOML - end common (local name): 0,,n_sect,0,address
1794                    symbol_section = section_info.GetSection (nlist.n_sect, nlist.n_value);
1795                    // Fall through
1796
1797                case StabEndCommon:
1798                    // N_ECOMM - end common: name,,n_sect,0,0
1799                    // Set the size of the N_BCOMM to the terminating index of this N_ECOMM/N_ECOML
1800                    // so that we can always skip the entire symbol if we need to navigate
1801                    // more quickly at the source level when parsing STABS
1802                    if ( !N_COMM_indexes.empty() )
1803                    {
1804                        symbol_ptr = symtab->SymbolAtIndex(N_COMM_indexes.back());
1805                        symbol_ptr->SetByteSize(sym_idx + 1);
1806                        symbol_ptr->SetSizeIsSibling(true);
1807                        N_COMM_indexes.pop_back();
1808                    }
1809                    type = eSymbolTypeScopeEnd;
1810                    break;
1811
1812                case StabLength:
1813                    // N_LENG - second stab entry with length information
1814                    type = eSymbolTypeAdditional;
1815                    break;
1816
1817                default: break;
1818                }
1819            }
1820            else
1821            {
1822                //uint8_t n_pext    = NlistMaskPrivateExternal & nlist.n_type;
1823                uint8_t n_type  = NlistMaskType & nlist.n_type;
1824                sym[sym_idx].SetExternal((NlistMaskExternal & nlist.n_type) != 0);
1825
1826                switch (n_type)
1827                {
1828                case NListTypeIndirect:         // N_INDR - Fall through
1829                case NListTypePreboundUndefined:// N_PBUD - Fall through
1830                case NListTypeUndefined:        // N_UNDF
1831                    type = eSymbolTypeUndefined;
1832                    break;
1833
1834                case NListTypeAbsolute:         // N_ABS
1835                    type = eSymbolTypeAbsolute;
1836                    break;
1837
1838                case NListTypeSection:          // N_SECT
1839                    {
1840                        symbol_section = section_info.GetSection (nlist.n_sect, nlist.n_value);
1841
1842                        if (symbol_section == NULL)
1843                        {
1844                            // TODO: warn about this?
1845                            add_nlist = false;
1846                            break;
1847                        }
1848
1849                        if (TEXT_eh_frame_sectID == nlist.n_sect)
1850                        {
1851                            type = eSymbolTypeException;
1852                        }
1853                        else
1854                        {
1855                            uint32_t section_type = symbol_section->Get() & SectionFlagMaskSectionType;
1856
1857                            switch (section_type)
1858                            {
1859                            case SectionTypeRegular:                     break; // regular section
1860                            //case SectionTypeZeroFill:                 type = eSymbolTypeData;    break; // zero fill on demand section
1861                            case SectionTypeCStringLiterals:            type = eSymbolTypeData;    break; // section with only literal C strings
1862                            case SectionType4ByteLiterals:              type = eSymbolTypeData;    break; // section with only 4 byte literals
1863                            case SectionType8ByteLiterals:              type = eSymbolTypeData;    break; // section with only 8 byte literals
1864                            case SectionTypeLiteralPointers:            type = eSymbolTypeTrampoline; break; // section with only pointers to literals
1865                            case SectionTypeNonLazySymbolPointers:      type = eSymbolTypeTrampoline; break; // section with only non-lazy symbol pointers
1866                            case SectionTypeLazySymbolPointers:         type = eSymbolTypeTrampoline; break; // section with only lazy symbol pointers
1867                            case SectionTypeSymbolStubs:                type = eSymbolTypeTrampoline; break; // section with only symbol stubs, byte size of stub in the reserved2 field
1868                            case SectionTypeModuleInitFunctionPointers: type = eSymbolTypeCode;    break; // section with only function pointers for initialization
1869                            case SectionTypeModuleTermFunctionPointers: type = eSymbolTypeCode;    break; // section with only function pointers for termination
1870                            //case SectionTypeCoalesced:                type = eSymbolType;    break; // section contains symbols that are to be coalesced
1871                            //case SectionTypeZeroFillLarge:            type = eSymbolTypeData;    break; // zero fill on demand section (that can be larger than 4 gigabytes)
1872                            case SectionTypeInterposing:                type = eSymbolTypeTrampoline;  break; // section with only pairs of function pointers for interposing
1873                            case SectionType16ByteLiterals:             type = eSymbolTypeData;    break; // section with only 16 byte literals
1874                            case SectionTypeDTraceObjectFormat:         type = eSymbolTypeInstrumentation; break;
1875                            case SectionTypeLazyDylibSymbolPointers:    type = eSymbolTypeTrampoline; break;
1876                            default: break;
1877                            }
1878
1879                            if (type == eSymbolTypeInvalid)
1880                            {
1881                                const char *symbol_sect_name = symbol_section->GetName().AsCString();
1882                                if (symbol_section->IsDescendant (text_section_sp.get()))
1883                                {
1884                                    if (symbol_section->IsClear(SectionAttrUserPureInstructions |
1885                                                                SectionAttrUserSelfModifyingCode |
1886                                                                SectionAttrSytemSomeInstructions))
1887                                        type = eSymbolTypeData;
1888                                    else
1889                                        type = eSymbolTypeCode;
1890                                }
1891                                else
1892                                if (symbol_section->IsDescendant(data_section_sp.get()))
1893                                {
1894                                    if (symbol_sect_name && ::strstr (symbol_sect_name, "__objc") == symbol_sect_name)
1895                                    {
1896                                        type = eSymbolTypeRuntime;
1897
1898                                        if (symbol_name &&
1899                                            symbol_name[0] == '_' &&
1900                                            symbol_name[1] == 'O' &&
1901                                            symbol_name[2] == 'B')
1902                                        {
1903                                            llvm::StringRef symbol_name_ref(symbol_name);
1904                                            static const llvm::StringRef g_objc_v2_prefix_class ("_OBJC_CLASS_$_");
1905                                            static const llvm::StringRef g_objc_v2_prefix_metaclass ("_OBJC_METACLASS_$_");
1906                                            static const llvm::StringRef g_objc_v2_prefix_ivar ("_OBJC_IVAR_$_");
1907                                            if (symbol_name_ref.startswith(g_objc_v2_prefix_class))
1908                                            {
1909                                                symbol_name_non_abi_mangled = symbol_name + 1;
1910                                                symbol_name = symbol_name + g_objc_v2_prefix_class.size();
1911                                                type = eSymbolTypeObjCClass;
1912                                            }
1913                                            else if (symbol_name_ref.startswith(g_objc_v2_prefix_metaclass))
1914                                            {
1915                                                symbol_name_non_abi_mangled = symbol_name + 1;
1916                                                symbol_name = symbol_name + g_objc_v2_prefix_metaclass.size();
1917                                                type = eSymbolTypeObjCMetaClass;
1918                                            }
1919                                            else if (symbol_name_ref.startswith(g_objc_v2_prefix_ivar))
1920                                            {
1921                                                symbol_name_non_abi_mangled = symbol_name + 1;
1922                                                symbol_name = symbol_name + g_objc_v2_prefix_ivar.size();
1923                                                type = eSymbolTypeObjCIVar;
1924                                            }
1925                                        }
1926                                    }
1927                                    else
1928                                    if (symbol_sect_name && ::strstr (symbol_sect_name, "__gcc_except_tab") == symbol_sect_name)
1929                                    {
1930                                        type = eSymbolTypeException;
1931                                    }
1932                                    else
1933                                    {
1934                                        type = eSymbolTypeData;
1935                                    }
1936                                }
1937                                else
1938                                if (symbol_sect_name && ::strstr (symbol_sect_name, "__IMPORT") == symbol_sect_name)
1939                                {
1940                                    type = eSymbolTypeTrampoline;
1941                                }
1942                                else
1943                                if (symbol_section->IsDescendant(objc_section_sp.get()))
1944                                {
1945                                    type = eSymbolTypeRuntime;
1946                                    if (symbol_name && symbol_name[0] == '.')
1947                                    {
1948                                        llvm::StringRef symbol_name_ref(symbol_name);
1949                                        static const llvm::StringRef g_objc_v1_prefix_class (".objc_class_name_");
1950                                        if (symbol_name_ref.startswith(g_objc_v1_prefix_class))
1951                                        {
1952                                            symbol_name_non_abi_mangled = symbol_name;
1953                                            symbol_name = symbol_name + g_objc_v1_prefix_class.size();
1954                                            type = eSymbolTypeObjCClass;
1955                                        }
1956                                    }
1957                                }
1958                            }
1959                        }
1960                    }
1961                    break;
1962                }
1963            }
1964
1965            if (add_nlist)
1966            {
1967                uint64_t symbol_value = nlist.n_value;
1968                bool symbol_name_is_mangled = false;
1969
1970                if (symbol_name_non_abi_mangled)
1971                {
1972                    sym[sym_idx].GetMangled().SetMangledName (symbol_name_non_abi_mangled);
1973                    sym[sym_idx].GetMangled().SetDemangledName (symbol_name);
1974                }
1975                else
1976                {
1977                    if (symbol_name && symbol_name[0] == '_')
1978                    {
1979                        symbol_name_is_mangled = symbol_name[1] == '_';
1980                        symbol_name++;  // Skip the leading underscore
1981                    }
1982
1983                    if (symbol_name)
1984                    {
1985                        sym[sym_idx].GetMangled().SetValue(symbol_name, symbol_name_is_mangled);
1986                    }
1987                }
1988
1989                if (is_debug == false)
1990                {
1991                    if (type == eSymbolTypeCode)
1992                    {
1993                        // See if we can find a N_FUN entry for any code symbols.
1994                        // If we do find a match, and the name matches, then we
1995                        // can merge the two into just the function symbol to avoid
1996                        // duplicate entries in the symbol table
1997                        ValueToSymbolIndexMap::const_iterator pos = N_FUN_addr_to_sym_idx.find (nlist.n_value);
1998                        if (pos != N_FUN_addr_to_sym_idx.end())
1999                        {
2000                            if ((symbol_name_is_mangled == true && sym[sym_idx].GetMangled().GetMangledName() == sym[pos->second].GetMangled().GetMangledName()) ||
2001                                (symbol_name_is_mangled == false && sym[sym_idx].GetMangled().GetDemangledName() == sym[pos->second].GetMangled().GetDemangledName()))
2002                            {
2003                                m_nlist_idx_to_sym_idx[nlist_idx] = pos->second;
2004                                // We just need the flags from the linker symbol, so put these flags
2005                                // into the N_FUN flags to avoid duplicate symbols in the symbol table
2006                                sym[pos->second].SetFlags (nlist.n_type << 16 | nlist.n_desc);
2007                                sym[sym_idx].Clear();
2008                                continue;
2009                            }
2010                        }
2011                    }
2012                    else if (type == eSymbolTypeData)
2013                    {
2014                        // See if we can find a N_STSYM entry for any data symbols.
2015                        // If we do find a match, and the name matches, then we
2016                        // can merge the two into just the Static symbol to avoid
2017                        // duplicate entries in the symbol table
2018                        ValueToSymbolIndexMap::const_iterator pos = N_STSYM_addr_to_sym_idx.find (nlist.n_value);
2019                        if (pos != N_STSYM_addr_to_sym_idx.end())
2020                        {
2021                            if ((symbol_name_is_mangled == true && sym[sym_idx].GetMangled().GetMangledName() == sym[pos->second].GetMangled().GetMangledName()) ||
2022                                (symbol_name_is_mangled == false && sym[sym_idx].GetMangled().GetDemangledName() == sym[pos->second].GetMangled().GetDemangledName()))
2023                            {
2024                                m_nlist_idx_to_sym_idx[nlist_idx] = pos->second;
2025                                // We just need the flags from the linker symbol, so put these flags
2026                                // into the N_STSYM flags to avoid duplicate symbols in the symbol table
2027                                sym[pos->second].SetFlags (nlist.n_type << 16 | nlist.n_desc);
2028                                sym[sym_idx].Clear();
2029                                continue;
2030                            }
2031                        }
2032                    }
2033                }
2034                if (symbol_section)
2035                {
2036                    const addr_t section_file_addr = symbol_section->GetFileAddress();
2037                    if (symbol_byte_size == 0 && function_starts_count > 0)
2038                    {
2039                        addr_t symbol_lookup_file_addr = nlist.n_value;
2040                        // Do an exact address match for non-ARM addresses, else get the closest since
2041                        // the symbol might be a thumb symbol which has an address with bit zero set
2042                        FunctionStarts::Entry *func_start_entry = function_starts.FindEntry (symbol_lookup_file_addr, !is_arm);
2043                        if (is_arm && func_start_entry)
2044                        {
2045                            // Verify that the function start address is the symbol address (ARM)
2046                            // or the symbol address + 1 (thumb)
2047                            if (func_start_entry->addr != symbol_lookup_file_addr &&
2048                                func_start_entry->addr != (symbol_lookup_file_addr + 1))
2049                            {
2050                                // Not the right entry, NULL it out...
2051                                func_start_entry = NULL;
2052                            }
2053                        }
2054                        if (func_start_entry)
2055                        {
2056                            func_start_entry->data = true;
2057
2058                            addr_t symbol_file_addr = func_start_entry->addr;
2059                            uint32_t symbol_flags = 0;
2060                            if (is_arm)
2061                            {
2062                                if (symbol_file_addr & 1)
2063                                    symbol_flags = MACHO_NLIST_ARM_SYMBOL_IS_THUMB;
2064                                symbol_file_addr &= 0xfffffffffffffffeull;
2065                            }
2066
2067                            const FunctionStarts::Entry *next_func_start_entry = function_starts.FindNextEntry (func_start_entry);
2068                            const addr_t section_end_file_addr = section_file_addr + symbol_section->GetByteSize();
2069                            if (next_func_start_entry)
2070                            {
2071                                addr_t next_symbol_file_addr = next_func_start_entry->addr;
2072                                // Be sure the clear the Thumb address bit when we calculate the size
2073                                // from the current and next address
2074                                if (is_arm)
2075                                    next_symbol_file_addr &= 0xfffffffffffffffeull;
2076                                symbol_byte_size = std::min<lldb::addr_t>(next_symbol_file_addr - symbol_file_addr, section_end_file_addr - symbol_file_addr);
2077                            }
2078                            else
2079                            {
2080                                symbol_byte_size = section_end_file_addr - symbol_file_addr;
2081                            }
2082                        }
2083                    }
2084                    symbol_value -= section_file_addr;
2085                }
2086
2087                sym[sym_idx].SetID (nlist_idx);
2088                sym[sym_idx].SetType (type);
2089                sym[sym_idx].GetAddress().SetSection (symbol_section);
2090                sym[sym_idx].GetAddress().SetOffset (symbol_value);
2091                sym[sym_idx].SetFlags (nlist.n_type << 16 | nlist.n_desc);
2092
2093                if (symbol_byte_size > 0)
2094                    sym[sym_idx].SetByteSize(symbol_byte_size);
2095
2096                ++sym_idx;
2097            }
2098            else
2099            {
2100                sym[sym_idx].Clear();
2101            }
2102
2103        }
2104
2105        // STAB N_GSYM entries end up having a symbol type eSymbolTypeGlobal and when the symbol value
2106        // is zero, the address of the global ends up being in a non-STAB entry. Try and fix up all
2107        // such entries by figuring out what the address for the global is by looking up this non-STAB
2108        // entry and copying the value into the debug symbol's value to save us the hassle in the
2109        // debug symbol parser.
2110
2111        Symbol *global_symbol = NULL;
2112        for (nlist_idx = 0;
2113             nlist_idx < symtab_load_command.nsyms && (global_symbol = symtab->FindSymbolWithType (eSymbolTypeData, Symtab::eDebugYes, Symtab::eVisibilityAny, nlist_idx)) != NULL;
2114             nlist_idx++)
2115        {
2116            if (global_symbol->GetAddress().GetFileAddress() == 0)
2117            {
2118                std::vector<uint32_t> indexes;
2119                if (symtab->AppendSymbolIndexesWithName (global_symbol->GetMangled().GetName(), indexes) > 0)
2120                {
2121                    std::vector<uint32_t>::const_iterator pos;
2122                    std::vector<uint32_t>::const_iterator end = indexes.end();
2123                    for (pos = indexes.begin(); pos != end; ++pos)
2124                    {
2125                        symbol_ptr = symtab->SymbolAtIndex(*pos);
2126                        if (symbol_ptr != global_symbol && symbol_ptr->IsDebug() == false)
2127                        {
2128                            global_symbol->GetAddress() = symbol_ptr->GetAddress();
2129                            break;
2130                        }
2131                    }
2132                }
2133            }
2134        }
2135
2136        uint32_t synthetic_sym_id = symtab_load_command.nsyms;
2137
2138
2139        if (function_starts_count > 0)
2140        {
2141            char synthetic_function_symbol[PATH_MAX];
2142            uint32_t num_synthetic_function_symbols = 0;
2143            for (i=0; i<function_starts_count; ++i)
2144            {
2145                if (function_starts.GetEntryRef (i).data == false)
2146                    ++num_synthetic_function_symbols;
2147            }
2148
2149            if (num_synthetic_function_symbols > 0)
2150            {
2151                if (num_syms < sym_idx + num_synthetic_function_symbols)
2152                {
2153                    num_syms = sym_idx + num_synthetic_function_symbols;
2154                    sym = symtab->Resize (num_syms);
2155                }
2156                uint32_t synthetic_function_symbol_idx = 0;
2157                for (i=0; i<function_starts_count; ++i)
2158                {
2159                    const FunctionStarts::Entry *func_start_entry = function_starts.GetEntryAtIndex (i);
2160                    if (func_start_entry->data == false)
2161                    {
2162                        addr_t symbol_file_addr = func_start_entry->addr;
2163                        uint32_t symbol_flags = 0;
2164                        if (is_arm)
2165                        {
2166                            if (symbol_file_addr & 1)
2167                                symbol_flags = MACHO_NLIST_ARM_SYMBOL_IS_THUMB;
2168                            symbol_file_addr &= 0xfffffffffffffffeull;
2169                        }
2170                        Address symbol_addr;
2171                        if (module_sp->ResolveFileAddress (symbol_file_addr, symbol_addr))
2172                        {
2173                            SectionSP symbol_section (symbol_addr.GetSection());
2174                            uint32_t symbol_byte_size = 0;
2175                            if (symbol_section)
2176                            {
2177                                const addr_t section_file_addr = symbol_section->GetFileAddress();
2178                                const FunctionStarts::Entry *next_func_start_entry = function_starts.FindNextEntry (func_start_entry);
2179                                const addr_t section_end_file_addr = section_file_addr + symbol_section->GetByteSize();
2180                                if (next_func_start_entry)
2181                                {
2182                                    addr_t next_symbol_file_addr = next_func_start_entry->addr;
2183                                    if (is_arm)
2184                                        next_symbol_file_addr &= 0xfffffffffffffffeull;
2185                                    symbol_byte_size = std::min<lldb::addr_t>(next_symbol_file_addr - symbol_file_addr, section_end_file_addr - symbol_file_addr);
2186                                }
2187                                else
2188                                {
2189                                    symbol_byte_size = section_end_file_addr - symbol_file_addr;
2190                                }
2191                                snprintf (synthetic_function_symbol,
2192                                          sizeof(synthetic_function_symbol),
2193                                          "___lldb_unnamed_function%u$$%s",
2194                                          ++synthetic_function_symbol_idx,
2195                                          module_sp->GetFileSpec().GetFilename().GetCString());
2196                                sym[sym_idx].SetID (synthetic_sym_id++);
2197                                sym[sym_idx].GetMangled().SetDemangledName(synthetic_function_symbol);
2198                                sym[sym_idx].SetType (eSymbolTypeCode);
2199                                sym[sym_idx].SetIsSynthetic (true);
2200                                sym[sym_idx].GetAddress() = symbol_addr;
2201                                if (symbol_flags)
2202                                    sym[sym_idx].SetFlags (symbol_flags);
2203                                if (symbol_byte_size)
2204                                    sym[sym_idx].SetByteSize (symbol_byte_size);
2205                                ++sym_idx;
2206                            }
2207                        }
2208                    }
2209                }
2210            }
2211        }
2212
2213        // Trim our symbols down to just what we ended up with after
2214        // removing any symbols.
2215        if (sym_idx < num_syms)
2216        {
2217            num_syms = sym_idx;
2218            sym = symtab->Resize (num_syms);
2219        }
2220
2221        // Now synthesize indirect symbols
2222        if (m_dysymtab.nindirectsyms != 0)
2223        {
2224            DataExtractor indirect_symbol_index_data (m_data, m_dysymtab.indirectsymoff, m_dysymtab.nindirectsyms * 4);
2225
2226            if (indirect_symbol_index_data.GetByteSize())
2227            {
2228                NListIndexToSymbolIndexMap::const_iterator end_index_pos = m_nlist_idx_to_sym_idx.end();
2229
2230                for (uint32_t sect_idx = 1; sect_idx < m_mach_sections.size(); ++sect_idx)
2231                {
2232                    if ((m_mach_sections[sect_idx].flags & SectionFlagMaskSectionType) == SectionTypeSymbolStubs)
2233                    {
2234                        uint32_t symbol_stub_byte_size = m_mach_sections[sect_idx].reserved2;
2235                        if (symbol_stub_byte_size == 0)
2236                            continue;
2237
2238                        const uint32_t num_symbol_stubs = m_mach_sections[sect_idx].size / symbol_stub_byte_size;
2239
2240                        if (num_symbol_stubs == 0)
2241                            continue;
2242
2243                        const uint32_t symbol_stub_index_offset = m_mach_sections[sect_idx].reserved1;
2244                        for (uint32_t stub_idx = 0; stub_idx < num_symbol_stubs; ++stub_idx)
2245                        {
2246                            const uint32_t symbol_stub_index = symbol_stub_index_offset + stub_idx;
2247                            const lldb::addr_t symbol_stub_addr = m_mach_sections[sect_idx].addr + (stub_idx * symbol_stub_byte_size);
2248                            uint32_t symbol_stub_offset = symbol_stub_index * 4;
2249                            if (indirect_symbol_index_data.ValidOffsetForDataOfSize(symbol_stub_offset, 4))
2250                            {
2251                                const uint32_t stub_sym_id = indirect_symbol_index_data.GetU32 (&symbol_stub_offset);
2252                                if (stub_sym_id & (IndirectSymbolAbsolute | IndirectSymbolLocal))
2253                                    continue;
2254
2255                                NListIndexToSymbolIndexMap::const_iterator index_pos = m_nlist_idx_to_sym_idx.find (stub_sym_id);
2256                                Symbol *stub_symbol = NULL;
2257                                if (index_pos != end_index_pos)
2258                                {
2259                                    // We have a remapping from the original nlist index to
2260                                    // a current symbol index, so just look this up by index
2261                                    stub_symbol = symtab->SymbolAtIndex (index_pos->second);
2262                                }
2263                                else
2264                                {
2265                                    // We need to lookup a symbol using the original nlist
2266                                    // symbol index since this index is coming from the
2267                                    // S_SYMBOL_STUBS
2268                                    stub_symbol = symtab->FindSymbolByID (stub_sym_id);
2269                                }
2270
2271                                assert (stub_symbol);
2272                                if (stub_symbol)
2273                                {
2274                                    Address so_addr(symbol_stub_addr, section_list);
2275
2276                                    if (stub_symbol->GetType() == eSymbolTypeUndefined)
2277                                    {
2278                                        // Change the external symbol into a trampoline that makes sense
2279                                        // These symbols were N_UNDF N_EXT, and are useless to us, so we
2280                                        // can re-use them so we don't have to make up a synthetic symbol
2281                                        // for no good reason.
2282                                        stub_symbol->SetType (eSymbolTypeTrampoline);
2283                                        stub_symbol->SetExternal (false);
2284                                        stub_symbol->GetAddress() = so_addr;
2285                                        stub_symbol->SetByteSize (symbol_stub_byte_size);
2286                                    }
2287                                    else
2288                                    {
2289                                        // Make a synthetic symbol to describe the trampoline stub
2290                                        Mangled stub_symbol_mangled_name(stub_symbol->GetMangled());
2291                                        if (sym_idx >= num_syms)
2292                                        {
2293                                            sym = symtab->Resize (++num_syms);
2294                                            stub_symbol = NULL;  // this pointer no longer valid
2295                                        }
2296                                        sym[sym_idx].SetID (synthetic_sym_id++);
2297                                        sym[sym_idx].GetMangled() = stub_symbol_mangled_name;
2298                                        sym[sym_idx].SetType (eSymbolTypeTrampoline);
2299                                        sym[sym_idx].SetIsSynthetic (true);
2300                                        sym[sym_idx].GetAddress() = so_addr;
2301                                        sym[sym_idx].SetByteSize (symbol_stub_byte_size);
2302                                        ++sym_idx;
2303                                    }
2304                                }
2305                            }
2306                        }
2307                    }
2308                }
2309            }
2310        }
2311        return symtab->GetNumSymbols();
2312    }
2313    return 0;
2314}
2315
2316
2317void
2318ObjectFileMachO::Dump (Stream *s)
2319{
2320    ModuleSP module_sp(GetModule());
2321    if (module_sp)
2322    {
2323        lldb_private::Mutex::Locker locker(module_sp->GetMutex());
2324        s->Printf("%p: ", this);
2325        s->Indent();
2326        if (m_header.magic == HeaderMagic64 || m_header.magic == HeaderMagic64Swapped)
2327            s->PutCString("ObjectFileMachO64");
2328        else
2329            s->PutCString("ObjectFileMachO32");
2330
2331        ArchSpec header_arch(eArchTypeMachO, m_header.cputype, m_header.cpusubtype);
2332
2333        *s << ", file = '" << m_file << "', arch = " << header_arch.GetArchitectureName() << "\n";
2334
2335        if (m_sections_ap.get())
2336            m_sections_ap->Dump(s, NULL, true, UINT32_MAX);
2337
2338        if (m_symtab_ap.get())
2339            m_symtab_ap->Dump(s, NULL, eSortOrderNone);
2340    }
2341}
2342
2343
2344bool
2345ObjectFileMachO::GetUUID (lldb_private::UUID* uuid)
2346{
2347    ModuleSP module_sp(GetModule());
2348    if (module_sp)
2349    {
2350        lldb_private::Mutex::Locker locker(module_sp->GetMutex());
2351        struct uuid_command load_cmd;
2352        uint32_t offset = MachHeaderSizeFromMagic(m_header.magic);
2353        uint32_t i;
2354        for (i=0; i<m_header.ncmds; ++i)
2355        {
2356            const uint32_t cmd_offset = offset;
2357            if (m_data.GetU32(&offset, &load_cmd, 2) == NULL)
2358                break;
2359
2360            if (load_cmd.cmd == LoadCommandUUID)
2361            {
2362                const uint8_t *uuid_bytes = m_data.PeekData(offset, 16);
2363                if (uuid_bytes)
2364                {
2365                    uuid->SetBytes (uuid_bytes);
2366                    return true;
2367                }
2368                return false;
2369            }
2370            offset = cmd_offset + load_cmd.cmdsize;
2371        }
2372    }
2373    return false;
2374}
2375
2376
2377uint32_t
2378ObjectFileMachO::GetDependentModules (FileSpecList& files)
2379{
2380    uint32_t count = 0;
2381    ModuleSP module_sp(GetModule());
2382    if (module_sp)
2383    {
2384        lldb_private::Mutex::Locker locker(module_sp->GetMutex());
2385        struct load_command load_cmd;
2386        uint32_t offset = MachHeaderSizeFromMagic(m_header.magic);
2387        const bool resolve_path = false; // Don't resolve the dependend file paths since they may not reside on this system
2388        uint32_t i;
2389        for (i=0; i<m_header.ncmds; ++i)
2390        {
2391            const uint32_t cmd_offset = offset;
2392            if (m_data.GetU32(&offset, &load_cmd, 2) == NULL)
2393                break;
2394
2395            switch (load_cmd.cmd)
2396            {
2397            case LoadCommandDylibLoad:
2398            case LoadCommandDylibLoadWeak:
2399            case LoadCommandDylibReexport:
2400            case LoadCommandDynamicLinkerLoad:
2401            case LoadCommandFixedVMShlibLoad:
2402            case LoadCommandDylibLoadUpward:
2403                {
2404                    uint32_t name_offset = cmd_offset + m_data.GetU32(&offset);
2405                    const char *path = m_data.PeekCStr(name_offset);
2406                    // Skip any path that starts with '@' since these are usually:
2407                    // @executable_path/.../file
2408                    // @rpath/.../file
2409                    if (path && path[0] != '@')
2410                    {
2411                        FileSpec file_spec(path, resolve_path);
2412                        if (files.AppendIfUnique(file_spec))
2413                            count++;
2414                    }
2415                }
2416                break;
2417
2418            default:
2419                break;
2420            }
2421            offset = cmd_offset + load_cmd.cmdsize;
2422        }
2423    }
2424    return count;
2425}
2426
2427lldb_private::Address
2428ObjectFileMachO::GetEntryPointAddress ()
2429{
2430    // If the object file is not an executable it can't hold the entry point.  m_entry_point_address
2431    // is initialized to an invalid address, so we can just return that.
2432    // If m_entry_point_address is valid it means we've found it already, so return the cached value.
2433
2434    if (!IsExecutable() || m_entry_point_address.IsValid())
2435        return m_entry_point_address;
2436
2437    // Otherwise, look for the UnixThread or Thread command.  The data for the Thread command is given in
2438    // /usr/include/mach-o.h, but it is basically:
2439    //
2440    //  uint32_t flavor  - this is the flavor argument you would pass to thread_get_state
2441    //  uint32_t count   - this is the count of longs in the thread state data
2442    //  struct XXX_thread_state state - this is the structure from <machine/thread_status.h> corresponding to the flavor.
2443    //  <repeat this trio>
2444    //
2445    // So we just keep reading the various register flavors till we find the GPR one, then read the PC out of there.
2446    // FIXME: We will need to have a "RegisterContext data provider" class at some point that can get all the registers
2447    // out of data in this form & attach them to a given thread.  That should underlie the MacOS X User process plugin,
2448    // and we'll also need it for the MacOS X Core File process plugin.  When we have that we can also use it here.
2449    //
2450    // For now we hard-code the offsets and flavors we need:
2451    //
2452    //
2453
2454    ModuleSP module_sp(GetModule());
2455    if (module_sp)
2456    {
2457        lldb_private::Mutex::Locker locker(module_sp->GetMutex());
2458        struct load_command load_cmd;
2459        uint32_t offset = MachHeaderSizeFromMagic(m_header.magic);
2460        uint32_t i;
2461        lldb::addr_t start_address = LLDB_INVALID_ADDRESS;
2462        bool done = false;
2463
2464        for (i=0; i<m_header.ncmds; ++i)
2465        {
2466            const uint32_t cmd_offset = offset;
2467            if (m_data.GetU32(&offset, &load_cmd, 2) == NULL)
2468                break;
2469
2470            switch (load_cmd.cmd)
2471            {
2472            case LoadCommandUnixThread:
2473            case LoadCommandThread:
2474                {
2475                    while (offset < cmd_offset + load_cmd.cmdsize)
2476                    {
2477                        uint32_t flavor = m_data.GetU32(&offset);
2478                        uint32_t count = m_data.GetU32(&offset);
2479                        if (count == 0)
2480                        {
2481                            // We've gotten off somehow, log and exit;
2482                            return m_entry_point_address;
2483                        }
2484
2485                        switch (m_header.cputype)
2486                        {
2487                        case llvm::MachO::CPUTypeARM:
2488                           if (flavor == 1) // ARM_THREAD_STATE from mach/arm/thread_status.h
2489                           {
2490                               offset += 60;  // This is the offset of pc in the GPR thread state data structure.
2491                               start_address = m_data.GetU32(&offset);
2492                               done = true;
2493                            }
2494                        break;
2495                        case llvm::MachO::CPUTypeI386:
2496                           if (flavor == 1) // x86_THREAD_STATE32 from mach/i386/thread_status.h
2497                           {
2498                               offset += 40;  // This is the offset of eip in the GPR thread state data structure.
2499                               start_address = m_data.GetU32(&offset);
2500                               done = true;
2501                            }
2502                        break;
2503                        case llvm::MachO::CPUTypeX86_64:
2504                           if (flavor == 4) // x86_THREAD_STATE64 from mach/i386/thread_status.h
2505                           {
2506                               offset += 16 * 8;  // This is the offset of rip in the GPR thread state data structure.
2507                               start_address = m_data.GetU64(&offset);
2508                               done = true;
2509                            }
2510                        break;
2511                        default:
2512                            return m_entry_point_address;
2513                        }
2514                        // Haven't found the GPR flavor yet, skip over the data for this flavor:
2515                        if (done)
2516                            break;
2517                        offset += count * 4;
2518                    }
2519                }
2520                break;
2521            case LoadCommandMain:
2522                {
2523                    ConstString text_segment_name ("__TEXT");
2524                    uint64_t entryoffset = m_data.GetU64(&offset);
2525                    SectionSP text_segment_sp = GetSectionList()->FindSectionByName(text_segment_name);
2526                    if (text_segment_sp)
2527                    {
2528                        done = true;
2529                        start_address = text_segment_sp->GetFileAddress() + entryoffset;
2530                    }
2531                }
2532
2533            default:
2534                break;
2535            }
2536            if (done)
2537                break;
2538
2539            // Go to the next load command:
2540            offset = cmd_offset + load_cmd.cmdsize;
2541        }
2542
2543        if (start_address != LLDB_INVALID_ADDRESS)
2544        {
2545            // We got the start address from the load commands, so now resolve that address in the sections
2546            // of this ObjectFile:
2547            if (!m_entry_point_address.ResolveAddressUsingFileSections (start_address, GetSectionList()))
2548            {
2549                m_entry_point_address.Clear();
2550            }
2551        }
2552        else
2553        {
2554            // We couldn't read the UnixThread load command - maybe it wasn't there.  As a fallback look for the
2555            // "start" symbol in the main executable.
2556
2557            ModuleSP module_sp (GetModule());
2558
2559            if (module_sp)
2560            {
2561                SymbolContextList contexts;
2562                SymbolContext context;
2563                if (module_sp->FindSymbolsWithNameAndType(ConstString ("start"), eSymbolTypeCode, contexts))
2564                {
2565                    if (contexts.GetContextAtIndex(0, context))
2566                        m_entry_point_address = context.symbol->GetAddress();
2567                }
2568            }
2569        }
2570    }
2571
2572    return m_entry_point_address;
2573
2574}
2575
2576lldb_private::Address
2577ObjectFileMachO::GetHeaderAddress ()
2578{
2579    lldb_private::Address header_addr;
2580    SectionList *section_list = GetSectionList();
2581    if (section_list)
2582    {
2583        SectionSP text_segment_sp (section_list->FindSectionByName (GetSegmentNameTEXT()));
2584        if (text_segment_sp)
2585        {
2586            header_addr.SetSection (text_segment_sp);
2587            header_addr.SetOffset (0);
2588        }
2589    }
2590    return header_addr;
2591}
2592
2593uint32_t
2594ObjectFileMachO::GetNumThreadContexts ()
2595{
2596    ModuleSP module_sp(GetModule());
2597    if (module_sp)
2598    {
2599        lldb_private::Mutex::Locker locker(module_sp->GetMutex());
2600        if (!m_thread_context_offsets_valid)
2601        {
2602            m_thread_context_offsets_valid = true;
2603            uint32_t offset = MachHeaderSizeFromMagic(m_header.magic);
2604            FileRangeArray::Entry file_range;
2605            thread_command thread_cmd;
2606            for (uint32_t i=0; i<m_header.ncmds; ++i)
2607            {
2608                const uint32_t cmd_offset = offset;
2609                if (m_data.GetU32(&offset, &thread_cmd, 2) == NULL)
2610                    break;
2611
2612                if (thread_cmd.cmd == LoadCommandThread)
2613                {
2614                    file_range.SetRangeBase (offset);
2615                    file_range.SetByteSize (thread_cmd.cmdsize - 8);
2616                    m_thread_context_offsets.Append (file_range);
2617                }
2618                offset = cmd_offset + thread_cmd.cmdsize;
2619            }
2620        }
2621    }
2622    return m_thread_context_offsets.GetSize();
2623}
2624
2625lldb::RegisterContextSP
2626ObjectFileMachO::GetThreadContextAtIndex (uint32_t idx, lldb_private::Thread &thread)
2627{
2628    lldb::RegisterContextSP reg_ctx_sp;
2629
2630    ModuleSP module_sp(GetModule());
2631    if (module_sp)
2632    {
2633        lldb_private::Mutex::Locker locker(module_sp->GetMutex());
2634        if (!m_thread_context_offsets_valid)
2635            GetNumThreadContexts ();
2636
2637        const FileRangeArray::Entry *thread_context_file_range = m_thread_context_offsets.GetEntryAtIndex (idx);
2638
2639        DataExtractor data (m_data,
2640                            thread_context_file_range->GetRangeBase(),
2641                            thread_context_file_range->GetByteSize());
2642
2643        switch (m_header.cputype)
2644        {
2645            case llvm::MachO::CPUTypeARM:
2646                reg_ctx_sp.reset (new RegisterContextDarwin_arm_Mach (thread, data));
2647                break;
2648
2649            case llvm::MachO::CPUTypeI386:
2650                reg_ctx_sp.reset (new RegisterContextDarwin_i386_Mach (thread, data));
2651                break;
2652
2653            case llvm::MachO::CPUTypeX86_64:
2654                reg_ctx_sp.reset (new RegisterContextDarwin_x86_64_Mach (thread, data));
2655                break;
2656        }
2657    }
2658    return reg_ctx_sp;
2659}
2660
2661
2662ObjectFile::Type
2663ObjectFileMachO::CalculateType()
2664{
2665    switch (m_header.filetype)
2666    {
2667        case HeaderFileTypeObject:                                          // 0x1u MH_OBJECT
2668            if (GetAddressByteSize () == 4)
2669            {
2670                // 32 bit kexts are just object files, but they do have a valid
2671                // UUID load command.
2672                UUID uuid;
2673                if (GetUUID(&uuid))
2674                {
2675                    // this checking for the UUID load command is not enough
2676                    // we could eventually look for the symbol named
2677                    // "OSKextGetCurrentIdentifier" as this is required of kexts
2678                    if (m_strata == eStrataInvalid)
2679                        m_strata = eStrataKernel;
2680                    return eTypeSharedLibrary;
2681                }
2682            }
2683            return eTypeObjectFile;
2684
2685        case HeaderFileTypeExecutable:          return eTypeExecutable;     // 0x2u MH_EXECUTE
2686        case HeaderFileTypeFixedVMShlib:        return eTypeSharedLibrary;  // 0x3u MH_FVMLIB
2687        case HeaderFileTypeCore:                return eTypeCoreFile;       // 0x4u MH_CORE
2688        case HeaderFileTypePreloadedExecutable: return eTypeSharedLibrary;  // 0x5u MH_PRELOAD
2689        case HeaderFileTypeDynamicShlib:        return eTypeSharedLibrary;  // 0x6u MH_DYLIB
2690        case HeaderFileTypeDynamicLinkEditor:   return eTypeDynamicLinker;  // 0x7u MH_DYLINKER
2691        case HeaderFileTypeBundle:              return eTypeSharedLibrary;  // 0x8u MH_BUNDLE
2692        case HeaderFileTypeDynamicShlibStub:    return eTypeStubLibrary;    // 0x9u MH_DYLIB_STUB
2693        case HeaderFileTypeDSYM:                return eTypeDebugInfo;      // 0xAu MH_DSYM
2694        case HeaderFileTypeKextBundle:          return eTypeSharedLibrary;  // 0xBu MH_KEXT_BUNDLE
2695        default:
2696            break;
2697    }
2698    return eTypeUnknown;
2699}
2700
2701ObjectFile::Strata
2702ObjectFileMachO::CalculateStrata()
2703{
2704    switch (m_header.filetype)
2705    {
2706        case HeaderFileTypeObject:      // 0x1u MH_OBJECT
2707            {
2708                // 32 bit kexts are just object files, but they do have a valid
2709                // UUID load command.
2710                UUID uuid;
2711                if (GetUUID(&uuid))
2712                {
2713                    // this checking for the UUID load command is not enough
2714                    // we could eventually look for the symbol named
2715                    // "OSKextGetCurrentIdentifier" as this is required of kexts
2716                    if (m_type == eTypeInvalid)
2717                        m_type = eTypeSharedLibrary;
2718
2719                    return eStrataKernel;
2720                }
2721            }
2722            return eStrataUnknown;
2723
2724        case HeaderFileTypeExecutable:                                     // 0x2u MH_EXECUTE
2725            // Check for the MH_DYLDLINK bit in the flags
2726            if (m_header.flags & HeaderFlagBitIsDynamicLinkObject)
2727            {
2728                return eStrataUser;
2729            }
2730            else
2731            {
2732                SectionList *section_list = GetSectionList();
2733                if (section_list)
2734                {
2735                    static ConstString g_kld_section_name ("__KLD");
2736                    if (section_list->FindSectionByName(g_kld_section_name))
2737                        return eStrataKernel;
2738                }
2739            }
2740            return eStrataRawImage;
2741
2742        case HeaderFileTypeFixedVMShlib:        return eStrataUser;         // 0x3u MH_FVMLIB
2743        case HeaderFileTypeCore:                return eStrataUnknown;      // 0x4u MH_CORE
2744        case HeaderFileTypePreloadedExecutable: return eStrataRawImage;     // 0x5u MH_PRELOAD
2745        case HeaderFileTypeDynamicShlib:        return eStrataUser;         // 0x6u MH_DYLIB
2746        case HeaderFileTypeDynamicLinkEditor:   return eStrataUser;         // 0x7u MH_DYLINKER
2747        case HeaderFileTypeBundle:              return eStrataUser;         // 0x8u MH_BUNDLE
2748        case HeaderFileTypeDynamicShlibStub:    return eStrataUser;         // 0x9u MH_DYLIB_STUB
2749        case HeaderFileTypeDSYM:                return eStrataUnknown;      // 0xAu MH_DSYM
2750        case HeaderFileTypeKextBundle:          return eStrataKernel;       // 0xBu MH_KEXT_BUNDLE
2751        default:
2752            break;
2753    }
2754    return eStrataUnknown;
2755}
2756
2757
2758uint32_t
2759ObjectFileMachO::GetVersion (uint32_t *versions, uint32_t num_versions)
2760{
2761    ModuleSP module_sp(GetModule());
2762    if (module_sp)
2763    {
2764        lldb_private::Mutex::Locker locker(module_sp->GetMutex());
2765        struct dylib_command load_cmd;
2766        uint32_t offset = MachHeaderSizeFromMagic(m_header.magic);
2767        uint32_t version_cmd = 0;
2768        uint64_t version = 0;
2769        uint32_t i;
2770        for (i=0; i<m_header.ncmds; ++i)
2771        {
2772            const uint32_t cmd_offset = offset;
2773            if (m_data.GetU32(&offset, &load_cmd, 2) == NULL)
2774                break;
2775
2776            if (load_cmd.cmd == LoadCommandDylibIdent)
2777            {
2778                if (version_cmd == 0)
2779                {
2780                    version_cmd = load_cmd.cmd;
2781                    if (m_data.GetU32(&offset, &load_cmd.dylib, 4) == NULL)
2782                        break;
2783                    version = load_cmd.dylib.current_version;
2784                }
2785                break; // Break for now unless there is another more complete version
2786                       // number load command in the future.
2787            }
2788            offset = cmd_offset + load_cmd.cmdsize;
2789        }
2790
2791        if (version_cmd == LoadCommandDylibIdent)
2792        {
2793            if (versions != NULL && num_versions > 0)
2794            {
2795                if (num_versions > 0)
2796                    versions[0] = (version & 0xFFFF0000ull) >> 16;
2797                if (num_versions > 1)
2798                    versions[1] = (version & 0x0000FF00ull) >> 8;
2799                if (num_versions > 2)
2800                    versions[2] = (version & 0x000000FFull);
2801                // Fill in an remaining version numbers with invalid values
2802                for (i=3; i<num_versions; ++i)
2803                    versions[i] = UINT32_MAX;
2804            }
2805            // The LC_ID_DYLIB load command has a version with 3 version numbers
2806            // in it, so always return 3
2807            return 3;
2808        }
2809    }
2810    return false;
2811}
2812
2813bool
2814ObjectFileMachO::GetArchitecture (ArchSpec &arch)
2815{
2816    ModuleSP module_sp(GetModule());
2817    if (module_sp)
2818    {
2819        lldb_private::Mutex::Locker locker(module_sp->GetMutex());
2820        arch.SetArchitecture (eArchTypeMachO, m_header.cputype, m_header.cpusubtype);
2821
2822        // Files with type MH_PRELOAD are currently used in cases where the image
2823        // debugs at the addresses in the file itself. Below we set the OS to
2824        // unknown to make sure we use the DynamicLoaderStatic()...
2825        if (m_header.filetype == HeaderFileTypePreloadedExecutable)
2826        {
2827            arch.GetTriple().setOS (llvm::Triple::UnknownOS);
2828        }
2829        return true;
2830    }
2831    return false;
2832}
2833
2834
2835//------------------------------------------------------------------
2836// PluginInterface protocol
2837//------------------------------------------------------------------
2838const char *
2839ObjectFileMachO::GetPluginName()
2840{
2841    return "ObjectFileMachO";
2842}
2843
2844const char *
2845ObjectFileMachO::GetShortPluginName()
2846{
2847    return GetPluginNameStatic();
2848}
2849
2850uint32_t
2851ObjectFileMachO::GetPluginVersion()
2852{
2853    return 1;
2854}
2855
2856