blob: c11b2c31c3fec7251a73adfe98d5945d69aeb23e [file] [log] [blame]
Jim Grosbache0934be2012-01-16 23:50:58 +00001//===-- RuntimeDyldMachO.cpp - Run-time dynamic linker for MC-JIT -*- C++ -*-=//
Danil Malyshevcf852dc2011-07-13 07:57:58 +00002//
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// Implementation of the MC-JIT runtime dynamic linker.
11//
12//===----------------------------------------------------------------------===//
13
14#define DEBUG_TYPE "dyld"
15#include "llvm/ADT/OwningPtr.h"
16#include "llvm/ADT/StringRef.h"
17#include "llvm/ADT/STLExtras.h"
Eli Bendersky76463fd2012-01-22 07:05:02 +000018#include "RuntimeDyldMachO.h"
Danil Malyshevcf852dc2011-07-13 07:57:58 +000019using namespace llvm;
20using namespace llvm::object;
21
22namespace llvm {
23
24bool RuntimeDyldMachO::
Jim Grosbach61425c02012-01-16 22:26:39 +000025resolveRelocation(uint8_t *Address, uint64_t Value, bool isPCRel,
26 unsigned Type, unsigned Size, int64_t Addend) {
Danil Malyshevcf852dc2011-07-13 07:57:58 +000027 // This just dispatches to the proper target specific routine.
28 switch (CPUType) {
Craig Topper85814382012-02-07 05:05:23 +000029 default: llvm_unreachable("Unsupported CPU type!");
Danil Malyshevcf852dc2011-07-13 07:57:58 +000030 case mach::CTM_x86_64:
31 return resolveX86_64Relocation((uintptr_t)Address, (uintptr_t)Value,
Jim Grosbach61425c02012-01-16 22:26:39 +000032 isPCRel, Type, Size, Addend);
Danil Malyshevcf852dc2011-07-13 07:57:58 +000033 case mach::CTM_ARM:
34 return resolveARMRelocation((uintptr_t)Address, (uintptr_t)Value,
Jim Grosbach61425c02012-01-16 22:26:39 +000035 isPCRel, Type, Size, Addend);
Danil Malyshevcf852dc2011-07-13 07:57:58 +000036 }
Danil Malyshevcf852dc2011-07-13 07:57:58 +000037}
38
39bool RuntimeDyldMachO::
Jim Grosbach61425c02012-01-16 22:26:39 +000040resolveX86_64Relocation(uintptr_t Address, uintptr_t Value, bool isPCRel,
41 unsigned Type, unsigned Size, int64_t Addend) {
Danil Malyshevcf852dc2011-07-13 07:57:58 +000042 // If the relocation is PC-relative, the value to be encoded is the
43 // pointer difference.
44 if (isPCRel)
45 // FIXME: It seems this value needs to be adjusted by 4 for an effective PC
46 // address. Is that expected? Only for branches, perhaps?
47 Value -= Address + 4;
48
49 switch(Type) {
50 default:
51 llvm_unreachable("Invalid relocation type!");
Jim Grosbach652ca2f2012-01-16 23:50:49 +000052 case macho::RIT_X86_64_Signed1:
53 case macho::RIT_X86_64_Signed2:
54 case macho::RIT_X86_64_Signed4:
55 case macho::RIT_X86_64_Signed:
Danil Malyshevcf852dc2011-07-13 07:57:58 +000056 case macho::RIT_X86_64_Unsigned:
57 case macho::RIT_X86_64_Branch: {
Jim Grosbach652ca2f2012-01-16 23:50:49 +000058 Value += Addend;
Danil Malyshevcf852dc2011-07-13 07:57:58 +000059 // Mask in the target value a byte at a time (we don't have an alignment
60 // guarantee for the target address, so this is safest).
61 uint8_t *p = (uint8_t*)Address;
62 for (unsigned i = 0; i < Size; ++i) {
63 *p++ = (uint8_t)Value;
64 Value >>= 8;
65 }
66 return false;
67 }
Danil Malyshevcf852dc2011-07-13 07:57:58 +000068 case macho::RIT_X86_64_GOTLoad:
69 case macho::RIT_X86_64_GOT:
70 case macho::RIT_X86_64_Subtractor:
Danil Malyshevcf852dc2011-07-13 07:57:58 +000071 case macho::RIT_X86_64_TLV:
72 return Error("Relocation type not implemented yet!");
73 }
Danil Malyshevcf852dc2011-07-13 07:57:58 +000074}
75
Jim Grosbach61425c02012-01-16 22:26:39 +000076bool RuntimeDyldMachO::
77resolveARMRelocation(uintptr_t Address, uintptr_t Value, bool isPCRel,
78 unsigned Type, unsigned Size, int64_t Addend) {
Danil Malyshevcf852dc2011-07-13 07:57:58 +000079 // If the relocation is PC-relative, the value to be encoded is the
80 // pointer difference.
81 if (isPCRel) {
82 Value -= Address;
83 // ARM PCRel relocations have an effective-PC offset of two instructions
84 // (four bytes in Thumb mode, 8 bytes in ARM mode).
85 // FIXME: For now, assume ARM mode.
86 Value -= 8;
87 }
88
89 switch(Type) {
90 default:
91 llvm_unreachable("Invalid relocation type!");
92 case macho::RIT_Vanilla: {
Danil Malyshevcf852dc2011-07-13 07:57:58 +000093 // Mask in the target value a byte at a time (we don't have an alignment
94 // guarantee for the target address, so this is safest).
95 uint8_t *p = (uint8_t*)Address;
96 for (unsigned i = 0; i < Size; ++i) {
97 *p++ = (uint8_t)Value;
98 Value >>= 8;
99 }
100 break;
101 }
102 case macho::RIT_ARM_Branch24Bit: {
103 // Mask the value into the target address. We know instructions are
104 // 32-bit aligned, so we can do it all at once.
105 uint32_t *p = (uint32_t*)Address;
106 // The low two bits of the value are not encoded.
107 Value >>= 2;
108 // Mask the value to 24 bits.
109 Value &= 0xffffff;
110 // FIXME: If the destination is a Thumb function (and the instruction
111 // is a non-predicated BL instruction), we need to change it to a BLX
112 // instruction instead.
113
114 // Insert the value into the instruction.
115 *p = (*p & ~0xffffff) | Value;
116 break;
117 }
118 case macho::RIT_ARM_ThumbBranch22Bit:
119 case macho::RIT_ARM_ThumbBranch32Bit:
120 case macho::RIT_ARM_Half:
121 case macho::RIT_ARM_HalfDifference:
122 case macho::RIT_Pair:
123 case macho::RIT_Difference:
124 case macho::RIT_ARM_LocalDifference:
125 case macho::RIT_ARM_PreboundLazyPointer:
126 return Error("Relocation type not implemented yet!");
127 }
128 return false;
129}
130
131bool RuntimeDyldMachO::
132loadSegment32(const MachOObject *Obj,
133 const MachOObject::LoadCommandInfo *SegmentLCI,
134 const InMemoryStruct<macho::SymtabLoadCommand> &SymtabLC) {
Jim Grosbach61425c02012-01-16 22:26:39 +0000135 // FIXME: This should really be combined w/ loadSegment64. Templatized
136 // function on the 32/64 datatypes maybe?
Danil Malyshevcf852dc2011-07-13 07:57:58 +0000137 InMemoryStruct<macho::SegmentLoadCommand> SegmentLC;
138 Obj->ReadSegmentLoadCommand(*SegmentLCI, SegmentLC);
139 if (!SegmentLC)
140 return Error("unable to load segment load command");
141
Jim Grosbach61425c02012-01-16 22:26:39 +0000142
143 SmallVector<unsigned, 16> SectionMap;
Danil Malyshevcf852dc2011-07-13 07:57:58 +0000144 for (unsigned SectNum = 0; SectNum != SegmentLC->NumSections; ++SectNum) {
145 InMemoryStruct<macho::Section> Sect;
146 Obj->ReadSection(*SegmentLCI, SectNum, Sect);
147 if (!Sect)
148 return Error("unable to load section: '" + Twine(SectNum) + "'");
149
Jim Grosbach61425c02012-01-16 22:26:39 +0000150 // Allocate memory via the MM for the section.
151 uint8_t *Buffer;
152 uint32_t SectionID = Sections.size();
Sean Callanan996d6fd2012-03-01 00:15:29 +0000153 if (Sect->Flags == 0x80000400)
Jim Grosbach61425c02012-01-16 22:26:39 +0000154 Buffer = MemMgr->allocateCodeSection(Sect->Size, Sect->Align, SectionID);
155 else
156 Buffer = MemMgr->allocateDataSection(Sect->Size, Sect->Align, SectionID);
Danil Malyshevcf852dc2011-07-13 07:57:58 +0000157
Jim Grosbach61425c02012-01-16 22:26:39 +0000158 DEBUG(dbgs() << "Loading "
159 << ((Sect->Flags == 0x80000400) ? "text" : "data")
160 << " (ID #" << SectionID << ")"
161 << " '" << Sect->SegmentName << ","
162 << Sect->Name << "' of size " << Sect->Size
163 << " to address " << Buffer << ".\n");
Danil Malyshevcf852dc2011-07-13 07:57:58 +0000164
Jim Grosbach61425c02012-01-16 22:26:39 +0000165 // Copy the payload from the object file into the allocated buffer.
Danil Malyshevcf852dc2011-07-13 07:57:58 +0000166 uint8_t *Base = (uint8_t*)Obj->getData(SegmentLC->FileOffset,
167 SegmentLC->FileSize).data();
Jim Grosbach61425c02012-01-16 22:26:39 +0000168 memcpy(Buffer, Base + Sect->Address, Sect->Size);
Danil Malyshevcf852dc2011-07-13 07:57:58 +0000169
Jim Grosbach61425c02012-01-16 22:26:39 +0000170 // Remember what got allocated for this SectionID.
171 Sections.push_back(sys::MemoryBlock(Buffer, Sect->Size));
Jim Grosbach020f4e82012-01-16 23:50:55 +0000172 SectionLocalMemToID[Buffer] = SectionID;
Jim Grosbach61425c02012-01-16 22:26:39 +0000173
174 // By default, the load address of a section is its memory buffer.
175 SectionLoadAddress.push_back((uint64_t)Buffer);
176
177 // Keep a map of object file section numbers to corresponding SectionIDs
178 // while processing the file.
179 SectionMap.push_back(SectionID);
180 }
181
182 // Process the symbol table.
183 SmallVector<StringRef, 64> SymbolNames;
184 processSymbols32(Obj, SectionMap, SymbolNames, SymtabLC);
185
186 // Process the relocations for each section we're loading.
187 Relocations.grow(Relocations.size() + SegmentLC->NumSections);
188 for (unsigned SectNum = 0; SectNum != SegmentLC->NumSections; ++SectNum) {
189 InMemoryStruct<macho::Section> Sect;
190 Obj->ReadSection(*SegmentLCI, SectNum, Sect);
191 if (!Sect)
192 return Error("unable to load section: '" + Twine(SectNum) + "'");
Danil Malyshevcf852dc2011-07-13 07:57:58 +0000193 for (unsigned j = 0; j != Sect->NumRelocationTableEntries; ++j) {
194 InMemoryStruct<macho::RelocationEntry> RE;
195 Obj->ReadRelocationEntry(Sect->RelocationTableOffset, j, RE);
196 if (RE->Word0 & macho::RF_Scattered)
197 return Error("NOT YET IMPLEMENTED: scattered relocations.");
198 // Word0 of the relocation is the offset into the section where the
199 // relocation should be applied. We need to translate that into an
200 // offset into a function since that's our atom.
201 uint32_t Offset = RE->Word0;
Danil Malyshevcf852dc2011-07-13 07:57:58 +0000202 bool isExtern = (RE->Word1 >> 27) & 1;
Jim Grosbach61425c02012-01-16 22:26:39 +0000203
204 // FIXME: Get the relocation addend from the target address.
205 // FIXME: VERY imporant for internal relocations.
206
Danil Malyshevcf852dc2011-07-13 07:57:58 +0000207 // Figure out the source symbol of the relocation. If isExtern is true,
208 // this relocation references the symbol table, otherwise it references
209 // a section in the same object, numbered from 1 through NumSections
210 // (SectionBases is [0, NumSections-1]).
Danil Malyshevcf852dc2011-07-13 07:57:58 +0000211 uint32_t SourceNum = RE->Word1 & 0xffffff; // 24-bit value
Jim Grosbach61425c02012-01-16 22:26:39 +0000212 if (!isExtern) {
213 assert(SourceNum > 0 && "Invalid relocation section number!");
214 unsigned SectionID = SectionMap[SourceNum - 1];
215 unsigned TargetID = SectionMap[SectNum];
216 DEBUG(dbgs() << "Internal relocation at Section #"
217 << TargetID << " + " << Offset
218 << " from Section #"
219 << SectionID << " (Word1: "
220 << format("0x%x", RE->Word1) << ")\n");
Danil Malyshevcf852dc2011-07-13 07:57:58 +0000221
Jim Grosbach61425c02012-01-16 22:26:39 +0000222 // Store the relocation information. It will get resolved when
223 // the section addresses are assigned.
224 Relocations[SectionID].push_back(RelocationEntry(TargetID,
225 Offset,
226 RE->Word1,
227 0 /*Addend*/));
228 } else {
229 StringRef SourceName = SymbolNames[SourceNum];
Danil Malyshevcf852dc2011-07-13 07:57:58 +0000230
Jim Grosbach61425c02012-01-16 22:26:39 +0000231 // Now store the relocation information. Associate it with the source
232 // symbol. Just add it to the unresolved list and let the general
233 // path post-load resolve it if we know where the symbol is.
234 UnresolvedRelocations[SourceName].push_back(RelocationEntry(SectNum,
235 Offset,
236 RE->Word1,
237 0 /*Addend*/));
238 DEBUG(dbgs() << "Relocation at Section #" << SectNum << " + " << Offset
239 << " from '" << SourceName << "(Word1: "
240 << format("0x%x", RE->Word1) << ")\n");
241 }
Danil Malyshevcf852dc2011-07-13 07:57:58 +0000242 }
243 }
Jim Grosbach61425c02012-01-16 22:26:39 +0000244
245 // Resolve the addresses of any symbols that were defined in this segment.
246 for (int i = 0, e = SymbolNames.size(); i != e; ++i)
247 resolveSymbol(SymbolNames[i]);
248
Danil Malyshevcf852dc2011-07-13 07:57:58 +0000249 return false;
250}
251
252
253bool RuntimeDyldMachO::
254loadSegment64(const MachOObject *Obj,
255 const MachOObject::LoadCommandInfo *SegmentLCI,
256 const InMemoryStruct<macho::SymtabLoadCommand> &SymtabLC) {
257 InMemoryStruct<macho::Segment64LoadCommand> Segment64LC;
258 Obj->ReadSegment64LoadCommand(*SegmentLCI, Segment64LC);
259 if (!Segment64LC)
260 return Error("unable to load segment load command");
261
Jim Grosbach61425c02012-01-16 22:26:39 +0000262
263 SmallVector<unsigned, 16> SectionMap;
Danil Malyshevcf852dc2011-07-13 07:57:58 +0000264 for (unsigned SectNum = 0; SectNum != Segment64LC->NumSections; ++SectNum) {
265 InMemoryStruct<macho::Section64> Sect;
266 Obj->ReadSection64(*SegmentLCI, SectNum, Sect);
267 if (!Sect)
268 return Error("unable to load section: '" + Twine(SectNum) + "'");
269
Jim Grosbach61425c02012-01-16 22:26:39 +0000270 // Allocate memory via the MM for the section.
271 uint8_t *Buffer;
272 uint32_t SectionID = Sections.size();
Jim Grosbach93391342012-01-21 00:21:53 +0000273 unsigned Align = 1 << Sect->Align; // .o file has log2 alignment.
Jim Grosbachb442d7c2012-01-20 22:44:03 +0000274 if (Sect->Flags == 0x80000400)
Jim Grosbach93391342012-01-21 00:21:53 +0000275 Buffer = MemMgr->allocateCodeSection(Sect->Size, Align, SectionID);
Jim Grosbach61425c02012-01-16 22:26:39 +0000276 else
Jim Grosbach93391342012-01-21 00:21:53 +0000277 Buffer = MemMgr->allocateDataSection(Sect->Size, Align, SectionID);
Danil Malyshevcf852dc2011-07-13 07:57:58 +0000278
Jim Grosbach61425c02012-01-16 22:26:39 +0000279 DEBUG(dbgs() << "Loading "
280 << ((Sect->Flags == 0x80000400) ? "text" : "data")
281 << " (ID #" << SectionID << ")"
282 << " '" << Sect->SegmentName << ","
283 << Sect->Name << "' of size " << Sect->Size
Jim Grosbach93391342012-01-21 00:21:53 +0000284 << " (align " << Align << ")"
Jim Grosbach61425c02012-01-16 22:26:39 +0000285 << " to address " << Buffer << ".\n");
Danil Malyshevcf852dc2011-07-13 07:57:58 +0000286
Jim Grosbach61425c02012-01-16 22:26:39 +0000287 // Copy the payload from the object file into the allocated buffer.
Danil Malyshevcf852dc2011-07-13 07:57:58 +0000288 uint8_t *Base = (uint8_t*)Obj->getData(Segment64LC->FileOffset,
289 Segment64LC->FileSize).data();
Jim Grosbach61425c02012-01-16 22:26:39 +0000290 memcpy(Buffer, Base + Sect->Address, Sect->Size);
Danil Malyshevcf852dc2011-07-13 07:57:58 +0000291
Jim Grosbach61425c02012-01-16 22:26:39 +0000292 // Remember what got allocated for this SectionID.
293 Sections.push_back(sys::MemoryBlock(Buffer, Sect->Size));
Jim Grosbach020f4e82012-01-16 23:50:55 +0000294 SectionLocalMemToID[Buffer] = SectionID;
Jim Grosbach61425c02012-01-16 22:26:39 +0000295
296 // By default, the load address of a section is its memory buffer.
297 SectionLoadAddress.push_back((uint64_t)Buffer);
298
299 // Keep a map of object file section numbers to corresponding SectionIDs
300 // while processing the file.
301 SectionMap.push_back(SectionID);
302 }
303
304 // Process the symbol table.
305 SmallVector<StringRef, 64> SymbolNames;
306 processSymbols64(Obj, SectionMap, SymbolNames, SymtabLC);
307
308 // Process the relocations for each section we're loading.
309 Relocations.grow(Relocations.size() + Segment64LC->NumSections);
310 for (unsigned SectNum = 0; SectNum != Segment64LC->NumSections; ++SectNum) {
311 InMemoryStruct<macho::Section64> Sect;
312 Obj->ReadSection64(*SegmentLCI, SectNum, Sect);
313 if (!Sect)
314 return Error("unable to load section: '" + Twine(SectNum) + "'");
Danil Malyshevcf852dc2011-07-13 07:57:58 +0000315 for (unsigned j = 0; j != Sect->NumRelocationTableEntries; ++j) {
316 InMemoryStruct<macho::RelocationEntry> RE;
317 Obj->ReadRelocationEntry(Sect->RelocationTableOffset, j, RE);
318 if (RE->Word0 & macho::RF_Scattered)
319 return Error("NOT YET IMPLEMENTED: scattered relocations.");
320 // Word0 of the relocation is the offset into the section where the
321 // relocation should be applied. We need to translate that into an
322 // offset into a function since that's our atom.
323 uint32_t Offset = RE->Word0;
Danil Malyshevcf852dc2011-07-13 07:57:58 +0000324 bool isExtern = (RE->Word1 >> 27) & 1;
Jim Grosbach61425c02012-01-16 22:26:39 +0000325
326 // FIXME: Get the relocation addend from the target address.
327 // FIXME: VERY imporant for internal relocations.
328
Danil Malyshevcf852dc2011-07-13 07:57:58 +0000329 // Figure out the source symbol of the relocation. If isExtern is true,
330 // this relocation references the symbol table, otherwise it references
331 // a section in the same object, numbered from 1 through NumSections
332 // (SectionBases is [0, NumSections-1]).
Danil Malyshevcf852dc2011-07-13 07:57:58 +0000333 uint32_t SourceNum = RE->Word1 & 0xffffff; // 24-bit value
Jim Grosbach61425c02012-01-16 22:26:39 +0000334 if (!isExtern) {
335 assert(SourceNum > 0 && "Invalid relocation section number!");
336 unsigned SectionID = SectionMap[SourceNum - 1];
337 unsigned TargetID = SectionMap[SectNum];
338 DEBUG(dbgs() << "Internal relocation at Section #"
339 << TargetID << " + " << Offset
340 << " from Section #"
341 << SectionID << " (Word1: "
342 << format("0x%x", RE->Word1) << ")\n");
Danil Malyshevcf852dc2011-07-13 07:57:58 +0000343
Jim Grosbach61425c02012-01-16 22:26:39 +0000344 // Store the relocation information. It will get resolved when
345 // the section addresses are assigned.
346 Relocations[SectionID].push_back(RelocationEntry(TargetID,
347 Offset,
348 RE->Word1,
349 0 /*Addend*/));
350 } else {
351 StringRef SourceName = SymbolNames[SourceNum];
Danil Malyshevcf852dc2011-07-13 07:57:58 +0000352
Jim Grosbach61425c02012-01-16 22:26:39 +0000353 // Now store the relocation information. Associate it with the source
354 // symbol. Just add it to the unresolved list and let the general
355 // path post-load resolve it if we know where the symbol is.
356 UnresolvedRelocations[SourceName].push_back(RelocationEntry(SectNum,
357 Offset,
358 RE->Word1,
359 0 /*Addend*/));
360 DEBUG(dbgs() << "Relocation at Section #" << SectNum << " + " << Offset
361 << " from '" << SourceName << "(Word1: "
362 << format("0x%x", RE->Word1) << ")\n");
363 }
Danil Malyshevcf852dc2011-07-13 07:57:58 +0000364 }
365 }
Jim Grosbach61425c02012-01-16 22:26:39 +0000366
367 // Resolve the addresses of any symbols that were defined in this segment.
368 for (int i = 0, e = SymbolNames.size(); i != e; ++i)
369 resolveSymbol(SymbolNames[i]);
370
Danil Malyshevcf852dc2011-07-13 07:57:58 +0000371 return false;
372}
373
Jim Grosbach61425c02012-01-16 22:26:39 +0000374bool RuntimeDyldMachO::
375processSymbols32(const MachOObject *Obj,
376 SmallVectorImpl<unsigned> &SectionMap,
377 SmallVectorImpl<StringRef> &SymbolNames,
378 const InMemoryStruct<macho::SymtabLoadCommand> &SymtabLC) {
379 // FIXME: Combine w/ processSymbols64. Factor 64/32 datatype and such.
380 for (unsigned i = 0; i != SymtabLC->NumSymbolTableEntries; ++i) {
381 InMemoryStruct<macho::SymbolTableEntry> STE;
382 Obj->ReadSymbolTableEntry(SymtabLC->SymbolTableOffset, i, STE);
383 if (!STE)
384 return Error("unable to read symbol: '" + Twine(i) + "'");
385 // Get the symbol name.
386 StringRef Name = Obj->getStringAtIndex(STE->StringIndex);
387 SymbolNames.push_back(Name);
388
389 // FIXME: Check the symbol type and flags.
390 if (STE->Type != 0xF) // external, defined in this segment.
391 continue;
392 // Flags in the upper nibble we don't care about.
393 if ((STE->Flags & 0xf) != 0x0)
394 continue;
395
396 // Remember the symbol.
397 uint32_t SectionID = SectionMap[STE->SectionIndex - 1];
398 SymbolTable[Name] = SymbolLoc(SectionID, STE->Value);
399
400 DEBUG(dbgs() << "Symbol: '" << Name << "' @ "
401 << (getSectionAddress(SectionID) + STE->Value)
402 << "\n");
403 }
404 return false;
405}
406
407bool RuntimeDyldMachO::
408processSymbols64(const MachOObject *Obj,
409 SmallVectorImpl<unsigned> &SectionMap,
410 SmallVectorImpl<StringRef> &SymbolNames,
411 const InMemoryStruct<macho::SymtabLoadCommand> &SymtabLC) {
412 for (unsigned i = 0; i != SymtabLC->NumSymbolTableEntries; ++i) {
413 InMemoryStruct<macho::Symbol64TableEntry> STE;
414 Obj->ReadSymbol64TableEntry(SymtabLC->SymbolTableOffset, i, STE);
415 if (!STE)
416 return Error("unable to read symbol: '" + Twine(i) + "'");
417 // Get the symbol name.
418 StringRef Name = Obj->getStringAtIndex(STE->StringIndex);
419 SymbolNames.push_back(Name);
420
421 // FIXME: Check the symbol type and flags.
422 if (STE->Type != 0xF) // external, defined in this segment.
423 continue;
424 // Flags in the upper nibble we don't care about.
425 if ((STE->Flags & 0xf) != 0x0)
426 continue;
427
428 // Remember the symbol.
429 uint32_t SectionID = SectionMap[STE->SectionIndex - 1];
430 SymbolTable[Name] = SymbolLoc(SectionID, STE->Value);
431
432 DEBUG(dbgs() << "Symbol: '" << Name << "' @ "
433 << (getSectionAddress(SectionID) + STE->Value)
434 << "\n");
435 }
436 return false;
437}
438
439// resolveSymbol - Resolve any relocations to the specified symbol if
440// we know where it lives.
441void RuntimeDyldMachO::resolveSymbol(StringRef Name) {
442 StringMap<SymbolLoc>::const_iterator Loc = SymbolTable.find(Name);
443 if (Loc == SymbolTable.end())
444 return;
445
446 RelocationList &Relocs = UnresolvedRelocations[Name];
447 DEBUG(dbgs() << "Resolving symbol '" << Name << "'\n");
448 for (int i = 0, e = Relocs.size(); i != e; ++i) {
449 // Change the relocation to be section relative rather than symbol
450 // relative and move it to the resolved relocation list.
451 RelocationEntry Entry = Relocs[i];
452 Entry.Addend += Loc->second.second;
453 Relocations[Loc->second.first].push_back(Entry);
454 }
455 // FIXME: Keep a worklist of the relocations we've added so that we can
456 // resolve more selectively later.
457 Relocs.clear();
458}
459
Danil Malyshevcf852dc2011-07-13 07:57:58 +0000460bool RuntimeDyldMachO::loadObject(MemoryBuffer *InputBuffer) {
461 // If the linker is in an error state, don't do anything.
462 if (hasError())
463 return true;
464 // Load the Mach-O wrapper object.
465 std::string ErrorStr;
466 OwningPtr<MachOObject> Obj(
467 MachOObject::LoadFromBuffer(InputBuffer, &ErrorStr));
468 if (!Obj)
469 return Error("unable to load object: '" + ErrorStr + "'");
470
471 // Get the CPU type information from the header.
472 const macho::Header &Header = Obj->getHeader();
473
474 // FIXME: Error checking that the loaded object is compatible with
475 // the system we're running on.
476 CPUType = Header.CPUType;
477 CPUSubtype = Header.CPUSubtype;
478
479 // Validate that the load commands match what we expect.
480 const MachOObject::LoadCommandInfo *SegmentLCI = 0, *SymtabLCI = 0,
481 *DysymtabLCI = 0;
482 for (unsigned i = 0; i != Header.NumLoadCommands; ++i) {
483 const MachOObject::LoadCommandInfo &LCI = Obj->getLoadCommandInfo(i);
484 switch (LCI.Command.Type) {
485 case macho::LCT_Segment:
486 case macho::LCT_Segment64:
487 if (SegmentLCI)
488 return Error("unexpected input object (multiple segments)");
489 SegmentLCI = &LCI;
490 break;
491 case macho::LCT_Symtab:
492 if (SymtabLCI)
493 return Error("unexpected input object (multiple symbol tables)");
494 SymtabLCI = &LCI;
495 break;
496 case macho::LCT_Dysymtab:
497 if (DysymtabLCI)
498 return Error("unexpected input object (multiple symbol tables)");
499 DysymtabLCI = &LCI;
500 break;
501 default:
502 return Error("unexpected input object (unexpected load command");
503 }
504 }
505
506 if (!SymtabLCI)
507 return Error("no symbol table found in object");
508 if (!SegmentLCI)
Eli Bendersky1eb189b2012-01-06 07:49:17 +0000509 return Error("no segments found in object");
Danil Malyshevcf852dc2011-07-13 07:57:58 +0000510
511 // Read and register the symbol table data.
512 InMemoryStruct<macho::SymtabLoadCommand> SymtabLC;
513 Obj->ReadSymtabLoadCommand(*SymtabLCI, SymtabLC);
514 if (!SymtabLC)
515 return Error("unable to load symbol table load command");
516 Obj->RegisterStringTable(*SymtabLC);
517
518 // Read the dynamic link-edit information, if present (not present in static
519 // objects).
520 if (DysymtabLCI) {
521 InMemoryStruct<macho::DysymtabLoadCommand> DysymtabLC;
522 Obj->ReadDysymtabLoadCommand(*DysymtabLCI, DysymtabLC);
523 if (!DysymtabLC)
524 return Error("unable to load dynamic link-exit load command");
525
526 // FIXME: We don't support anything interesting yet.
527// if (DysymtabLC->LocalSymbolsIndex != 0)
528// return Error("NOT YET IMPLEMENTED: local symbol entries");
529// if (DysymtabLC->ExternalSymbolsIndex != 0)
530// return Error("NOT YET IMPLEMENTED: non-external symbol entries");
531// if (DysymtabLC->UndefinedSymbolsIndex != SymtabLC->NumSymbolTableEntries)
532// return Error("NOT YET IMPLEMENTED: undefined symbol entries");
533 }
534
535 // Load the segment load command.
536 if (SegmentLCI->Command.Type == macho::LCT_Segment) {
537 if (loadSegment32(Obj.get(), SegmentLCI, SymtabLC))
538 return true;
539 } else {
540 if (loadSegment64(Obj.get(), SegmentLCI, SymtabLC))
541 return true;
542 }
543
Jim Grosbach61425c02012-01-16 22:26:39 +0000544 // Assign the addresses of the sections from the object so that any
545 // relocations to them get set properly.
546 // FIXME: This is done directly from the client at the moment. We should
547 // default the values to the local storage, at least when the target arch
548 // is the same as the host arch.
549
Danil Malyshevcf852dc2011-07-13 07:57:58 +0000550 return false;
551}
552
553// Assign an address to a symbol name and resolve all the relocations
554// associated with it.
Jim Grosbach61425c02012-01-16 22:26:39 +0000555void RuntimeDyldMachO::reassignSectionAddress(unsigned SectionID,
556 uint64_t Addr) {
557 // The address to use for relocation resolution is not
558 // the address of the local section buffer. We must be doing
559 // a remote execution environment of some sort. Re-apply any
560 // relocations referencing this section with the given address.
561 //
562 // Addr is a uint64_t because we can't assume the pointer width
563 // of the target is the same as that of the host. Just use a generic
564 // "big enough" type.
Danil Malyshevcf852dc2011-07-13 07:57:58 +0000565
Jim Grosbach61425c02012-01-16 22:26:39 +0000566 SectionLoadAddress[SectionID] = Addr;
567
568 RelocationList &Relocs = Relocations[SectionID];
Danil Malyshevcf852dc2011-07-13 07:57:58 +0000569 for (unsigned i = 0, e = Relocs.size(); i != e; ++i) {
570 RelocationEntry &RE = Relocs[i];
Jim Grosbach61425c02012-01-16 22:26:39 +0000571 uint8_t *Target = (uint8_t*)Sections[RE.SectionID].base() + RE.Offset;
Danil Malyshevcf852dc2011-07-13 07:57:58 +0000572 bool isPCRel = (RE.Data >> 24) & 1;
573 unsigned Type = (RE.Data >> 28) & 0xf;
574 unsigned Size = 1 << ((RE.Data >> 25) & 3);
575
Jim Grosbach61425c02012-01-16 22:26:39 +0000576 DEBUG(dbgs() << "Resolving relocation at Section #" << RE.SectionID
577 << " + " << RE.Offset << " (" << format("%p", Target) << ")"
578 << " from Section #" << SectionID << " (" << format("%p", Addr) << ")"
Danil Malyshevcf852dc2011-07-13 07:57:58 +0000579 << "(" << (isPCRel ? "pcrel" : "absolute")
Jim Grosbach61425c02012-01-16 22:26:39 +0000580 << ", type: " << Type << ", Size: " << Size << ", Addend: "
581 << RE.Addend << ").\n");
Danil Malyshevcf852dc2011-07-13 07:57:58 +0000582
Jim Grosbach61425c02012-01-16 22:26:39 +0000583 resolveRelocation(Target, Addr, isPCRel, Type, Size, RE.Addend);
Danil Malyshevcf852dc2011-07-13 07:57:58 +0000584 }
585}
586
587bool RuntimeDyldMachO::isKnownFormat(const MemoryBuffer *InputBuffer) {
588 StringRef Magic = InputBuffer->getBuffer().slice(0, 4);
589 if (Magic == "\xFE\xED\xFA\xCE") return true;
590 if (Magic == "\xCE\xFA\xED\xFE") return true;
591 if (Magic == "\xFE\xED\xFA\xCF") return true;
592 if (Magic == "\xCF\xFA\xED\xFE") return true;
593 return false;
594}
595
596} // end namespace llvm