GNULDBackend.h revision 22add6ff3426df1a85089fe6a6e1597ee3b6f300 
1//===- GNULDBackend.h -----------------------------------------------------===//
2//
3//                     The MCLinker Project
4//
5// This file is distributed under the University of Illinois Open Source
6// License. See LICENSE.TXT for details.
7//
8//===----------------------------------------------------------------------===//
9#ifndef MCLD_TARGET_GNU_LDBACKEND_H
10#define MCLD_TARGET_GNU_LDBACKEND_H
11#ifdef ENABLE_UNITTEST
12#include <gtest.h>
13#endif
14#include <mcld/Target/TargetLDBackend.h>
15
16#include <llvm/Support/ELF.h>
17#include <mcld/ADT/HashTable.h>
18#include <mcld/ADT/HashEntry.h>
19#include <mcld/LD/ELFDynObjFileFormat.h>
20#include <mcld/LD/ELFExecFileFormat.h>
21#include <mcld/LD/ELFObjectFileFormat.h>
22#include <mcld/LD/GNUArchiveReader.h>
23#include <mcld/LD/ELFObjectReader.h>
24#include <mcld/LD/ELFDynObjReader.h>
25#include <mcld/LD/ELFDynObjWriter.h>
26#include <mcld/LD/ELFExecWriter.h>
27#include <mcld/LD/ELFObjectWriter.h>
28#include <mcld/LD/ELFSegment.h>
29#include <mcld/LD/ELFSegmentFactory.h>
30#include <mcld/Target/ELFDynamic.h>
31
32#include <mcld/Support/GCFactory.h>
33#include <mcld/Module.h>
34
35namespace mcld {
36
37class Module;
38class LinkerConfig;
39class Layout;
40class EhFrame;
41class EhFrameHdr;
42class BranchIslandFactory;
43class StubFactory;
44
45/** \class GNULDBackend
46 *  \brief GNULDBackend provides a common interface for all GNU Unix-OS
47 *  LDBackend.
48 */
49class GNULDBackend : public TargetLDBackend
50{
51protected:
52  GNULDBackend(const LinkerConfig& pConfig);
53
54public:
55  virtual ~GNULDBackend();
56
57  // -----  readers/writers  ----- //
58  GNUArchiveReader* createArchiveReader(Module& pModule);
59  ELFObjectReader* createObjectReader(FragmentLinker& pLinker);
60  ELFDynObjReader* createDynObjReader(FragmentLinker& pLinker);
61  ELFObjectWriter* createObjectWriter(FragmentLinker& pLinker);
62  ELFDynObjWriter* createDynObjWriter(FragmentLinker& pLinker);
63  ELFExecWriter*   createExecWriter(FragmentLinker& pLinker);
64
65  // -----  output sections  ----- //
66  /// initStdSections - initialize standard sections of the output file.
67  bool initStdSections(ObjectBuilder& pBuilder);
68
69  /// getOutputFormat - get the sections of the output file.
70  const ELFFileFormat* getOutputFormat() const;
71  ELFFileFormat*       getOutputFormat();
72
73  // -----  target symbols ----- //
74  /// initStandardSymbols - initialize standard symbols.
75  /// Some section symbols is undefined in input object, and linkers must set
76  /// up its value. Take __init_array_begin for example. This symbol is an
77  /// undefined symbol in input objects. FragmentLinker must finalize its value
78  /// to the begin of the .init_array section, then relocation enties to
79  /// __init_array_begin can be applied without emission of "undefined
80  /// reference to `__init_array_begin'".
81  bool initStandardSymbols(FragmentLinker& pLinker, Module& pModule);
82
83  /// finalizeSymbol - Linker checks pSymbol.reserved() if it's not zero,
84  /// then it will ask backend to finalize the symbol value.
85  /// @return ture - if backend set the symbol value sucessfully
86  /// @return false - if backend do not recognize the symbol
87  bool finalizeSymbols(FragmentLinker& pLinker) {
88    return (finalizeStandardSymbols(pLinker) &&
89            finalizeTargetSymbols(pLinker));
90  }
91
92  /// finalizeStandardSymbols - set the value of standard symbols
93  virtual bool finalizeStandardSymbols(FragmentLinker& pLinker);
94
95  /// finalizeTargetSymbols - set the value of target symbols
96  virtual bool finalizeTargetSymbols(FragmentLinker& pLinker) = 0;
97
98  /// finalizeTLSSymbol - set the value of a TLS symbol
99  virtual bool finalizeTLSSymbol(LDSymbol& pSymbol);
100
101  size_t sectionStartOffset() const;
102
103  /// The return value of machine() it the same as e_machine in the ELF header*/
104  virtual uint32_t machine() const = 0;
105
106  /// ELFVersion - the value of e_ident[EI_VERSION]
107  virtual uint8_t ELFVersion() const
108  { return llvm::ELF::EV_CURRENT; }
109
110  /// OSABI - the value of e_ident[EI_OSABI]
111  virtual uint8_t OSABI() const = 0;
112
113  /// ABIVersion - the value of e_ident[EI_ABIVRESION]
114  virtual uint8_t ABIVersion() const = 0;
115
116  /// flags - the value of ElfXX_Ehdr::e_flags
117  virtual uint64_t flags() const = 0;
118
119  /// entry - the symbol name of the entry point
120  virtual const char* entry() const
121  { return "_start"; }
122
123  /// dyld - the name of the default dynamic linker
124  /// target may override this function if needed.
125  /// @ref gnu ld, bfd/elf32-i386.c:521
126  virtual const char* dyld() const
127  { return "/usr/lib/libc.so.1"; }
128
129  /// defaultTextSegmentAddr - target should specify its own default start address
130  /// of the text segment. esp. for exec.
131  virtual uint64_t defaultTextSegmentAddr() const
132  { return 0x0; }
133
134  bool hasTextRel() const
135  { return m_bHasTextRel; }
136
137  bool hasStaticTLS() const
138  { return m_bHasStaticTLS; }
139
140  /// segmentStartAddr - this function returns the start address of the segment
141  uint64_t segmentStartAddr(const FragmentLinker& pLinker) const;
142
143  /// partialScanRelocation - When doing partial linking, fix the relocation
144  /// offset after section merge
145  void partialScanRelocation(Relocation& pReloc,
146                             FragmentLinker& pLinker,
147                             Module& pModule,
148                             const LDSection& pSection);
149
150  /// sizeNamePools - compute the size of regular name pools
151  /// In ELF executable files, regular name pools are .symtab, .strtab.,
152  /// .dynsym, .dynstr, and .hash
153  virtual void sizeNamePools(const Module& pModule, bool pIsStaticLink);
154
155  /// emitSectionData - emit target-dependent section data
156  virtual uint64_t emitSectionData(const LDSection& pSection,
157                                   MemoryRegion& pRegion) const = 0;
158
159  /// emitRegNamePools - emit regular name pools - .symtab, .strtab
160  virtual void emitRegNamePools(const Module& pModule,
161                                MemoryArea& pOutput);
162
163  /// emitNamePools - emit dynamic name pools - .dyntab, .dynstr, .hash
164  virtual void emitDynNamePools(const Module& pModule,
165                                MemoryArea& pOutput);
166
167  /// sizeInterp - compute the size of program interpreter's name
168  /// In ELF executables, this is the length of dynamic linker's path name
169  virtual void sizeInterp();
170
171  /// emitInterp - emit the .interp
172  virtual void emitInterp(MemoryArea& pOutput);
173
174  /// getSectionOrder - compute the layout order of the section
175  /// Layout calls this function to get the default order of the pSectHdr.
176  /// If the pSectHdr.type() is LDFileFormat::Target, then getSectionOrder()
177  /// will call getTargetSectionOrder().
178  ///
179  /// If targets favors certain order for general sections, please override
180  /// this function.
181  ///
182  /// @see getTargetSectionOrder
183  virtual unsigned int getSectionOrder(const LDSection& pSectHdr) const;
184
185  /// getTargetSectionOrder - compute the layout order of target section
186  /// If the target favors certain order for the given gSectHdr, please
187  /// override this function.
188  ///
189  /// By default, this function returns the maximun order, and pSectHdr
190  /// will be the last section to be laid out.
191  virtual unsigned int getTargetSectionOrder(const LDSection& pSectHdr) const
192  { return (unsigned int)-1; }
193
194  /// numOfSegments - return the number of segments
195  /// if the target favors other ways to emit program header, please override
196  /// this function
197  virtual unsigned int numOfSegments() const
198  { return m_ELFSegmentTable.size(); }
199
200  /// elfSegmentTable - return the reference of the elf segment table
201  ELFSegmentFactory& elfSegmentTable()
202  { return m_ELFSegmentTable; }
203
204  /// elfSegmentTable - return the reference of the elf segment table
205  const ELFSegmentFactory& elfSegmentTable() const
206  { return m_ELFSegmentTable; }
207
208  /// commonPageSize - the common page size of the target machine, and we set it
209  /// to 4K here. If target favors the different size, please override this
210  /// function
211  virtual uint64_t commonPageSize() const;
212
213  /// abiPageSize - the abi page size of the target machine, and we set it to 4K
214  /// here. If target favors the different size, please override this function
215  virtual uint64_t abiPageSize() const;
216
217  /// getSymbolIdx - get the symbol index of ouput symbol table
218  size_t getSymbolIdx(LDSymbol* pSymbol) const;
219
220  /// isDefaultExecStack - target should specify whether the stack is default
221  /// executable. If target favors another choice, please override this function
222  virtual bool isDefaultExecStack() const
223  { return true; }
224
225  /// allocateCommonSymbols - allocate common symbols in the corresponding
226  /// sections.
227  /// Different concrete target backend may overlap this function.
228  virtual bool allocateCommonSymbols(Module& pModule);
229
230  /// isSymbolPreemtible - whether the symbol can be preemted by other
231  /// link unit
232  /// @ref Google gold linker, symtab.h:551
233  bool isSymbolPreemptible(const ResolveInfo& pSym) const;
234
235  /// symbolNeedsDynRel - return whether the symbol needs a dynamic relocation
236  /// @ref Google gold linker, symtab.h:645
237  bool symbolNeedsDynRel(const FragmentLinker& pLinker,
238                         const ResolveInfo& pSym,
239                         bool pSymHasPLT,
240                         bool isAbsReloc) const;
241
242  // getTDATASymbol - get section symbol of .tdata
243  LDSymbol& getTDATASymbol();
244  const LDSymbol& getTDATASymbol() const;
245
246  /// getTBSSSymbol - get section symbol of .tbss
247  LDSymbol& getTBSSSymbol();
248  const LDSymbol& getTBSSSymbol() const;
249
250  //  -----  relaxation  -----  //
251  /// initBRIslandFactory - initialize the branch island factory for relaxation
252  bool initBRIslandFactory();
253
254  /// initStubFactory - initialize the stub factory for relaxation
255  bool initStubFactory();
256
257  /// getBRIslandFactory
258  BranchIslandFactory* getBRIslandFactory() { return m_pBRIslandFactory; }
259
260  /// getStubFactory
261  StubFactory*         getStubFactory()     { return m_pStubFactory; }
262
263  /// maxBranchOffset - return the max (forward) branch offset of the backend.
264  /// Target can override this function if needed.
265  virtual uint64_t maxBranchOffset() { return (uint64_t)-1; }
266
267protected:
268  uint64_t getSymbolSize(const LDSymbol& pSymbol) const;
269
270  uint64_t getSymbolInfo(const LDSymbol& pSymbol) const;
271
272  uint64_t getSymbolValue(const LDSymbol& pSymbol) const;
273
274  uint64_t getSymbolShndx(const LDSymbol& pSymbol) const;
275
276  /// getHashBucketCount - calculate hash bucket count.
277  /// @ref Google gold linker, dynobj.cc:791
278  static unsigned getHashBucketCount(unsigned pNumOfSymbols, bool pIsGNUStyle);
279
280  /// isDynamicSymbol
281  /// @ref Google gold linker: symtab.cc:311
282  bool isDynamicSymbol(const LDSymbol& pSymbol);
283
284  /// isDynamicSymbol
285  /// @ref Google gold linker: symtab.cc:311
286  bool isDynamicSymbol(const ResolveInfo& pResolveInfo);
287
288  /// symbolNeedsPLT - return whether the symbol needs a PLT entry
289  /// @ref Google gold linker, symtab.h:596
290  bool symbolNeedsPLT(const FragmentLinker& pLinker,
291                      const ResolveInfo& pSym) const;
292
293  /// symbolNeedsCopyReloc - return whether the symbol needs a copy relocation
294  bool symbolNeedsCopyReloc(const FragmentLinker& pLinker,
295                            const Relocation& pReloc,
296                            const ResolveInfo& pSym) const;
297
298  /// symbolHasFinalValue - return true if the symbol's value can be decided at
299  /// link time
300  bool symbolFinalValueIsKnown(const FragmentLinker& pLinker,
301                               const ResolveInfo& pSym) const;
302
303  /// emitSymbol32 - emit an ELF32 symbol
304  void emitSymbol32(llvm::ELF::Elf32_Sym& pSym32,
305                    LDSymbol& pSymbol,
306                    char* pStrtab,
307                    size_t pStrtabsize,
308                    size_t pSymtabIdx);
309
310  /// emitSymbol64 - emit an ELF64 symbol
311  void emitSymbol64(llvm::ELF::Elf64_Sym& pSym64,
312                    LDSymbol& pSymbol,
313                    char* pStrtab,
314                    size_t pStrtabsize,
315                    size_t pSymtabIdx);
316
317  /// checkAndSetHasTextRel - check pSection flag to set HasTextRel
318  void checkAndSetHasTextRel(const LDSection& pSection);
319
320  void setHasStaticTLS(bool pVal = true)
321  { m_bHasStaticTLS = pVal; }
322
323private:
324  /// createProgramHdrs - base on output sections to create the program headers
325  void createProgramHdrs(Module& pModule, const FragmentLinker& pLinker);
326
327  /// doCreateProgramHdrs - backend can implement this function to create the
328  /// target-dependent segments
329  virtual void doCreateProgramHdrs(Module& pModule,
330                                   const FragmentLinker& pLinker) = 0;
331
332  /// setupProgramHdrs - set up the attributes of segments
333  ///  (i.e., offset, addresses, file/mem size, flag,  and alignment)
334  void setupProgramHdrs(const FragmentLinker& pLinker);
335
336  /// getSegmentFlag - give a section flag and return the corresponding segment
337  /// flag
338  inline uint32_t getSegmentFlag(const uint32_t pSectionFlag)
339  {
340    uint32_t flag = llvm::ELF::PF_R;
341    if (0 != (pSectionFlag & llvm::ELF::SHF_WRITE))
342      flag |= llvm::ELF::PF_W;
343    if (0 != (pSectionFlag & llvm::ELF::SHF_EXECINSTR))
344      flag |= llvm::ELF::PF_X;
345    return flag;
346  }
347
348  /// setupGNUStackInfo - setup the section flag of .note.GNU-stack in output
349  void setupGNUStackInfo(Module& pModule, FragmentLinker& pLinker);
350
351  /// setupRelro - setup the offset constraint of PT_RELRO
352  void setupRelro(Module& pModule);
353
354  /// setOutputSectionOffset - helper function to set a group of output sections'
355  /// offset, and set pSectBegin to pStartOffset if pStartOffset is not -1U.
356  void setOutputSectionOffset(Module& pModule,
357                              Module::iterator pSectBegin,
358                              Module::iterator pSectEnd,
359                              uint64_t pStartOffset = -1U);
360
361  /// setOutputSectionOffset - helper function to set output sections' address.
362  void setOutputSectionAddress(FragmentLinker& pLinker,
363                               Module& pModule,
364                               Module::iterator pSectBegin,
365                               Module::iterator pSectEnd);
366
367  /// preLayout - Backend can do any needed modification before layout
368  void preLayout(Module& pModule, FragmentLinker& pLinker);
369
370  /// postLayout -Backend can do any needed modification after layout
371  void postLayout(Module& pModule, FragmentLinker& pLinker);
372
373  /// preLayout - Backend can do any needed modification before layout
374  virtual void doPreLayout(FragmentLinker& pLinker) = 0;
375
376  /// postLayout -Backend can do any needed modification after layout
377  virtual void doPostLayout(Module& pModule, FragmentLinker& pLinker) = 0;
378
379  /// postProcessing - Backend can do any needed modification in the final stage
380  void postProcessing(FragmentLinker& pLinker, MemoryArea& pOutput);
381
382  /// dynamic - the dynamic section of the target machine.
383  virtual ELFDynamic& dynamic() = 0;
384
385  /// dynamic - the dynamic section of the target machine.
386  virtual const ELFDynamic& dynamic() const = 0;
387
388  /// relax - the relaxation pass
389  bool relax(Module& pModule, FragmentLinker& pLinker);
390
391  /// mayRelax - Backends should override this function if they need relaxation
392  virtual bool mayRelax() { return false; }
393
394  /// doRelax - Backend can orevride this function to add its relaxation
395  /// implementation. Return true if the output (e.g., .text) is "relaxed"
396  /// (i.e. layout is changed), and set pFinished to true if everything is fit,
397  /// otherwise set it to false.
398  virtual bool doRelax(Module& pModule, FragmentLinker& pLinker, bool& pFinished)
399  { return false; }
400
401  /// getRelEntrySize - the size in BYTE of rel type relocation
402  virtual size_t getRelEntrySize() = 0;
403
404  /// getRelEntrySize - the size in BYTE of rela type relocation
405  virtual size_t getRelaEntrySize() = 0;
406
407protected:
408  // Based on Kind in LDFileFormat to define basic section orders for ELF, and
409  // refer gold linker to add more enumerations to handle Regular and BSS kind
410  enum SectionOrder {
411    SHO_INTERP = 1,          // .interp
412    SHO_RO_NOTE,             // .note.ABI-tag, .note.gnu.build-id
413    SHO_NAMEPOOL,            // *.hash, .dynsym, .dynstr
414    SHO_RELOCATION,          // .rel.*, .rela.*
415    SHO_REL_PLT,             // .rel.plt should come after other .rel.*
416    SHO_INIT,                // .init
417    SHO_PLT,                 // .plt
418    SHO_TEXT,                // .text
419    SHO_FINI,                // .fini
420    SHO_RO,                  // .rodata
421    SHO_EXCEPTION,           // .eh_frame_hdr, .eh_frame, .gcc_except_table
422    SHO_TLS_DATA,            // .tdata
423    SHO_TLS_BSS,             // .tbss
424    SHO_RELRO_LOCAL,         // .data.rel.ro.local
425    SHO_RELRO,               // .data.rel.ro,
426    SHO_RELRO_LAST,          // for x86 to adjust .got if needed
427    SHO_NON_RELRO_FIRST,     // for x86 to adjust .got.plt if needed
428    SHO_DATA,                // .data
429    SHO_LARGE_DATA,          // .ldata
430    SHO_RW_NOTE,             //
431    SHO_SMALL_DATA,          // .sdata
432    SHO_SMALL_BSS,           // .sbss
433    SHO_BSS,                 // .bss
434    SHO_LARGE_BSS,           // .lbss
435    SHO_UNDEFINED,           // default order
436    SHO_STRTAB               // .strtab
437  };
438
439  struct SymCompare
440  {
441    bool operator()(const LDSymbol* X, const LDSymbol* Y) const
442    { return (X==Y); }
443  };
444
445  struct SymPtrHash
446  {
447    size_t operator()(const LDSymbol* pKey) const
448    {
449      return (unsigned((uintptr_t)pKey) >> 4) ^
450             (unsigned((uintptr_t)pKey) >> 9);
451    }
452  };
453
454  typedef HashEntry<LDSymbol*, size_t, SymCompare> SymHashEntryType;
455  typedef HashTable<SymHashEntryType,
456                    SymPtrHash,
457                    EntryFactory<SymHashEntryType> > HashTableType;
458
459
460protected:
461  ELFObjectReader* m_pObjectReader;
462
463  // -----  file formats  ----- //
464  ELFDynObjFileFormat* m_pDynObjFileFormat;
465  ELFExecFileFormat*   m_pExecFileFormat;
466  ELFObjectFileFormat* m_pObjectFileFormat;
467
468  // ELF segment factory
469  ELFSegmentFactory m_ELFSegmentTable;
470
471  // branch island factory
472  BranchIslandFactory* m_pBRIslandFactory;
473
474  // stub factory
475  StubFactory* m_pStubFactory;
476
477  // map the LDSymbol to its index in the output symbol table
478  HashTableType* m_pSymIndexMap;
479
480  /// m_pEhFrame - section .eh_frame
481  EhFrame* m_pEhFrame;
482
483  // section .eh_frame_hdr
484  EhFrameHdr* m_pEhFrameHdr;
485
486  // ----- dynamic flags ----- //
487  // DF_TEXTREL of DT_FLAGS
488  bool m_bHasTextRel;
489
490  // DF_STATIC_TLS of DT_FLAGS
491  bool m_bHasStaticTLS;
492
493  // -----  standard symbols  ----- //
494  // section symbols
495  LDSymbol* f_pPreInitArrayStart;
496  LDSymbol* f_pPreInitArrayEnd;
497  LDSymbol* f_pInitArrayStart;
498  LDSymbol* f_pInitArrayEnd;
499  LDSymbol* f_pFiniArrayStart;
500  LDSymbol* f_pFiniArrayEnd;
501  LDSymbol* f_pStack;
502  LDSymbol* f_pDynamic;
503
504  // section symbols for .tdata and .tbss
505  LDSymbol* f_pTDATA;
506  LDSymbol* f_pTBSS;
507
508  // segment symbols
509  LDSymbol* f_pExecutableStart;
510  LDSymbol* f_pEText;
511  LDSymbol* f_p_EText;
512  LDSymbol* f_p__EText;
513  LDSymbol* f_pEData;
514  LDSymbol* f_p_EData;
515  LDSymbol* f_pBSSStart;
516  LDSymbol* f_pEnd;
517  LDSymbol* f_p_End;
518};
519
520} // namespace of mcld
521
522#endif
523
524