1//===-- COFFDump.cpp - COFF-specific dumper ---------------------*- 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/// \file
11/// \brief This file implements the COFF-specific dumper for llvm-objdump.
12/// It outputs the Win64 EH data structures as plain text.
13/// The encoding of the unwind codes is decribed in MSDN:
14/// http://msdn.microsoft.com/en-us/library/ck9asaa9.aspx
15///
16//===----------------------------------------------------------------------===//
17
18#include "llvm-objdump.h"
19#include "llvm/Object/COFF.h"
20#include "llvm/Object/ObjectFile.h"
21#include "llvm/Support/Format.h"
22#include "llvm/Support/SourceMgr.h"
23#include "llvm/Support/Win64EH.h"
24#include "llvm/Support/raw_ostream.h"
25#include "llvm/Support/system_error.h"
26#include <algorithm>
27#include <cstring>
28
29using namespace llvm;
30using namespace object;
31using namespace llvm::Win64EH;
32
33// Returns the name of the unwind code.
34static StringRef getUnwindCodeTypeName(uint8_t Code) {
35  switch(Code) {
36  default: llvm_unreachable("Invalid unwind code");
37  case UOP_PushNonVol: return "UOP_PushNonVol";
38  case UOP_AllocLarge: return "UOP_AllocLarge";
39  case UOP_AllocSmall: return "UOP_AllocSmall";
40  case UOP_SetFPReg: return "UOP_SetFPReg";
41  case UOP_SaveNonVol: return "UOP_SaveNonVol";
42  case UOP_SaveNonVolBig: return "UOP_SaveNonVolBig";
43  case UOP_SaveXMM128: return "UOP_SaveXMM128";
44  case UOP_SaveXMM128Big: return "UOP_SaveXMM128Big";
45  case UOP_PushMachFrame: return "UOP_PushMachFrame";
46  }
47}
48
49// Returns the name of a referenced register.
50static StringRef getUnwindRegisterName(uint8_t Reg) {
51  switch(Reg) {
52  default: llvm_unreachable("Invalid register");
53  case 0: return "RAX";
54  case 1: return "RCX";
55  case 2: return "RDX";
56  case 3: return "RBX";
57  case 4: return "RSP";
58  case 5: return "RBP";
59  case 6: return "RSI";
60  case 7: return "RDI";
61  case 8: return "R8";
62  case 9: return "R9";
63  case 10: return "R10";
64  case 11: return "R11";
65  case 12: return "R12";
66  case 13: return "R13";
67  case 14: return "R14";
68  case 15: return "R15";
69  }
70}
71
72// Calculates the number of array slots required for the unwind code.
73static unsigned getNumUsedSlots(const UnwindCode &UnwindCode) {
74  switch (UnwindCode.getUnwindOp()) {
75  default: llvm_unreachable("Invalid unwind code");
76  case UOP_PushNonVol:
77  case UOP_AllocSmall:
78  case UOP_SetFPReg:
79  case UOP_PushMachFrame:
80    return 1;
81  case UOP_SaveNonVol:
82  case UOP_SaveXMM128:
83    return 2;
84  case UOP_SaveNonVolBig:
85  case UOP_SaveXMM128Big:
86    return 3;
87  case UOP_AllocLarge:
88    return (UnwindCode.getOpInfo() == 0) ? 2 : 3;
89  }
90}
91
92// Prints one unwind code. Because an unwind code can occupy up to 3 slots in
93// the unwind codes array, this function requires that the correct number of
94// slots is provided.
95static void printUnwindCode(ArrayRef<UnwindCode> UCs) {
96  assert(UCs.size() >= getNumUsedSlots(UCs[0]));
97  outs() <<  format("    0x%02x: ", unsigned(UCs[0].u.CodeOffset))
98         << getUnwindCodeTypeName(UCs[0].getUnwindOp());
99  switch (UCs[0].getUnwindOp()) {
100  case UOP_PushNonVol:
101    outs() << " " << getUnwindRegisterName(UCs[0].getOpInfo());
102    break;
103  case UOP_AllocLarge:
104    if (UCs[0].getOpInfo() == 0) {
105      outs() << " " << UCs[1].FrameOffset;
106    } else {
107      outs() << " " << UCs[1].FrameOffset
108                       + (static_cast<uint32_t>(UCs[2].FrameOffset) << 16);
109    }
110    break;
111  case UOP_AllocSmall:
112    outs() << " " << ((UCs[0].getOpInfo() + 1) * 8);
113    break;
114  case UOP_SetFPReg:
115    outs() << " ";
116    break;
117  case UOP_SaveNonVol:
118    outs() << " " << getUnwindRegisterName(UCs[0].getOpInfo())
119           << format(" [0x%04x]", 8 * UCs[1].FrameOffset);
120    break;
121  case UOP_SaveNonVolBig:
122    outs() << " " << getUnwindRegisterName(UCs[0].getOpInfo())
123           << format(" [0x%08x]", UCs[1].FrameOffset
124                    + (static_cast<uint32_t>(UCs[2].FrameOffset) << 16));
125    break;
126  case UOP_SaveXMM128:
127    outs() << " XMM" << static_cast<uint32_t>(UCs[0].getOpInfo())
128           << format(" [0x%04x]", 16 * UCs[1].FrameOffset);
129    break;
130  case UOP_SaveXMM128Big:
131    outs() << " XMM" << UCs[0].getOpInfo()
132           << format(" [0x%08x]", UCs[1].FrameOffset
133                           + (static_cast<uint32_t>(UCs[2].FrameOffset) << 16));
134    break;
135  case UOP_PushMachFrame:
136    outs() << " " << (UCs[0].getOpInfo() ? "w/o" : "w")
137           << " error code";
138    break;
139  }
140  outs() << "\n";
141}
142
143static void printAllUnwindCodes(ArrayRef<UnwindCode> UCs) {
144  for (const UnwindCode *I = UCs.begin(), *E = UCs.end(); I < E; ) {
145    unsigned UsedSlots = getNumUsedSlots(*I);
146    if (UsedSlots > UCs.size()) {
147      outs() << "Unwind data corrupted: Encountered unwind op "
148             << getUnwindCodeTypeName((*I).getUnwindOp())
149             << " which requires " << UsedSlots
150             << " slots, but only " << UCs.size()
151             << " remaining in buffer";
152      return ;
153    }
154    printUnwindCode(ArrayRef<UnwindCode>(I, E));
155    I += UsedSlots;
156  }
157}
158
159// Given a symbol sym this functions returns the address and section of it.
160static error_code resolveSectionAndAddress(const COFFObjectFile *Obj,
161                                           const SymbolRef &Sym,
162                                           const coff_section *&ResolvedSection,
163                                           uint64_t &ResolvedAddr) {
164  if (error_code ec = Sym.getAddress(ResolvedAddr)) return ec;
165  section_iterator iter(Obj->begin_sections());
166  if (error_code ec = Sym.getSection(iter)) return ec;
167  ResolvedSection = Obj->getCOFFSection(iter);
168  return object_error::success;
169}
170
171// Given a vector of relocations for a section and an offset into this section
172// the function returns the symbol used for the relocation at the offset.
173static error_code resolveSymbol(const std::vector<RelocationRef> &Rels,
174                                uint64_t Offset, SymbolRef &Sym) {
175  for (std::vector<RelocationRef>::const_iterator I = Rels.begin(),
176                                                  E = Rels.end();
177                                                  I != E; ++I) {
178    uint64_t Ofs;
179    if (error_code ec = I->getOffset(Ofs)) return ec;
180    if (Ofs == Offset) {
181      Sym = *I->getSymbol();
182      break;
183    }
184  }
185  return object_error::success;
186}
187
188// Given a vector of relocations for a section and an offset into this section
189// the function resolves the symbol used for the relocation at the offset and
190// returns the section content and the address inside the content pointed to
191// by the symbol.
192static error_code getSectionContents(const COFFObjectFile *Obj,
193                                     const std::vector<RelocationRef> &Rels,
194                                     uint64_t Offset,
195                                     ArrayRef<uint8_t> &Contents,
196                                     uint64_t &Addr) {
197  SymbolRef Sym;
198  if (error_code ec = resolveSymbol(Rels, Offset, Sym)) return ec;
199  const coff_section *Section;
200  if (error_code ec = resolveSectionAndAddress(Obj, Sym, Section, Addr))
201    return ec;
202  if (error_code ec = Obj->getSectionContents(Section, Contents)) return ec;
203  return object_error::success;
204}
205
206// Given a vector of relocations for a section and an offset into this section
207// the function returns the name of the symbol used for the relocation at the
208// offset.
209static error_code resolveSymbolName(const std::vector<RelocationRef> &Rels,
210                                    uint64_t Offset, StringRef &Name) {
211  SymbolRef Sym;
212  if (error_code ec = resolveSymbol(Rels, Offset, Sym)) return ec;
213  if (error_code ec = Sym.getName(Name)) return ec;
214  return object_error::success;
215}
216
217static void printCOFFSymbolAddress(llvm::raw_ostream &Out,
218                                   const std::vector<RelocationRef> &Rels,
219                                   uint64_t Offset, uint32_t Disp) {
220  StringRef Sym;
221  if (error_code ec = resolveSymbolName(Rels, Offset, Sym)) {
222    error(ec);
223    return ;
224  }
225  Out << Sym;
226  if (Disp > 0)
227    Out << format(" + 0x%04x", Disp);
228}
229
230void llvm::printCOFFUnwindInfo(const COFFObjectFile *Obj) {
231  const coff_file_header *Header;
232  if (error(Obj->getCOFFHeader(Header))) return;
233
234  if (Header->Machine != COFF::IMAGE_FILE_MACHINE_AMD64) {
235    errs() << "Unsupported image machine type "
236              "(currently only AMD64 is supported).\n";
237    return;
238  }
239
240  const coff_section *Pdata = 0;
241
242  error_code ec;
243  for (section_iterator SI = Obj->begin_sections(),
244                        SE = Obj->end_sections();
245                        SI != SE; SI.increment(ec)) {
246    if (error(ec)) return;
247
248    StringRef Name;
249    if (error(SI->getName(Name))) continue;
250
251    if (Name != ".pdata") continue;
252
253    Pdata = Obj->getCOFFSection(SI);
254    std::vector<RelocationRef> Rels;
255    for (relocation_iterator RI = SI->begin_relocations(),
256                             RE = SI->end_relocations();
257                             RI != RE; RI.increment(ec)) {
258      if (error(ec)) break;
259      Rels.push_back(*RI);
260    }
261
262    // Sort relocations by address.
263    std::sort(Rels.begin(), Rels.end(), RelocAddressLess);
264
265    ArrayRef<uint8_t> Contents;
266    if (error(Obj->getSectionContents(Pdata, Contents))) continue;
267    if (Contents.empty()) continue;
268
269    ArrayRef<RuntimeFunction> RFs(
270                  reinterpret_cast<const RuntimeFunction *>(Contents.data()),
271                                  Contents.size() / sizeof(RuntimeFunction));
272    for (const RuntimeFunction *I = RFs.begin(), *E = RFs.end(); I < E; ++I) {
273      const uint64_t SectionOffset = std::distance(RFs.begin(), I)
274                                     * sizeof(RuntimeFunction);
275
276      outs() << "Function Table:\n";
277
278      outs() << "  Start Address: ";
279      printCOFFSymbolAddress(outs(), Rels, SectionOffset +
280                             /*offsetof(RuntimeFunction, StartAddress)*/ 0,
281                             I->StartAddress);
282      outs() << "\n";
283
284      outs() << "  End Address: ";
285      printCOFFSymbolAddress(outs(), Rels, SectionOffset +
286                             /*offsetof(RuntimeFunction, EndAddress)*/ 4,
287                             I->EndAddress);
288      outs() << "\n";
289
290      outs() << "  Unwind Info Address: ";
291      printCOFFSymbolAddress(outs(), Rels, SectionOffset +
292                             /*offsetof(RuntimeFunction, UnwindInfoOffset)*/ 8,
293                             I->UnwindInfoOffset);
294      outs() << "\n";
295
296      ArrayRef<uint8_t> XContents;
297      uint64_t UnwindInfoOffset = 0;
298      if (error(getSectionContents(Obj, Rels, SectionOffset +
299                              /*offsetof(RuntimeFunction, UnwindInfoOffset)*/ 8,
300                                   XContents, UnwindInfoOffset))) continue;
301      if (XContents.empty()) continue;
302
303      UnwindInfoOffset += I->UnwindInfoOffset;
304      if (UnwindInfoOffset > XContents.size()) continue;
305
306      const Win64EH::UnwindInfo *UI =
307                            reinterpret_cast<const Win64EH::UnwindInfo *>
308                              (XContents.data() + UnwindInfoOffset);
309
310      // The casts to int are required in order to output the value as number.
311      // Without the casts the value would be interpreted as char data (which
312      // results in garbage output).
313      outs() << "  Version: " << static_cast<int>(UI->getVersion()) << "\n";
314      outs() << "  Flags: " << static_cast<int>(UI->getFlags());
315      if (UI->getFlags()) {
316          if (UI->getFlags() & UNW_ExceptionHandler)
317            outs() << " UNW_ExceptionHandler";
318          if (UI->getFlags() & UNW_TerminateHandler)
319            outs() << " UNW_TerminateHandler";
320          if (UI->getFlags() & UNW_ChainInfo)
321            outs() << " UNW_ChainInfo";
322      }
323      outs() << "\n";
324      outs() << "  Size of prolog: "
325             << static_cast<int>(UI->PrologSize) << "\n";
326      outs() << "  Number of Codes: "
327             << static_cast<int>(UI->NumCodes) << "\n";
328      // Maybe this should move to output of UOP_SetFPReg?
329      if (UI->getFrameRegister()) {
330        outs() << "  Frame register: "
331                << getUnwindRegisterName(UI->getFrameRegister())
332                << "\n";
333        outs() << "  Frame offset: "
334                << 16 * UI->getFrameOffset()
335                << "\n";
336      } else {
337        outs() << "  No frame pointer used\n";
338      }
339      if (UI->getFlags() & (UNW_ExceptionHandler | UNW_TerminateHandler)) {
340        // FIXME: Output exception handler data
341      } else if (UI->getFlags() & UNW_ChainInfo) {
342        // FIXME: Output chained unwind info
343      }
344
345      if (UI->NumCodes)
346        outs() << "  Unwind Codes:\n";
347
348      printAllUnwindCodes(ArrayRef<UnwindCode>(&UI->UnwindCodes[0],
349                          UI->NumCodes));
350
351      outs() << "\n\n";
352      outs().flush();
353    }
354  }
355}
356