1 //===- MapFile.cpp --------------------------------------------------------===//
2 //
3 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
4 // See https://llvm.org/LICENSE.txt for license information.
5 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
6 //
7 //===----------------------------------------------------------------------===//
8 //
9 // This file implements the /map option in the same format as link.exe
10 // (based on observations)
11 //
12 // Header (program name, timestamp info, preferred load address)
13 //
14 // Section list (Start = Section index:Base address):
15 // Start         Length     Name                   Class
16 // 0001:00001000 00000015H .text                   CODE
17 //
18 // Symbols list:
19 // Address        Publics by Value    Rva + Base          Lib:Object
20 // 0001:00001000  main                 0000000140001000    main.obj
21 // 0001:00001300  ?__scrt_common_main@@YAHXZ  0000000140001300 libcmt:exe_main.obj
22 //
23 // entry point at        0001:00000360
24 //
25 // Static symbols
26 //
27 // 0000:00000000  __guard_fids__       0000000140000000     libcmt : exe_main.obj
28 //===----------------------------------------------------------------------===//
29 
30 #include "MapFile.h"
31 #include "COFFLinkerContext.h"
32 #include "SymbolTable.h"
33 #include "Symbols.h"
34 #include "Writer.h"
35 #include "lld/Common/ErrorHandler.h"
36 #include "lld/Common/Timer.h"
37 #include "llvm/Support/Parallel.h"
38 #include "llvm/Support/Path.h"
39 #include "llvm/Support/TimeProfiler.h"
40 #include "llvm/Support/raw_ostream.h"
41 
42 using namespace llvm;
43 using namespace llvm::object;
44 using namespace lld;
45 using namespace lld::coff;
46 
47 // Print out the first two columns of a line.
48 static void writeHeader(raw_ostream &os, uint32_t sec, uint64_t addr) {
49   os << format(" %04x:%08llx", sec, addr);
50 }
51 
52 // Write the time stamp with the format used by link.exe
53 // It seems identical to strftime with "%c" on msvc build, but we need a
54 // locale-agnostic version.
55 static void writeFormattedTimestamp(raw_ostream &os, time_t tds) {
56   constexpr const char *const days[7] = {"Sun", "Mon", "Tue", "Wed",
57                                          "Thu", "Fri", "Sat"};
58   constexpr const char *const months[12] = {"Jan", "Feb", "Mar", "Apr",
59                                             "May", "Jun", "Jul", "Aug",
60                                             "Sep", "Oct", "Nov", "Dec"};
61   tm *time = localtime(&tds);
62   os << format("%s %s %2d %02d:%02d:%02d %d", days[time->tm_wday],
63                months[time->tm_mon], time->tm_mday, time->tm_hour, time->tm_min,
64                time->tm_sec, time->tm_year + 1900);
65 }
66 
67 static void sortUniqueSymbols(std::vector<Defined *> &syms,
68                               uint64_t imageBase) {
69   // Build helper vector
70   using SortEntry = std::pair<Defined *, size_t>;
71   std::vector<SortEntry> v;
72   v.resize(syms.size());
73   for (size_t i = 0, e = syms.size(); i < e; ++i)
74     v[i] = SortEntry(syms[i], i);
75 
76   // Remove duplicate symbol pointers
77   parallelSort(v, std::less<SortEntry>());
78   auto end = std::unique(v.begin(), v.end(),
79                          [](const SortEntry &a, const SortEntry &b) {
80                            return a.first == b.first;
81                          });
82   v.erase(end, v.end());
83 
84   // Sort by RVA then original order
85   parallelSort(v, [imageBase](const SortEntry &a, const SortEntry &b) {
86     // Add config.imageBase to avoid comparing "negative" RVAs.
87     // This can happen with symbols of Absolute kind
88     uint64_t rvaa = imageBase + a.first->getRVA();
89     uint64_t rvab = imageBase + b.first->getRVA();
90     return rvaa < rvab || (rvaa == rvab && a.second < b.second);
91   });
92 
93   syms.resize(v.size());
94   for (size_t i = 0, e = v.size(); i < e; ++i)
95     syms[i] = v[i].first;
96 }
97 
98 // Returns the lists of all symbols that we want to print out.
99 static void getSymbols(const COFFLinkerContext &ctx,
100                        std::vector<Defined *> &syms,
101                        std::vector<Defined *> &staticSyms) {
102 
103   for (ObjFile *file : ctx.objFileInstances)
104     for (Symbol *b : file->getSymbols()) {
105       if (!b || !b->isLive())
106         continue;
107       if (auto *sym = dyn_cast<DefinedCOFF>(b)) {
108         COFFSymbolRef symRef = sym->getCOFFSymbol();
109         if (!symRef.isSectionDefinition() &&
110             symRef.getStorageClass() != COFF::IMAGE_SYM_CLASS_LABEL) {
111           if (symRef.getStorageClass() == COFF::IMAGE_SYM_CLASS_STATIC)
112             staticSyms.push_back(sym);
113           else
114             syms.push_back(sym);
115         }
116       } else if (auto *sym = dyn_cast<Defined>(b)) {
117         syms.push_back(sym);
118       }
119     }
120 
121   for (ImportFile *file : ctx.importFileInstances) {
122     if (!file->live)
123       continue;
124 
125     if (!file->thunkSym)
126       continue;
127 
128     if (!file->thunkLive)
129       continue;
130 
131     if (auto *thunkSym = dyn_cast<Defined>(file->thunkSym))
132       syms.push_back(thunkSym);
133 
134     if (auto *impSym = dyn_cast_or_null<Defined>(file->impSym))
135       syms.push_back(impSym);
136   }
137 
138   sortUniqueSymbols(syms, ctx.config.imageBase);
139   sortUniqueSymbols(staticSyms, ctx.config.imageBase);
140 }
141 
142 // Construct a map from symbols to their stringified representations.
143 static DenseMap<Defined *, std::string>
144 getSymbolStrings(const COFFLinkerContext &ctx, ArrayRef<Defined *> syms) {
145   std::vector<std::string> str(syms.size());
146   parallelFor((size_t)0, syms.size(), [&](size_t i) {
147     raw_string_ostream os(str[i]);
148     Defined *sym = syms[i];
149 
150     uint16_t sectionIdx = 0;
151     uint64_t address = 0;
152     SmallString<128> fileDescr;
153 
154     if (auto *absSym = dyn_cast<DefinedAbsolute>(sym)) {
155       address = absSym->getVA();
156       fileDescr = "<absolute>";
157     } else if (isa<DefinedSynthetic>(sym)) {
158       fileDescr = "<linker-defined>";
159     } else if (isa<DefinedCommon>(sym)) {
160       fileDescr = "<common>";
161     } else if (Chunk *chunk = sym->getChunk()) {
162       address = sym->getRVA();
163       if (OutputSection *sec = ctx.getOutputSection(chunk))
164         address -= sec->header.VirtualAddress;
165 
166       sectionIdx = chunk->getOutputSectionIdx();
167 
168       InputFile *file;
169       if (auto *impSym = dyn_cast<DefinedImportData>(sym))
170         file = impSym->file;
171       else if (auto *thunkSym = dyn_cast<DefinedImportThunk>(sym))
172         file = thunkSym->wrappedSym->file;
173       else
174         file = sym->getFile();
175 
176       if (file) {
177         if (!file->parentName.empty()) {
178           fileDescr = sys::path::filename(file->parentName);
179           sys::path::replace_extension(fileDescr, "");
180           fileDescr += ":";
181         }
182         fileDescr += sys::path::filename(file->getName());
183       }
184     }
185     writeHeader(os, sectionIdx, address);
186     os << "       ";
187     os << left_justify(sym->getName(), 26);
188     os << " ";
189     os << format_hex_no_prefix((ctx.config.imageBase + sym->getRVA()), 16);
190     if (!fileDescr.empty()) {
191       os << "     "; // FIXME : Handle "f" and "i" flags sometimes generated
192                      // by link.exe in those spaces
193       os << fileDescr;
194     }
195   });
196 
197   DenseMap<Defined *, std::string> ret;
198   for (size_t i = 0, e = syms.size(); i < e; ++i)
199     ret[syms[i]] = std::move(str[i]);
200   return ret;
201 }
202 
203 void lld::coff::writeMapFile(COFFLinkerContext &ctx) {
204   if (ctx.config.mapFile.empty())
205     return;
206 
207   llvm::TimeTraceScope timeScope("Map file");
208   std::error_code ec;
209   raw_fd_ostream os(ctx.config.mapFile, ec, sys::fs::OF_None);
210   if (ec)
211     fatal("cannot open " + ctx.config.mapFile + ": " + ec.message());
212 
213   ScopedTimer t1(ctx.totalMapTimer);
214 
215   // Collect symbol info that we want to print out.
216   ScopedTimer t2(ctx.symbolGatherTimer);
217   std::vector<Defined *> syms;
218   std::vector<Defined *> staticSyms;
219   getSymbols(ctx, syms, staticSyms);
220   t2.stop();
221 
222   ScopedTimer t3(ctx.symbolStringsTimer);
223   DenseMap<Defined *, std::string> symStr = getSymbolStrings(ctx, syms);
224   DenseMap<Defined *, std::string> staticSymStr =
225       getSymbolStrings(ctx, staticSyms);
226   t3.stop();
227 
228   ScopedTimer t4(ctx.writeTimer);
229   SmallString<128> AppName = sys::path::filename(ctx.config.outputFile);
230   sys::path::replace_extension(AppName, "");
231 
232   // Print out the file header
233   os << " " << AppName << "\n";
234   os << "\n";
235 
236   os << " Timestamp is " << format_hex_no_prefix(ctx.config.timestamp, 8)
237      << " (";
238   if (ctx.config.repro) {
239     os << "Repro mode";
240   } else {
241     writeFormattedTimestamp(os, ctx.config.timestamp);
242   }
243   os << ")\n";
244 
245   os << "\n";
246   os << " Preferred load address is "
247      << format_hex_no_prefix(ctx.config.imageBase, 16) << "\n";
248   os << "\n";
249 
250   // Print out section table.
251   os << " Start         Length     Name                   Class\n";
252 
253   for (OutputSection *sec : ctx.outputSections) {
254     // Merge display of chunks with same sectionName
255     std::vector<std::pair<SectionChunk *, SectionChunk *>> ChunkRanges;
256     for (Chunk *c : sec->chunks) {
257       auto *sc = dyn_cast<SectionChunk>(c);
258       if (!sc)
259         continue;
260 
261       if (ChunkRanges.empty() ||
262           c->getSectionName() != ChunkRanges.back().first->getSectionName()) {
263         ChunkRanges.emplace_back(sc, sc);
264       } else {
265         ChunkRanges.back().second = sc;
266       }
267     }
268 
269     const bool isCodeSection =
270         (sec->header.Characteristics & COFF::IMAGE_SCN_CNT_CODE) &&
271         (sec->header.Characteristics & COFF::IMAGE_SCN_MEM_READ) &&
272         (sec->header.Characteristics & COFF::IMAGE_SCN_MEM_EXECUTE);
273     StringRef SectionClass = (isCodeSection ? "CODE" : "DATA");
274 
275     for (auto &cr : ChunkRanges) {
276       size_t size =
277           cr.second->getRVA() + cr.second->getSize() - cr.first->getRVA();
278 
279       auto address = cr.first->getRVA() - sec->header.VirtualAddress;
280       writeHeader(os, sec->sectionIndex, address);
281       os << " " << format_hex_no_prefix(size, 8) << "H";
282       os << " " << left_justify(cr.first->getSectionName(), 23);
283       os << " " << SectionClass;
284       os << '\n';
285     }
286   }
287 
288   // Print out the symbols table (without static symbols)
289   os << "\n";
290   os << "  Address         Publics by Value              Rva+Base"
291         "               Lib:Object\n";
292   os << "\n";
293   for (Defined *sym : syms)
294     os << symStr[sym] << '\n';
295 
296   // Print out the entry point.
297   os << "\n";
298 
299   uint16_t entrySecIndex = 0;
300   uint64_t entryAddress = 0;
301 
302   if (!ctx.config.noEntry) {
303     Defined *entry = dyn_cast_or_null<Defined>(ctx.config.entry);
304     if (entry) {
305       Chunk *chunk = entry->getChunk();
306       entrySecIndex = chunk->getOutputSectionIdx();
307       entryAddress =
308           entry->getRVA() - ctx.getOutputSection(chunk)->header.VirtualAddress;
309     }
310   }
311   os << " entry point at         ";
312   os << format("%04x:%08llx", entrySecIndex, entryAddress);
313   os << "\n";
314 
315   // Print out the static symbols
316   os << "\n";
317   os << " Static symbols\n";
318   os << "\n";
319   for (Defined *sym : staticSyms)
320     os << staticSymStr[sym] << '\n';
321 
322   // Print out the exported functions
323   if (ctx.config.mapInfo) {
324     os << "\n";
325     os << " Exports\n";
326     os << "\n";
327     os << "  ordinal    name\n\n";
328     for (Export &e : ctx.config.exports) {
329       os << format("  %7d", e.ordinal) << "    " << e.name << "\n";
330       if (!e.extName.empty() && e.extName != e.name)
331         os << "               exported name: " << e.extName << "\n";
332     }
333   }
334 
335   t4.stop();
336   t1.stop();
337 }
338