1 //===- Archive.cpp - ar File Format implementation --------------*- 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 // This file defines the ArchiveObjectFile class.
11 //
12 //===----------------------------------------------------------------------===//
13 
14 #include "llvm/Object/Archive.h"
15 #include "llvm/ADT/APInt.h"
16 #include "llvm/Support/Endian.h"
17 #include "llvm/Support/MemoryBuffer.h"
18 
19 using namespace llvm;
20 using namespace object;
21 
22 static const char *Magic = "!<arch>\n";
23 
24 static bool isInternalMember(const ArchiveMemberHeader &amh) {
25   static const char *const internals[] = {
26     "/",
27     "//",
28     "#_LLVM_SYM_TAB_#"
29   };
30 
31   StringRef name = amh.getName();
32   for (std::size_t i = 0; i < sizeof(internals) / sizeof(*internals); ++i) {
33     if (name == internals[i])
34       return true;
35   }
36   return false;
37 }
38 
39 void Archive::anchor() { }
40 
41 error_code Archive::Child::getName(StringRef &Result) const {
42   StringRef name = ToHeader(Data.data())->getName();
43   // Check if it's a special name.
44   if (name[0] == '/') {
45     if (name.size() == 1) { // Linker member.
46       Result = name;
47       return object_error::success;
48     }
49     if (name.size() == 2 && name[1] == '/') { // String table.
50       Result = name;
51       return object_error::success;
52     }
53     // It's a long name.
54     // Get the offset.
55     std::size_t offset;
56     if (name.substr(1).rtrim(" ").getAsInteger(10, offset))
57       llvm_unreachable("Long name offset is not an integer");
58     const char *addr = Parent->StringTable->Data.begin()
59                        + sizeof(ArchiveMemberHeader)
60                        + offset;
61     // Verify it.
62     if (Parent->StringTable == Parent->end_children()
63         || addr < (Parent->StringTable->Data.begin()
64                    + sizeof(ArchiveMemberHeader))
65         || addr > (Parent->StringTable->Data.begin()
66                    + sizeof(ArchiveMemberHeader)
67                    + Parent->StringTable->getSize()))
68       return object_error::parse_failed;
69 
70     // GNU long file names end with a /.
71     if (Parent->kind() == K_GNU) {
72       StringRef::size_type End = StringRef(addr).find('/');
73       Result = StringRef(addr, End);
74     } else {
75       Result = addr;
76     }
77     return object_error::success;
78   } else if (name.startswith("#1/")) {
79     uint64_t name_size;
80     if (name.substr(3).rtrim(" ").getAsInteger(10, name_size))
81       llvm_unreachable("Long name length is not an ingeter");
82     Result = Data.substr(sizeof(ArchiveMemberHeader), name_size);
83     return object_error::success;
84   }
85   // It's a simple name.
86   if (name[name.size() - 1] == '/')
87     Result = name.substr(0, name.size() - 1);
88   else
89     Result = name;
90   return object_error::success;
91 }
92 
93 error_code Archive::Child::getAsBinary(OwningPtr<Binary> &Result) const {
94   OwningPtr<Binary> ret;
95   OwningPtr<MemoryBuffer> Buff;
96   if (error_code ec = getMemoryBuffer(Buff))
97     return ec;
98   if (error_code ec = createBinary(Buff.take(), ret))
99     return ec;
100   Result.swap(ret);
101   return object_error::success;
102 }
103 
104 Archive::Archive(MemoryBuffer *source, error_code &ec)
105   : Binary(Binary::ID_Archive, source) {
106   // Check for sufficient magic.
107   if (!source || source->getBufferSize()
108                  < (8 + sizeof(ArchiveMemberHeader) + 2) // Smallest archive.
109               || StringRef(source->getBufferStart(), 8) != Magic) {
110     ec = object_error::invalid_file_type;
111     return;
112   }
113 
114   // Get the special members.
115   child_iterator i = begin_children(false);
116   child_iterator e = end_children();
117 
118   StringRef name;
119   if ((ec = i->getName(name)))
120     return;
121 
122   // Below is the pattern that is used to figure out the archive format
123   // GNU archive format
124   //  First member : / (points to the symbol table )
125   //  Second member : // (may exist, if it exists, points to the string table)
126   //  Note : The string table is used if the filename exceeds 15 characters
127   // BSD archive format
128   //  First member : __.SYMDEF (points to the symbol table)
129   //  There is no string table, if the filename exceeds 15 characters or has a
130   //  embedded space, the filename has #1/<size>, The size represents the size
131   //  of the filename that needs to be read after the archive header
132   // COFF archive format
133   //  First member : /
134   //  Second member : / (provides a directory of symbols)
135   //  Third member : // contains the string table, this is present even if the
136   //                    string table is empty
137   if (name == "/") {
138     SymbolTable = i;
139     StringTable = e;
140     if (i != e) ++i;
141     if (i == e) {
142       ec = object_error::parse_failed;
143       return;
144     }
145     if ((ec = i->getName(name)))
146       return;
147     if (name[0] != '/') {
148       Format = K_GNU;
149     } else if ((name.size() > 1) && (name == "//")) {
150       Format = K_GNU;
151       StringTable = i;
152       ++i;
153     } else  {
154       Format = K_COFF;
155       if (i != e) {
156         SymbolTable = i;
157         ++i;
158       }
159       if (i != e) {
160         StringTable = i;
161       }
162     }
163   } else if (name == "__.SYMDEF") {
164     Format = K_BSD;
165     SymbolTable = i;
166     StringTable = e;
167   }
168   ec = object_error::success;
169 }
170 
171 Archive::child_iterator Archive::begin_children(bool skip_internal) const {
172   const char *Loc = Data->getBufferStart() + strlen(Magic);
173   size_t Size = sizeof(ArchiveMemberHeader) +
174     ToHeader(Loc)->getSize();
175   Child c(this, StringRef(Loc, Size));
176   // Skip internals at the beginning of an archive.
177   if (skip_internal && isInternalMember(*ToHeader(Loc)))
178     return c.getNext();
179   return c;
180 }
181 
182 Archive::child_iterator Archive::end_children() const {
183   return Child(this, StringRef(0, 0));
184 }
185 
186 error_code Archive::Symbol::getName(StringRef &Result) const {
187   Result = StringRef(Parent->SymbolTable->getBuffer().begin() + StringIndex);
188   return object_error::success;
189 }
190 
191 error_code Archive::Symbol::getMember(child_iterator &Result) const {
192   const char *Buf = Parent->SymbolTable->getBuffer().begin();
193   const char *Offsets = Buf + 4;
194   uint32_t Offset = 0;
195   if (Parent->kind() == K_GNU) {
196     Offset = *(reinterpret_cast<const support::ubig32_t*>(Offsets)
197                + SymbolIndex);
198   } else if (Parent->kind() == K_BSD) {
199     llvm_unreachable("BSD format is not supported");
200   } else {
201     uint32_t MemberCount = *reinterpret_cast<const support::ulittle32_t*>(Buf);
202 
203     // Skip offsets.
204     Buf += sizeof(support::ulittle32_t)
205            + (MemberCount * sizeof(support::ulittle32_t));
206 
207     uint32_t SymbolCount = *reinterpret_cast<const support::ulittle32_t*>(Buf);
208 
209     if (SymbolIndex >= SymbolCount)
210       return object_error::parse_failed;
211 
212     // Skip SymbolCount to get to the indices table.
213     const char *Indices = Buf + sizeof(support::ulittle32_t);
214 
215     // Get the index of the offset in the file member offset table for this
216     // symbol.
217     uint16_t OffsetIndex =
218       *(reinterpret_cast<const support::ulittle16_t*>(Indices)
219         + SymbolIndex);
220     // Subtract 1 since OffsetIndex is 1 based.
221     --OffsetIndex;
222 
223     if (OffsetIndex >= MemberCount)
224       return object_error::parse_failed;
225 
226     Offset = *(reinterpret_cast<const support::ulittle32_t*>(Offsets)
227                + OffsetIndex);
228   }
229 
230   const char *Loc = Parent->getData().begin() + Offset;
231   size_t Size = sizeof(ArchiveMemberHeader) +
232     ToHeader(Loc)->getSize();
233   Result = Child(Parent, StringRef(Loc, Size));
234 
235   return object_error::success;
236 }
237 
238 Archive::Symbol Archive::Symbol::getNext() const {
239   Symbol t(*this);
240   // Go to one past next null.
241   t.StringIndex =
242       Parent->SymbolTable->getBuffer().find('\0', t.StringIndex) + 1;
243   ++t.SymbolIndex;
244   return t;
245 }
246 
247 Archive::symbol_iterator Archive::begin_symbols() const {
248   const char *buf = SymbolTable->getBuffer().begin();
249   if (kind() == K_GNU) {
250     uint32_t symbol_count = 0;
251     symbol_count = *reinterpret_cast<const support::ubig32_t*>(buf);
252     buf += sizeof(uint32_t) + (symbol_count * (sizeof(uint32_t)));
253   } else if (kind() == K_BSD) {
254     llvm_unreachable("BSD archive format is not supported");
255   } else {
256     uint32_t member_count = 0;
257     uint32_t symbol_count = 0;
258     member_count = *reinterpret_cast<const support::ulittle32_t*>(buf);
259     buf += 4 + (member_count * 4); // Skip offsets.
260     symbol_count = *reinterpret_cast<const support::ulittle32_t*>(buf);
261     buf += 4 + (symbol_count * 2); // Skip indices.
262   }
263   uint32_t string_start_offset = buf - SymbolTable->getBuffer().begin();
264   return symbol_iterator(Symbol(this, 0, string_start_offset));
265 }
266 
267 Archive::symbol_iterator Archive::end_symbols() const {
268   const char *buf = SymbolTable->getBuffer().begin();
269   uint32_t symbol_count = 0;
270   if (kind() == K_GNU) {
271     symbol_count = *reinterpret_cast<const support::ubig32_t*>(buf);
272     buf += sizeof(uint32_t) + (symbol_count * (sizeof(uint32_t)));
273   } else if (kind() == K_BSD) {
274     llvm_unreachable("BSD archive format is not supported");
275   } else {
276     uint32_t member_count = 0;
277     member_count = *reinterpret_cast<const support::ulittle32_t*>(buf);
278     buf += 4 + (member_count * 4); // Skip offsets.
279     symbol_count = *reinterpret_cast<const support::ulittle32_t*>(buf);
280   }
281   return symbol_iterator(
282     Symbol(this, symbol_count, 0));
283 }
284 
285 Archive::child_iterator Archive::findSym(StringRef name) const {
286   Archive::symbol_iterator bs = begin_symbols();
287   Archive::symbol_iterator es = end_symbols();
288   Archive::child_iterator result;
289 
290   StringRef symname;
291   for (; bs != es; ++bs) {
292     if (bs->getName(symname))
293         return end_children();
294     if (symname == name) {
295       if (bs->getMember(result))
296         return end_children();
297       return result;
298     }
299   }
300   return end_children();
301 }
302