blob: 27676cf454a51c025cdee4c4f7c1ff2e3a3e41e2 [file] [log] [blame]
Michael J. Spencera51d7d92011-09-27 19:36:55 +00001//===- 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"
Michael J. Spencerc8a55a62011-11-02 19:33:12 +000016#include "llvm/Support/Endian.h"
Michael J. Spencera51d7d92011-09-27 19:36:55 +000017#include "llvm/Support/MemoryBuffer.h"
18
19using namespace llvm;
20using namespace object;
21
Benjamin Kramere1a44272012-02-22 13:42:11 +000022static const char *Magic = "!<arch>\n";
Michael J. Spencera51d7d92011-09-27 19:36:55 +000023
Michael J. Spenceraaf98ea2011-10-25 22:30:42 +000024static bool isInternalMember(const ArchiveMemberHeader &amh) {
Craig Toppere3298102012-05-24 06:35:32 +000025 static const char *const internals[] = {
Michael J. Spenceraaf98ea2011-10-25 22:30:42 +000026 "/",
Rafael Espindola250bfb12013-06-14 23:25:53 +000027 "//"
Craig Toppere3298102012-05-24 06:35:32 +000028 };
Michael J. Spenceraaf98ea2011-10-25 22:30:42 +000029
30 StringRef name = amh.getName();
31 for (std::size_t i = 0; i < sizeof(internals) / sizeof(*internals); ++i) {
32 if (name == internals[i])
33 return true;
34 }
35 return false;
36}
37
David Blaikie2d24e2a2011-12-20 02:50:00 +000038void Archive::anchor() { }
39
Michael J. Spencera51d7d92011-09-27 19:36:55 +000040error_code Archive::Child::getName(StringRef &Result) const {
Rafael Espindola4a0bf542013-07-05 03:35:15 +000041 StringRef name = getRawName();
Michael J. Spencera51d7d92011-09-27 19:36:55 +000042 // Check if it's a special name.
43 if (name[0] == '/') {
44 if (name.size() == 1) { // Linker member.
45 Result = name;
46 return object_error::success;
47 }
48 if (name.size() == 2 && name[1] == '/') { // String table.
49 Result = name;
50 return object_error::success;
51 }
52 // It's a long name.
53 // Get the offset.
Michael J. Spencerbf82b072013-01-10 00:07:38 +000054 std::size_t offset;
Michael J. Spencer7932c412013-01-10 01:05:34 +000055 if (name.substr(1).rtrim(" ").getAsInteger(10, offset))
56 llvm_unreachable("Long name offset is not an integer");
Michael J. Spencera51d7d92011-09-27 19:36:55 +000057 const char *addr = Parent->StringTable->Data.begin()
58 + sizeof(ArchiveMemberHeader)
Michael J. Spencere92800d2013-01-09 22:58:43 +000059 + offset;
Michael J. Spencera51d7d92011-09-27 19:36:55 +000060 // Verify it.
61 if (Parent->StringTable == Parent->end_children()
62 || addr < (Parent->StringTable->Data.begin()
63 + sizeof(ArchiveMemberHeader))
64 || addr > (Parent->StringTable->Data.begin()
65 + sizeof(ArchiveMemberHeader)
66 + Parent->StringTable->getSize()))
67 return object_error::parse_failed;
Shankar Easwaran206252c2012-11-13 18:38:42 +000068
69 // GNU long file names end with a /.
70 if (Parent->kind() == K_GNU) {
71 StringRef::size_type End = StringRef(addr).find('/');
72 Result = StringRef(addr, End);
73 } else {
74 Result = addr;
75 }
Michael J. Spencera51d7d92011-09-27 19:36:55 +000076 return object_error::success;
Michael J. Spenceraaf98ea2011-10-25 22:30:42 +000077 } else if (name.startswith("#1/")) {
Michael J. Spencere92800d2013-01-09 22:58:43 +000078 uint64_t name_size;
Michael J. Spencer7932c412013-01-10 01:05:34 +000079 if (name.substr(3).rtrim(" ").getAsInteger(10, name_size))
80 llvm_unreachable("Long name length is not an ingeter");
81 Result = Data.substr(sizeof(ArchiveMemberHeader), name_size);
Michael J. Spenceraaf98ea2011-10-25 22:30:42 +000082 return object_error::success;
Michael J. Spencera51d7d92011-09-27 19:36:55 +000083 }
84 // It's a simple name.
85 if (name[name.size() - 1] == '/')
86 Result = name.substr(0, name.size() - 1);
87 else
88 Result = name;
89 return object_error::success;
90}
91
Michael J. Spencera51d7d92011-09-27 19:36:55 +000092error_code Archive::Child::getAsBinary(OwningPtr<Binary> &Result) const {
93 OwningPtr<Binary> ret;
Michael J. Spencer0f76e642013-02-03 10:48:50 +000094 OwningPtr<MemoryBuffer> Buff;
95 if (error_code ec = getMemoryBuffer(Buff))
96 return ec;
97 if (error_code ec = createBinary(Buff.take(), ret))
Michael J. Spencera51d7d92011-09-27 19:36:55 +000098 return ec;
99 Result.swap(ret);
100 return object_error::success;
101}
102
103Archive::Archive(MemoryBuffer *source, error_code &ec)
David Meyer6f9489a2012-03-09 20:41:57 +0000104 : Binary(Binary::ID_Archive, source) {
Michael J. Spencera51d7d92011-09-27 19:36:55 +0000105 // Check for sufficient magic.
106 if (!source || source->getBufferSize()
Rafael Espindola6d88f9b2013-07-04 19:40:23 +0000107 < (8 + sizeof(ArchiveMemberHeader)) // Smallest archive.
Michael J. Spencera51d7d92011-09-27 19:36:55 +0000108 || StringRef(source->getBufferStart(), 8) != Magic) {
109 ec = object_error::invalid_file_type;
110 return;
111 }
112
Michael J. Spencerc8a55a62011-11-02 19:33:12 +0000113 // Get the special members.
114 child_iterator i = begin_children(false);
Michael J. Spencera51d7d92011-09-27 19:36:55 +0000115 child_iterator e = end_children();
Michael J. Spencera51d7d92011-09-27 19:36:55 +0000116
Rafael Espindola09a7f602013-07-03 15:57:14 +0000117 if (i == e) {
118 ec = object_error::parse_failed;
119 return;
120 }
121
Rafael Espindola4a0bf542013-07-05 03:35:15 +0000122 StringRef Name = i->getRawName();
Michael J. Spencera51d7d92011-09-27 19:36:55 +0000123
Shankar Easwaran206252c2012-11-13 18:38:42 +0000124 // Below is the pattern that is used to figure out the archive format
125 // GNU archive format
Rafael Espindola09a7f602013-07-03 15:57:14 +0000126 // First member : / (may exist, if it exists, points to the symbol table )
Shankar Easwaran206252c2012-11-13 18:38:42 +0000127 // Second member : // (may exist, if it exists, points to the string table)
128 // Note : The string table is used if the filename exceeds 15 characters
129 // BSD archive format
130 // First member : __.SYMDEF (points to the symbol table)
131 // There is no string table, if the filename exceeds 15 characters or has a
132 // embedded space, the filename has #1/<size>, The size represents the size
133 // of the filename that needs to be read after the archive header
134 // COFF archive format
135 // First member : /
136 // Second member : / (provides a directory of symbols)
Rui Ueyama891c0cd2013-06-03 00:27:03 +0000137 // Third member : // (may exist, if it exists, contains the string table)
138 // Note: Microsoft PE/COFF Spec 8.3 says that the third member is present
139 // even if the string table is empty. However, lib.exe does not in fact
140 // seem to create the third member if there's no member whose filename
141 // exceeds 15 characters. So the third member is optional.
Rafael Espindola09a7f602013-07-03 15:57:14 +0000142
Rafael Espindola4a0bf542013-07-05 03:35:15 +0000143 if (Name == "__.SYMDEF") {
Rafael Espindola09a7f602013-07-03 15:57:14 +0000144 Format = K_BSD;
145 SymbolTable = i;
146 ec = object_error::success;
147 return;
148 }
149
Rafael Espindola4a0bf542013-07-05 03:35:15 +0000150 if (Name == "/") {
Shankar Easwaran206252c2012-11-13 18:38:42 +0000151 SymbolTable = i;
Rafael Espindola09a7f602013-07-03 15:57:14 +0000152
153 ++i;
Michael J. Spencerbf82b072013-01-10 00:07:38 +0000154 if (i == e) {
155 ec = object_error::parse_failed;
156 return;
157 }
Rafael Espindola4a0bf542013-07-05 03:35:15 +0000158 Name = i->getRawName();
Rafael Espindola09a7f602013-07-03 15:57:14 +0000159 }
160
Rafael Espindola4a0bf542013-07-05 03:35:15 +0000161 if (Name == "//") {
Rafael Espindola09a7f602013-07-03 15:57:14 +0000162 Format = K_GNU;
163 StringTable = i;
164 ec = object_error::success;
165 return;
166 }
167
Rafael Espindola4a0bf542013-07-05 03:35:15 +0000168 if (Name[0] != '/') {
Rafael Espindola09a7f602013-07-03 15:57:14 +0000169 Format = K_GNU;
170 ec = object_error::success;
171 return;
172 }
173
Rafael Espindola4a0bf542013-07-05 03:35:15 +0000174 if (Name != "/") {
Rafael Espindola09a7f602013-07-03 15:57:14 +0000175 ec = object_error::parse_failed;
176 return;
177 }
178
179 Format = K_COFF;
180 SymbolTable = i;
181
182 ++i;
183 if (i == e) {
184 ec = object_error::success;
185 return;
186 }
187
Rafael Espindola4a0bf542013-07-05 03:35:15 +0000188 Name = i->getRawName();
Rafael Espindola09a7f602013-07-03 15:57:14 +0000189
Rafael Espindola4a0bf542013-07-05 03:35:15 +0000190 if (Name == "//")
Rafael Espindola09a7f602013-07-03 15:57:14 +0000191 StringTable = i;
192
Michael J. Spencera51d7d92011-09-27 19:36:55 +0000193 ec = object_error::success;
194}
195
Michael J. Spenceraaf98ea2011-10-25 22:30:42 +0000196Archive::child_iterator Archive::begin_children(bool skip_internal) const {
Benjamin Kramere1a44272012-02-22 13:42:11 +0000197 const char *Loc = Data->getBufferStart() + strlen(Magic);
Michael J. Spencera51d7d92011-09-27 19:36:55 +0000198 size_t Size = sizeof(ArchiveMemberHeader) +
199 ToHeader(Loc)->getSize();
Michael J. Spenceraaf98ea2011-10-25 22:30:42 +0000200 Child c(this, StringRef(Loc, Size));
201 // Skip internals at the beginning of an archive.
202 if (skip_internal && isInternalMember(*ToHeader(Loc)))
203 return c.getNext();
204 return c;
Michael J. Spencera51d7d92011-09-27 19:36:55 +0000205}
206
Michael J. Spencer58618932011-10-08 00:17:45 +0000207Archive::child_iterator Archive::end_children() const {
Michael J. Spencera51d7d92011-09-27 19:36:55 +0000208 return Child(this, StringRef(0, 0));
209}
Michael J. Spencerc8a55a62011-11-02 19:33:12 +0000210
211error_code Archive::Symbol::getName(StringRef &Result) const {
Michael J. Spencer0f76e642013-02-03 10:48:50 +0000212 Result = StringRef(Parent->SymbolTable->getBuffer().begin() + StringIndex);
Michael J. Spencerc8a55a62011-11-02 19:33:12 +0000213 return object_error::success;
214}
215
216error_code Archive::Symbol::getMember(child_iterator &Result) const {
Michael J. Spencer0f76e642013-02-03 10:48:50 +0000217 const char *Buf = Parent->SymbolTable->getBuffer().begin();
Michael J. Spencer768a7072012-11-14 00:04:13 +0000218 const char *Offsets = Buf + 4;
219 uint32_t Offset = 0;
Shankar Easwaran206252c2012-11-13 18:38:42 +0000220 if (Parent->kind() == K_GNU) {
Michael J. Spencer768a7072012-11-14 00:04:13 +0000221 Offset = *(reinterpret_cast<const support::ubig32_t*>(Offsets)
222 + SymbolIndex);
Shankar Easwaran206252c2012-11-13 18:38:42 +0000223 } else if (Parent->kind() == K_BSD) {
Matt Beaumont-Gay7af4b9b2012-11-14 17:58:11 +0000224 llvm_unreachable("BSD format is not supported");
Shankar Easwaran206252c2012-11-13 18:38:42 +0000225 } else {
Michael J. Spencer768a7072012-11-14 00:04:13 +0000226 uint32_t MemberCount = *reinterpret_cast<const support::ulittle32_t*>(Buf);
227
228 // Skip offsets.
229 Buf += sizeof(support::ulittle32_t)
230 + (MemberCount * sizeof(support::ulittle32_t));
231
232 uint32_t SymbolCount = *reinterpret_cast<const support::ulittle32_t*>(Buf);
233
234 if (SymbolIndex >= SymbolCount)
235 return object_error::parse_failed;
236
Matt Beaumont-Gayf1c2a6b2012-11-14 00:21:27 +0000237 // Skip SymbolCount to get to the indices table.
238 const char *Indices = Buf + sizeof(support::ulittle32_t);
Michael J. Spencer768a7072012-11-14 00:04:13 +0000239
240 // Get the index of the offset in the file member offset table for this
241 // symbol.
242 uint16_t OffsetIndex =
Matt Beaumont-Gayf1c2a6b2012-11-14 00:21:27 +0000243 *(reinterpret_cast<const support::ulittle16_t*>(Indices)
Shankar Easwaran206252c2012-11-13 18:38:42 +0000244 + SymbolIndex);
Michael J. Spencer768a7072012-11-14 00:04:13 +0000245 // Subtract 1 since OffsetIndex is 1 based.
246 --OffsetIndex;
247
248 if (OffsetIndex >= MemberCount)
249 return object_error::parse_failed;
250
251 Offset = *(reinterpret_cast<const support::ulittle32_t*>(Offsets)
252 + OffsetIndex);
Shankar Easwaran206252c2012-11-13 18:38:42 +0000253 }
Michael J. Spencerc8a55a62011-11-02 19:33:12 +0000254
Michael J. Spencer768a7072012-11-14 00:04:13 +0000255 const char *Loc = Parent->getData().begin() + Offset;
Michael J. Spencerc8a55a62011-11-02 19:33:12 +0000256 size_t Size = sizeof(ArchiveMemberHeader) +
257 ToHeader(Loc)->getSize();
258 Result = Child(Parent, StringRef(Loc, Size));
259
260 return object_error::success;
261}
262
263Archive::Symbol Archive::Symbol::getNext() const {
264 Symbol t(*this);
Benjamin Kramerefd2d5e2011-11-04 13:52:17 +0000265 // Go to one past next null.
266 t.StringIndex =
Michael J. Spencer0f76e642013-02-03 10:48:50 +0000267 Parent->SymbolTable->getBuffer().find('\0', t.StringIndex) + 1;
Michael J. Spencerc8a55a62011-11-02 19:33:12 +0000268 ++t.SymbolIndex;
269 return t;
270}
271
272Archive::symbol_iterator Archive::begin_symbols() const {
Michael J. Spencer0f76e642013-02-03 10:48:50 +0000273 const char *buf = SymbolTable->getBuffer().begin();
Shankar Easwaran206252c2012-11-13 18:38:42 +0000274 if (kind() == K_GNU) {
275 uint32_t symbol_count = 0;
276 symbol_count = *reinterpret_cast<const support::ubig32_t*>(buf);
277 buf += sizeof(uint32_t) + (symbol_count * (sizeof(uint32_t)));
278 } else if (kind() == K_BSD) {
Matt Beaumont-Gay7af4b9b2012-11-14 17:58:11 +0000279 llvm_unreachable("BSD archive format is not supported");
Shankar Easwaran206252c2012-11-13 18:38:42 +0000280 } else {
281 uint32_t member_count = 0;
282 uint32_t symbol_count = 0;
283 member_count = *reinterpret_cast<const support::ulittle32_t*>(buf);
284 buf += 4 + (member_count * 4); // Skip offsets.
285 symbol_count = *reinterpret_cast<const support::ulittle32_t*>(buf);
286 buf += 4 + (symbol_count * 2); // Skip indices.
287 }
Michael J. Spencer0f76e642013-02-03 10:48:50 +0000288 uint32_t string_start_offset = buf - SymbolTable->getBuffer().begin();
Michael J. Spencerc8a55a62011-11-02 19:33:12 +0000289 return symbol_iterator(Symbol(this, 0, string_start_offset));
290}
291
292Archive::symbol_iterator Archive::end_symbols() const {
Michael J. Spencer0f76e642013-02-03 10:48:50 +0000293 const char *buf = SymbolTable->getBuffer().begin();
Shankar Easwaran206252c2012-11-13 18:38:42 +0000294 uint32_t symbol_count = 0;
295 if (kind() == K_GNU) {
296 symbol_count = *reinterpret_cast<const support::ubig32_t*>(buf);
297 buf += sizeof(uint32_t) + (symbol_count * (sizeof(uint32_t)));
298 } else if (kind() == K_BSD) {
Matt Beaumont-Gay7af4b9b2012-11-14 17:58:11 +0000299 llvm_unreachable("BSD archive format is not supported");
Shankar Easwaran206252c2012-11-13 18:38:42 +0000300 } else {
301 uint32_t member_count = 0;
302 member_count = *reinterpret_cast<const support::ulittle32_t*>(buf);
303 buf += 4 + (member_count * 4); // Skip offsets.
304 symbol_count = *reinterpret_cast<const support::ulittle32_t*>(buf);
305 }
Michael J. Spencerc8a55a62011-11-02 19:33:12 +0000306 return symbol_iterator(
307 Symbol(this, symbol_count, 0));
308}
Shankar Easwaran206252c2012-11-13 18:38:42 +0000309
310Archive::child_iterator Archive::findSym(StringRef name) const {
311 Archive::symbol_iterator bs = begin_symbols();
312 Archive::symbol_iterator es = end_symbols();
313 Archive::child_iterator result;
314
315 StringRef symname;
316 for (; bs != es; ++bs) {
317 if (bs->getName(symname))
318 return end_children();
319 if (symname == name) {
320 if (bs->getMember(result))
321 return end_children();
322 return result;
323 }
324 }
325 return end_children();
326}