blob: 0d12be8f049514b7bb60c590b1087c6642fe1f6c [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"
Chandler Carruthd04a8d42012-12-03 16:50:05 +000017#include "llvm/Support/ErrorHandling.h"
Michael J. Spencera51d7d92011-09-27 19:36:55 +000018#include "llvm/Support/MemoryBuffer.h"
19
20using namespace llvm;
21using namespace object;
22
Benjamin Kramere1a44272012-02-22 13:42:11 +000023static const char *Magic = "!<arch>\n";
Michael J. Spencera51d7d92011-09-27 19:36:55 +000024
Michael J. Spencerfe944e82011-10-25 22:31:11 +000025namespace {
Michael J. Spencera51d7d92011-09-27 19:36:55 +000026struct ArchiveMemberHeader {
27 char Name[16];
28 char LastModified[12];
29 char UID[6];
30 char GID[6];
31 char AccessMode[8];
Dmitri Gribenko77592fe2012-06-09 00:01:45 +000032 char Size[10]; ///< Size of data, not including header or padding.
Michael J. Spencera51d7d92011-09-27 19:36:55 +000033 char Terminator[2];
34
35 ///! Get the name without looking up long names.
36 StringRef getName() const {
Michael J. Spenceraaf98ea2011-10-25 22:30:42 +000037 char EndCond;
38 if (Name[0] == '/' || Name[0] == '#')
39 EndCond = ' ';
40 else
41 EndCond = '/';
Michael J. Spencera51d7d92011-09-27 19:36:55 +000042 StringRef::size_type end = StringRef(Name, sizeof(Name)).find(EndCond);
43 if (end == StringRef::npos)
44 end = sizeof(Name);
45 assert(end <= sizeof(Name) && end > 0);
46 // Don't include the EndCond if there is one.
47 return StringRef(Name, end);
48 }
49
50 uint64_t getSize() const {
Michael J. Spencere92800d2013-01-09 22:58:43 +000051 uint64_t ret;
Michael J. Spencerbf82b072013-01-10 00:07:38 +000052 StringRef(Size, sizeof(Size)).rtrim(" ").getAsInteger(10, ret);
Michael J. Spencere92800d2013-01-09 22:58:43 +000053 return ret;
Michael J. Spencera51d7d92011-09-27 19:36:55 +000054 }
55};
Michael J. Spencerfe944e82011-10-25 22:31:11 +000056}
Michael J. Spencera51d7d92011-09-27 19:36:55 +000057
Michael J. Spencerfe944e82011-10-25 22:31:11 +000058static const ArchiveMemberHeader *ToHeader(const char *base) {
Michael J. Spencera51d7d92011-09-27 19:36:55 +000059 return reinterpret_cast<const ArchiveMemberHeader *>(base);
60}
Michael J. Spencerfe944e82011-10-25 22:31:11 +000061
Michael J. Spencera51d7d92011-09-27 19:36:55 +000062
Michael J. Spenceraaf98ea2011-10-25 22:30:42 +000063static bool isInternalMember(const ArchiveMemberHeader &amh) {
Craig Toppere3298102012-05-24 06:35:32 +000064 static const char *const internals[] = {
Michael J. Spenceraaf98ea2011-10-25 22:30:42 +000065 "/",
66 "//",
67 "#_LLVM_SYM_TAB_#"
Craig Toppere3298102012-05-24 06:35:32 +000068 };
Michael J. Spenceraaf98ea2011-10-25 22:30:42 +000069
70 StringRef name = amh.getName();
71 for (std::size_t i = 0; i < sizeof(internals) / sizeof(*internals); ++i) {
72 if (name == internals[i])
73 return true;
74 }
75 return false;
76}
77
David Blaikie2d24e2a2011-12-20 02:50:00 +000078void Archive::anchor() { }
79
Michael J. Spencera51d7d92011-09-27 19:36:55 +000080Archive::Child Archive::Child::getNext() const {
81 size_t SpaceToSkip = sizeof(ArchiveMemberHeader) +
82 ToHeader(Data.data())->getSize();
83 // If it's odd, add 1 to make it even.
84 if (SpaceToSkip & 1)
85 ++SpaceToSkip;
86
87 const char *NextLoc = Data.data() + SpaceToSkip;
88
89 // Check to see if this is past the end of the archive.
90 if (NextLoc >= Parent->Data->getBufferEnd())
91 return Child(Parent, StringRef(0, 0));
92
93 size_t NextSize = sizeof(ArchiveMemberHeader) +
94 ToHeader(NextLoc)->getSize();
95
96 return Child(Parent, StringRef(NextLoc, NextSize));
97}
98
99error_code Archive::Child::getName(StringRef &Result) const {
100 StringRef name = ToHeader(Data.data())->getName();
101 // Check if it's a special name.
102 if (name[0] == '/') {
103 if (name.size() == 1) { // Linker member.
104 Result = name;
105 return object_error::success;
106 }
107 if (name.size() == 2 && name[1] == '/') { // String table.
108 Result = name;
109 return object_error::success;
110 }
111 // It's a long name.
112 // Get the offset.
Michael J. Spencerbf82b072013-01-10 00:07:38 +0000113 std::size_t offset;
Michael J. Spencera51d7d92011-09-27 19:36:55 +0000114 name.substr(1).getAsInteger(10, offset);
115 const char *addr = Parent->StringTable->Data.begin()
116 + sizeof(ArchiveMemberHeader)
Michael J. Spencere92800d2013-01-09 22:58:43 +0000117 + offset;
Michael J. Spencera51d7d92011-09-27 19:36:55 +0000118 // Verify it.
119 if (Parent->StringTable == Parent->end_children()
120 || addr < (Parent->StringTable->Data.begin()
121 + sizeof(ArchiveMemberHeader))
122 || addr > (Parent->StringTable->Data.begin()
123 + sizeof(ArchiveMemberHeader)
124 + Parent->StringTable->getSize()))
125 return object_error::parse_failed;
Shankar Easwaran206252c2012-11-13 18:38:42 +0000126
127 // GNU long file names end with a /.
128 if (Parent->kind() == K_GNU) {
129 StringRef::size_type End = StringRef(addr).find('/');
130 Result = StringRef(addr, End);
131 } else {
132 Result = addr;
133 }
Michael J. Spencera51d7d92011-09-27 19:36:55 +0000134 return object_error::success;
Michael J. Spenceraaf98ea2011-10-25 22:30:42 +0000135 } else if (name.startswith("#1/")) {
Michael J. Spencere92800d2013-01-09 22:58:43 +0000136 uint64_t name_size;
Michael J. Spenceraaf98ea2011-10-25 22:30:42 +0000137 name.substr(3).getAsInteger(10, name_size);
Michael J. Spencere92800d2013-01-09 22:58:43 +0000138 Result = Data.substr(0, name_size);
Michael J. Spenceraaf98ea2011-10-25 22:30:42 +0000139 return object_error::success;
Michael J. Spencera51d7d92011-09-27 19:36:55 +0000140 }
141 // It's a simple name.
142 if (name[name.size() - 1] == '/')
143 Result = name.substr(0, name.size() - 1);
144 else
145 Result = name;
146 return object_error::success;
147}
148
149uint64_t Archive::Child::getSize() const {
Michael J. Spenceraaf98ea2011-10-25 22:30:42 +0000150 uint64_t size = ToHeader(Data.data())->getSize();
151 // Don't include attached name.
152 StringRef name = ToHeader(Data.data())->getName();
153 if (name.startswith("#1/")) {
Michael J. Spencere92800d2013-01-09 22:58:43 +0000154 uint64_t name_size;
Michael J. Spenceraaf98ea2011-10-25 22:30:42 +0000155 name.substr(3).getAsInteger(10, name_size);
Michael J. Spencere92800d2013-01-09 22:58:43 +0000156 size -= name_size;
Michael J. Spenceraaf98ea2011-10-25 22:30:42 +0000157 }
158 return size;
Michael J. Spencera51d7d92011-09-27 19:36:55 +0000159}
160
161MemoryBuffer *Archive::Child::getBuffer() const {
162 StringRef name;
163 if (getName(name)) return NULL;
Michael J. Spenceraaf98ea2011-10-25 22:30:42 +0000164 int size = sizeof(ArchiveMemberHeader);
165 if (name.startswith("#1/")) {
Michael J. Spencere92800d2013-01-09 22:58:43 +0000166 uint64_t name_size;
Michael J. Spenceraaf98ea2011-10-25 22:30:42 +0000167 name.substr(3).getAsInteger(10, name_size);
Michael J. Spencere92800d2013-01-09 22:58:43 +0000168 size += name_size;
Michael J. Spenceraaf98ea2011-10-25 22:30:42 +0000169 }
170 return MemoryBuffer::getMemBuffer(Data.substr(size, getSize()),
Michael J. Spencera51d7d92011-09-27 19:36:55 +0000171 name,
172 false);
173}
174
175error_code Archive::Child::getAsBinary(OwningPtr<Binary> &Result) const {
176 OwningPtr<Binary> ret;
177 if (error_code ec =
178 createBinary(getBuffer(), ret))
179 return ec;
180 Result.swap(ret);
181 return object_error::success;
182}
183
184Archive::Archive(MemoryBuffer *source, error_code &ec)
David Meyer6f9489a2012-03-09 20:41:57 +0000185 : Binary(Binary::ID_Archive, source) {
Michael J. Spencera51d7d92011-09-27 19:36:55 +0000186 // Check for sufficient magic.
187 if (!source || source->getBufferSize()
188 < (8 + sizeof(ArchiveMemberHeader) + 2) // Smallest archive.
189 || StringRef(source->getBufferStart(), 8) != Magic) {
190 ec = object_error::invalid_file_type;
191 return;
192 }
193
Michael J. Spencerc8a55a62011-11-02 19:33:12 +0000194 // Get the special members.
195 child_iterator i = begin_children(false);
Michael J. Spencera51d7d92011-09-27 19:36:55 +0000196 child_iterator e = end_children();
Michael J. Spencera51d7d92011-09-27 19:36:55 +0000197
Shankar Easwaran206252c2012-11-13 18:38:42 +0000198 StringRef name;
199 if ((ec = i->getName(name)))
200 return;
Michael J. Spencera51d7d92011-09-27 19:36:55 +0000201
Shankar Easwaran206252c2012-11-13 18:38:42 +0000202 // Below is the pattern that is used to figure out the archive format
203 // GNU archive format
204 // First member : / (points to the symbol table )
205 // Second member : // (may exist, if it exists, points to the string table)
206 // Note : The string table is used if the filename exceeds 15 characters
207 // BSD archive format
208 // First member : __.SYMDEF (points to the symbol table)
209 // There is no string table, if the filename exceeds 15 characters or has a
210 // embedded space, the filename has #1/<size>, The size represents the size
211 // of the filename that needs to be read after the archive header
212 // COFF archive format
213 // First member : /
214 // Second member : / (provides a directory of symbols)
215 // Third member : // contains the string table, this is present even if the
216 // string table is empty
217 if (name == "/") {
218 SymbolTable = i;
219 StringTable = e;
220 if (i != e) ++i;
Michael J. Spencerbf82b072013-01-10 00:07:38 +0000221 if (i == e) {
222 ec = object_error::parse_failed;
223 return;
224 }
Shankar Easwaran206252c2012-11-13 18:38:42 +0000225 if ((ec = i->getName(name)))
226 return;
227 if (name[0] != '/') {
228 Format = K_GNU;
229 } else if ((name.size() > 1) && (name == "//")) {
230 Format = K_GNU;
231 StringTable = i;
232 ++i;
233 } else {
234 Format = K_COFF;
235 if (i != e) {
236 SymbolTable = i;
237 ++i;
238 }
239 if (i != e) {
240 StringTable = i;
241 }
242 }
243 } else if (name == "__.SYMDEF") {
244 Format = K_BSD;
245 SymbolTable = i;
246 StringTable = e;
247 }
Michael J. Spencera51d7d92011-09-27 19:36:55 +0000248 ec = object_error::success;
249}
250
Michael J. Spenceraaf98ea2011-10-25 22:30:42 +0000251Archive::child_iterator Archive::begin_children(bool skip_internal) const {
Benjamin Kramere1a44272012-02-22 13:42:11 +0000252 const char *Loc = Data->getBufferStart() + strlen(Magic);
Michael J. Spencera51d7d92011-09-27 19:36:55 +0000253 size_t Size = sizeof(ArchiveMemberHeader) +
254 ToHeader(Loc)->getSize();
Michael J. Spenceraaf98ea2011-10-25 22:30:42 +0000255 Child c(this, StringRef(Loc, Size));
256 // Skip internals at the beginning of an archive.
257 if (skip_internal && isInternalMember(*ToHeader(Loc)))
258 return c.getNext();
259 return c;
Michael J. Spencera51d7d92011-09-27 19:36:55 +0000260}
261
Michael J. Spencer58618932011-10-08 00:17:45 +0000262Archive::child_iterator Archive::end_children() const {
Michael J. Spencera51d7d92011-09-27 19:36:55 +0000263 return Child(this, StringRef(0, 0));
264}
Michael J. Spencerc8a55a62011-11-02 19:33:12 +0000265
266error_code Archive::Symbol::getName(StringRef &Result) const {
267 Result =
268 StringRef(Parent->SymbolTable->getBuffer()->getBufferStart() + StringIndex);
269 return object_error::success;
270}
271
272error_code Archive::Symbol::getMember(child_iterator &Result) const {
Michael J. Spencer768a7072012-11-14 00:04:13 +0000273 const char *Buf = Parent->SymbolTable->getBuffer()->getBufferStart();
274 const char *Offsets = Buf + 4;
275 uint32_t Offset = 0;
Shankar Easwaran206252c2012-11-13 18:38:42 +0000276 if (Parent->kind() == K_GNU) {
Michael J. Spencer768a7072012-11-14 00:04:13 +0000277 Offset = *(reinterpret_cast<const support::ubig32_t*>(Offsets)
278 + SymbolIndex);
Shankar Easwaran206252c2012-11-13 18:38:42 +0000279 } else if (Parent->kind() == K_BSD) {
Matt Beaumont-Gay7af4b9b2012-11-14 17:58:11 +0000280 llvm_unreachable("BSD format is not supported");
Shankar Easwaran206252c2012-11-13 18:38:42 +0000281 } else {
Michael J. Spencer768a7072012-11-14 00:04:13 +0000282 uint32_t MemberCount = *reinterpret_cast<const support::ulittle32_t*>(Buf);
283
284 // Skip offsets.
285 Buf += sizeof(support::ulittle32_t)
286 + (MemberCount * sizeof(support::ulittle32_t));
287
288 uint32_t SymbolCount = *reinterpret_cast<const support::ulittle32_t*>(Buf);
289
290 if (SymbolIndex >= SymbolCount)
291 return object_error::parse_failed;
292
Matt Beaumont-Gayf1c2a6b2012-11-14 00:21:27 +0000293 // Skip SymbolCount to get to the indices table.
294 const char *Indices = Buf + sizeof(support::ulittle32_t);
Michael J. Spencer768a7072012-11-14 00:04:13 +0000295
296 // Get the index of the offset in the file member offset table for this
297 // symbol.
298 uint16_t OffsetIndex =
Matt Beaumont-Gayf1c2a6b2012-11-14 00:21:27 +0000299 *(reinterpret_cast<const support::ulittle16_t*>(Indices)
Shankar Easwaran206252c2012-11-13 18:38:42 +0000300 + SymbolIndex);
Michael J. Spencer768a7072012-11-14 00:04:13 +0000301 // Subtract 1 since OffsetIndex is 1 based.
302 --OffsetIndex;
303
304 if (OffsetIndex >= MemberCount)
305 return object_error::parse_failed;
306
307 Offset = *(reinterpret_cast<const support::ulittle32_t*>(Offsets)
308 + OffsetIndex);
Shankar Easwaran206252c2012-11-13 18:38:42 +0000309 }
Michael J. Spencerc8a55a62011-11-02 19:33:12 +0000310
Michael J. Spencer768a7072012-11-14 00:04:13 +0000311 const char *Loc = Parent->getData().begin() + Offset;
Michael J. Spencerc8a55a62011-11-02 19:33:12 +0000312 size_t Size = sizeof(ArchiveMemberHeader) +
313 ToHeader(Loc)->getSize();
314 Result = Child(Parent, StringRef(Loc, Size));
315
316 return object_error::success;
317}
318
319Archive::Symbol Archive::Symbol::getNext() const {
320 Symbol t(*this);
Benjamin Kramerefd2d5e2011-11-04 13:52:17 +0000321 // Go to one past next null.
322 t.StringIndex =
323 Parent->SymbolTable->getBuffer()->getBuffer().find('\0', t.StringIndex) + 1;
Michael J. Spencerc8a55a62011-11-02 19:33:12 +0000324 ++t.SymbolIndex;
325 return t;
326}
327
328Archive::symbol_iterator Archive::begin_symbols() const {
329 const char *buf = SymbolTable->getBuffer()->getBufferStart();
Shankar Easwaran206252c2012-11-13 18:38:42 +0000330 if (kind() == K_GNU) {
331 uint32_t symbol_count = 0;
332 symbol_count = *reinterpret_cast<const support::ubig32_t*>(buf);
333 buf += sizeof(uint32_t) + (symbol_count * (sizeof(uint32_t)));
334 } else if (kind() == K_BSD) {
Matt Beaumont-Gay7af4b9b2012-11-14 17:58:11 +0000335 llvm_unreachable("BSD archive format is not supported");
Shankar Easwaran206252c2012-11-13 18:38:42 +0000336 } else {
337 uint32_t member_count = 0;
338 uint32_t symbol_count = 0;
339 member_count = *reinterpret_cast<const support::ulittle32_t*>(buf);
340 buf += 4 + (member_count * 4); // Skip offsets.
341 symbol_count = *reinterpret_cast<const support::ulittle32_t*>(buf);
342 buf += 4 + (symbol_count * 2); // Skip indices.
343 }
Michael J. Spencerc8a55a62011-11-02 19:33:12 +0000344 uint32_t string_start_offset =
345 buf - SymbolTable->getBuffer()->getBufferStart();
346 return symbol_iterator(Symbol(this, 0, string_start_offset));
347}
348
349Archive::symbol_iterator Archive::end_symbols() const {
350 const char *buf = SymbolTable->getBuffer()->getBufferStart();
Shankar Easwaran206252c2012-11-13 18:38:42 +0000351 uint32_t symbol_count = 0;
352 if (kind() == K_GNU) {
353 symbol_count = *reinterpret_cast<const support::ubig32_t*>(buf);
354 buf += sizeof(uint32_t) + (symbol_count * (sizeof(uint32_t)));
355 } else if (kind() == K_BSD) {
Matt Beaumont-Gay7af4b9b2012-11-14 17:58:11 +0000356 llvm_unreachable("BSD archive format is not supported");
Shankar Easwaran206252c2012-11-13 18:38:42 +0000357 } else {
358 uint32_t member_count = 0;
359 member_count = *reinterpret_cast<const support::ulittle32_t*>(buf);
360 buf += 4 + (member_count * 4); // Skip offsets.
361 symbol_count = *reinterpret_cast<const support::ulittle32_t*>(buf);
362 }
Michael J. Spencerc8a55a62011-11-02 19:33:12 +0000363 return symbol_iterator(
364 Symbol(this, symbol_count, 0));
365}
Shankar Easwaran206252c2012-11-13 18:38:42 +0000366
367Archive::child_iterator Archive::findSym(StringRef name) const {
368 Archive::symbol_iterator bs = begin_symbols();
369 Archive::symbol_iterator es = end_symbols();
370 Archive::child_iterator result;
371
372 StringRef symname;
373 for (; bs != es; ++bs) {
374 if (bs->getName(symname))
375 return end_children();
376 if (symname == name) {
377 if (bs->getMember(result))
378 return end_children();
379 return result;
380 }
381 }
382 return end_children();
383}