1 //===- Archive.cpp - ar File Format implementation --------------*- C++ -*-===//
3 // The LLVM Compiler Infrastructure
5 // This file is distributed under the University of Illinois Open Source
6 // License. See LICENSE.TXT for details.
8 //===----------------------------------------------------------------------===//
10 // This file defines the ArchiveObjectFile class.
12 //===----------------------------------------------------------------------===//
14 #include "llvm/Object/Archive.h"
15 #include "llvm/ADT/APInt.h"
16 #include "llvm/Support/Endian.h"
17 #include "llvm/Support/MemoryBuffer.h"
20 using namespace object;
22 static const char *Magic = "!<arch>\n";
24 static bool isInternalMember(const ArchiveMemberHeader &amh) {
25 static const char *const internals[] = {
30 StringRef name = amh.getName();
31 for (std::size_t i = 0; i < sizeof(internals) / sizeof(*internals); ++i) {
32 if (name == internals[i])
38 void Archive::anchor() { }
40 error_code Archive::Child::getName(StringRef &Result) const {
41 StringRef name = ToHeader(Data.data())->getName();
42 // Check if it's a special name.
44 if (name.size() == 1) { // Linker member.
46 return object_error::success;
48 if (name.size() == 2 && name[1] == '/') { // String table.
50 return object_error::success;
55 if (name.substr(1).rtrim(" ").getAsInteger(10, offset))
56 llvm_unreachable("Long name offset is not an integer");
57 const char *addr = Parent->StringTable->Data.begin()
58 + sizeof(ArchiveMemberHeader)
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;
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);
76 return object_error::success;
77 } else if (name.startswith("#1/")) {
79 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);
82 return object_error::success;
84 // It's a simple name.
85 if (name[name.size() - 1] == '/')
86 Result = name.substr(0, name.size() - 1);
89 return object_error::success;
92 error_code Archive::Child::getAsBinary(OwningPtr<Binary> &Result) const {
93 OwningPtr<Binary> ret;
94 OwningPtr<MemoryBuffer> Buff;
95 if (error_code ec = getMemoryBuffer(Buff))
97 if (error_code ec = createBinary(Buff.take(), ret))
100 return object_error::success;
103 Archive::Archive(MemoryBuffer *source, error_code &ec)
104 : Binary(Binary::ID_Archive, source) {
105 // Check for sufficient magic.
106 if (!source || source->getBufferSize()
107 < (8 + sizeof(ArchiveMemberHeader) + 2) // Smallest archive.
108 || StringRef(source->getBufferStart(), 8) != Magic) {
109 ec = object_error::invalid_file_type;
113 // Get the special members.
114 child_iterator i = begin_children(false);
115 child_iterator e = end_children();
118 if ((ec = i->getName(name)))
121 // Below is the pattern that is used to figure out the archive format
122 // GNU archive format
123 // First member : / (points to the symbol table )
124 // Second member : // (may exist, if it exists, points to the string table)
125 // Note : The string table is used if the filename exceeds 15 characters
126 // BSD archive format
127 // First member : __.SYMDEF (points to the symbol table)
128 // There is no string table, if the filename exceeds 15 characters or has a
129 // embedded space, the filename has #1/<size>, The size represents the size
130 // of the filename that needs to be read after the archive header
131 // COFF archive format
133 // Second member : / (provides a directory of symbols)
134 // Third member : // (may exist, if it exists, contains the string table)
135 // Note: Microsoft PE/COFF Spec 8.3 says that the third member is present
136 // even if the string table is empty. However, lib.exe does not in fact
137 // seem to create the third member if there's no member whose filename
138 // exceeds 15 characters. So the third member is optional.
144 ec = object_error::parse_failed;
147 if ((ec = i->getName(name)))
149 if (name[0] != '/') {
151 } else if ((name.size() > 1) && (name == "//")) {
162 if ((ec = i->getName(name)))
168 } else if (name == "__.SYMDEF") {
173 ec = object_error::success;
176 Archive::child_iterator Archive::begin_children(bool skip_internal) const {
177 const char *Loc = Data->getBufferStart() + strlen(Magic);
178 size_t Size = sizeof(ArchiveMemberHeader) +
179 ToHeader(Loc)->getSize();
180 Child c(this, StringRef(Loc, Size));
181 // Skip internals at the beginning of an archive.
182 if (skip_internal && isInternalMember(*ToHeader(Loc)))
187 Archive::child_iterator Archive::end_children() const {
188 return Child(this, StringRef(0, 0));
191 error_code Archive::Symbol::getName(StringRef &Result) const {
192 Result = StringRef(Parent->SymbolTable->getBuffer().begin() + StringIndex);
193 return object_error::success;
196 error_code Archive::Symbol::getMember(child_iterator &Result) const {
197 const char *Buf = Parent->SymbolTable->getBuffer().begin();
198 const char *Offsets = Buf + 4;
200 if (Parent->kind() == K_GNU) {
201 Offset = *(reinterpret_cast<const support::ubig32_t*>(Offsets)
203 } else if (Parent->kind() == K_BSD) {
204 llvm_unreachable("BSD format is not supported");
206 uint32_t MemberCount = *reinterpret_cast<const support::ulittle32_t*>(Buf);
209 Buf += sizeof(support::ulittle32_t)
210 + (MemberCount * sizeof(support::ulittle32_t));
212 uint32_t SymbolCount = *reinterpret_cast<const support::ulittle32_t*>(Buf);
214 if (SymbolIndex >= SymbolCount)
215 return object_error::parse_failed;
217 // Skip SymbolCount to get to the indices table.
218 const char *Indices = Buf + sizeof(support::ulittle32_t);
220 // Get the index of the offset in the file member offset table for this
222 uint16_t OffsetIndex =
223 *(reinterpret_cast<const support::ulittle16_t*>(Indices)
225 // Subtract 1 since OffsetIndex is 1 based.
228 if (OffsetIndex >= MemberCount)
229 return object_error::parse_failed;
231 Offset = *(reinterpret_cast<const support::ulittle32_t*>(Offsets)
235 const char *Loc = Parent->getData().begin() + Offset;
236 size_t Size = sizeof(ArchiveMemberHeader) +
237 ToHeader(Loc)->getSize();
238 Result = Child(Parent, StringRef(Loc, Size));
240 return object_error::success;
243 Archive::Symbol Archive::Symbol::getNext() const {
245 // Go to one past next null.
247 Parent->SymbolTable->getBuffer().find('\0', t.StringIndex) + 1;
252 Archive::symbol_iterator Archive::begin_symbols() const {
253 const char *buf = SymbolTable->getBuffer().begin();
254 if (kind() == K_GNU) {
255 uint32_t symbol_count = 0;
256 symbol_count = *reinterpret_cast<const support::ubig32_t*>(buf);
257 buf += sizeof(uint32_t) + (symbol_count * (sizeof(uint32_t)));
258 } else if (kind() == K_BSD) {
259 llvm_unreachable("BSD archive format is not supported");
261 uint32_t member_count = 0;
262 uint32_t symbol_count = 0;
263 member_count = *reinterpret_cast<const support::ulittle32_t*>(buf);
264 buf += 4 + (member_count * 4); // Skip offsets.
265 symbol_count = *reinterpret_cast<const support::ulittle32_t*>(buf);
266 buf += 4 + (symbol_count * 2); // Skip indices.
268 uint32_t string_start_offset = buf - SymbolTable->getBuffer().begin();
269 return symbol_iterator(Symbol(this, 0, string_start_offset));
272 Archive::symbol_iterator Archive::end_symbols() const {
273 const char *buf = SymbolTable->getBuffer().begin();
274 uint32_t symbol_count = 0;
275 if (kind() == K_GNU) {
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) {
279 llvm_unreachable("BSD archive format is not supported");
281 uint32_t member_count = 0;
282 member_count = *reinterpret_cast<const support::ulittle32_t*>(buf);
283 buf += 4 + (member_count * 4); // Skip offsets.
284 symbol_count = *reinterpret_cast<const support::ulittle32_t*>(buf);
286 return symbol_iterator(
287 Symbol(this, symbol_count, 0));
290 Archive::child_iterator Archive::findSym(StringRef name) const {
291 Archive::symbol_iterator bs = begin_symbols();
292 Archive::symbol_iterator es = end_symbols();
293 Archive::child_iterator result;
296 for (; bs != es; ++bs) {
297 if (bs->getName(symname))
298 return end_children();
299 if (symname == name) {
300 if (bs->getMember(result))
301 return end_children();
305 return end_children();