blob: 0c89baa0d7de424fdfa266c921bf02b35b07762e [file] [log] [blame]
Reid Spencerf9d7a512004-11-14 21:58:33 +00001//===-- ArchiveReader.cpp - Read LLVM archive files -------------*- C++ -*-===//
Misha Brukman2b37d7c2005-04-21 21:13:18 +00002//
John Criswellb576c942003-10-20 19:43:21 +00003// The LLVM Compiler Infrastructure
4//
Chris Lattner4ee451d2007-12-29 20:36:04 +00005// This file is distributed under the University of Illinois Open Source
6// License. See LICENSE.TXT for details.
Misha Brukman2b37d7c2005-04-21 21:13:18 +00007//
John Criswellb576c942003-10-20 19:43:21 +00008//===----------------------------------------------------------------------===//
Chris Lattner968cfd02003-04-19 21:45:34 +00009//
Gabor Greifa99be512007-07-05 17:07:56 +000010// Builds up standard unix archive files (.a) containing LLVM bitcode.
Chris Lattner968cfd02003-04-19 21:45:34 +000011//
12//===----------------------------------------------------------------------===//
13
Reid Spencerf9d7a512004-11-14 21:58:33 +000014#include "ArchiveInternals.h"
Rafael Espindola603d6b52012-01-23 03:41:53 +000015#include "llvm/ADT/SmallPtrSet.h"
Chris Lattnere07c15c2007-05-06 06:18:07 +000016#include "llvm/Bitcode/ReaderWriter.h"
Chris Lattnere07c15c2007-05-06 06:18:07 +000017#include "llvm/Support/MemoryBuffer.h"
Chris Lattnerc1d56242007-05-06 09:28:33 +000018#include "llvm/Module.h"
Dan Gohmand68a0762009-01-05 17:59:02 +000019#include <cstdlib>
Duraid Madina87487462005-12-26 14:31:26 +000020#include <memory>
Chris Lattner3446ae82004-01-10 19:00:15 +000021using namespace llvm;
Brian Gaeked0fde302003-11-11 22:41:34 +000022
Reid Spencerf9d7a512004-11-14 21:58:33 +000023/// Read a variable-bit-rate encoded unsigned integer
Dan Gohman844731a2008-05-13 00:00:25 +000024static inline unsigned readInteger(const char*&At, const char*End) {
Reid Spencerf9d7a512004-11-14 21:58:33 +000025 unsigned Shift = 0;
26 unsigned Result = 0;
Misha Brukman2b37d7c2005-04-21 21:13:18 +000027
Reid Spencerf9d7a512004-11-14 21:58:33 +000028 do {
Misha Brukman2b37d7c2005-04-21 21:13:18 +000029 if (At == End)
Reid Spencer8d8a7ff2006-07-07 20:56:50 +000030 return Result;
Reid Spencerf9d7a512004-11-14 21:58:33 +000031 Result |= (unsigned)((*At++) & 0x7F) << Shift;
32 Shift += 7;
33 } while (At[-1] & 0x80);
34 return Result;
Chris Lattner968cfd02003-04-19 21:45:34 +000035}
36
Reid Spencerf9d7a512004-11-14 21:58:33 +000037// Completely parse the Archive's symbol table and populate symTab member var.
Reid Spencer8d8a7ff2006-07-07 20:56:50 +000038bool
39Archive::parseSymbolTable(const void* data, unsigned size, std::string* error) {
Reid Spencerf9d7a512004-11-14 21:58:33 +000040 const char* At = (const char*) data;
41 const char* End = At + size;
42 while (At < End) {
43 unsigned offset = readInteger(At, End);
Reid Spencer8d8a7ff2006-07-07 20:56:50 +000044 if (At == End) {
45 if (error)
46 *error = "Ran out of data reading vbr_uint for symtab offset!";
47 return false;
48 }
Reid Spencerf9d7a512004-11-14 21:58:33 +000049 unsigned length = readInteger(At, End);
Reid Spencer8d8a7ff2006-07-07 20:56:50 +000050 if (At == End) {
51 if (error)
52 *error = "Ran out of data reading vbr_uint for symtab length!";
53 return false;
54 }
55 if (At + length > End) {
56 if (error)
57 *error = "Malformed symbol table: length not consistent with size";
58 return false;
59 }
Reid Spencerf9d7a512004-11-14 21:58:33 +000060 // we don't care if it can't be inserted (duplicate entry)
Reid Spencer9a29db42004-11-20 07:29:40 +000061 symTab.insert(std::make_pair(std::string(At, length), offset));
Reid Spencerf9d7a512004-11-14 21:58:33 +000062 At += length;
63 }
64 symTabSize = size;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +000065 return true;
Chris Lattner968cfd02003-04-19 21:45:34 +000066}
67
Reid Spencerf9d7a512004-11-14 21:58:33 +000068// This member parses an ArchiveMemberHeader that is presumed to be pointed to
69// by At. The At pointer is updated to the byte just after the header, which
Misha Brukman2b37d7c2005-04-21 21:13:18 +000070// can be variable in size.
Reid Spencerf9d7a512004-11-14 21:58:33 +000071ArchiveMember*
Reid Spencer8d8a7ff2006-07-07 20:56:50 +000072Archive::parseMemberHeader(const char*& At, const char* End, std::string* error)
73{
Reid Spencercd5561a2006-12-15 19:44:51 +000074 if (At + sizeof(ArchiveMemberHeader) >= End) {
75 if (error)
76 *error = "Unexpected end of file";
77 return 0;
78 }
Reid Spencerf9d7a512004-11-14 21:58:33 +000079
80 // Cast archive member header
81 ArchiveMemberHeader* Hdr = (ArchiveMemberHeader*)At;
82 At += sizeof(ArchiveMemberHeader);
83
Misha Brukman2b37d7c2005-04-21 21:13:18 +000084 // Extract the size and determine if the file is
Reid Spencerf9d7a512004-11-14 21:58:33 +000085 // compressed or not (negative length).
86 int flags = 0;
87 int MemberSize = atoi(Hdr->size);
88 if (MemberSize < 0) {
89 flags |= ArchiveMember::CompressedFlag;
90 MemberSize = -MemberSize;
91 }
92
93 // Check the size of the member for sanity
Reid Spencer8d8a7ff2006-07-07 20:56:50 +000094 if (At + MemberSize > End) {
95 if (error)
96 *error = "invalid member length in archive file";
97 return 0;
98 }
Reid Spencerf9d7a512004-11-14 21:58:33 +000099
100 // Check the member signature
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000101 if (!Hdr->checkSignature()) {
102 if (error)
103 *error = "invalid file member signature";
104 return 0;
105 }
Reid Spencerf9d7a512004-11-14 21:58:33 +0000106
107 // Convert and check the member name
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000108 // The empty name ( '/' and 15 blanks) is for a foreign (non-LLVM) symbol
109 // table. The special name "//" and 14 blanks is for a string table, used
Reid Spencerf9d7a512004-11-14 21:58:33 +0000110 // for long file names. This library doesn't generate either of those but
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000111 // it will accept them. If the name starts with #1/ and the remainder is
112 // digits, then those digits specify the length of the name that is
113 // stored immediately following the header. The special name
Gabor Greifa99be512007-07-05 17:07:56 +0000114 // __LLVM_SYM_TAB__ identifies the symbol table for LLVM bitcode.
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000115 // Anything else is a regular, short filename that is terminated with
Reid Spencerf9d7a512004-11-14 21:58:33 +0000116 // a '/' and blanks.
117
118 std::string pathname;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000119 switch (Hdr->name[0]) {
120 case '#':
121 if (Hdr->name[1] == '1' && Hdr->name[2] == '/') {
122 if (isdigit(Hdr->name[3])) {
123 unsigned len = atoi(&Hdr->name[3]);
Chris Lattner46333062010-02-04 06:19:43 +0000124 const char *nulp = (const char *)memchr(At, '\0', len);
Chris Lattner73382422010-04-25 04:44:26 +0000125 pathname.assign(At, nulp != 0 ? (uintptr_t)(nulp - At) : len);
Reid Spencerdd95e8d2004-11-17 16:13:11 +0000126 At += len;
127 MemberSize -= len;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000128 flags |= ArchiveMember::HasLongFilenameFlag;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000129 } else {
130 if (error)
131 *error = "invalid long filename";
132 return 0;
133 }
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000134 } else if (Hdr->name[1] == '_' &&
Reid Spencer9a29db42004-11-20 07:29:40 +0000135 (0 == memcmp(Hdr->name, ARFILE_LLVM_SYMTAB_NAME, 16))) {
Reid Spencerf9d7a512004-11-14 21:58:33 +0000136 // The member is using a long file name (>15 chars) format.
137 // This format is standard for 4.4BSD and Mac OSX operating
138 // systems. LLVM uses it similarly. In this format, the
139 // remainder of the name field (after #1/) specifies the
140 // length of the file name which occupy the first bytes of
141 // the member's data. The pathname already has the #1/ stripped.
142 pathname.assign(ARFILE_LLVM_SYMTAB_NAME);
143 flags |= ArchiveMember::LLVMSymbolTableFlag;
144 }
145 break;
146 case '/':
147 if (Hdr->name[1]== '/') {
Reid Spencer9a29db42004-11-20 07:29:40 +0000148 if (0 == memcmp(Hdr->name, ARFILE_STRTAB_NAME, 16)) {
Reid Spencerf9d7a512004-11-14 21:58:33 +0000149 pathname.assign(ARFILE_STRTAB_NAME);
150 flags |= ArchiveMember::StringTableFlag;
151 } else {
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000152 if (error)
153 *error = "invalid string table name";
154 return 0;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000155 }
156 } else if (Hdr->name[1] == ' ') {
Reid Spencer9a29db42004-11-20 07:29:40 +0000157 if (0 == memcmp(Hdr->name, ARFILE_SVR4_SYMTAB_NAME, 16)) {
158 pathname.assign(ARFILE_SVR4_SYMTAB_NAME);
159 flags |= ArchiveMember::SVR4SymbolTableFlag;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000160 } else {
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000161 if (error)
162 *error = "invalid SVR4 symbol table name";
163 return 0;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000164 }
165 } else if (isdigit(Hdr->name[1])) {
166 unsigned index = atoi(&Hdr->name[1]);
167 if (index < strtab.length()) {
168 const char* namep = strtab.c_str() + index;
169 const char* endp = strtab.c_str() + strtab.length();
170 const char* p = namep;
171 const char* last_p = p;
172 while (p < endp) {
173 if (*p == '\n' && *last_p == '/') {
Reid Spencer9a29db42004-11-20 07:29:40 +0000174 pathname.assign(namep, last_p - namep);
Reid Spencerf9d7a512004-11-14 21:58:33 +0000175 flags |= ArchiveMember::HasLongFilenameFlag;
176 break;
177 }
178 last_p = p;
179 p++;
180 }
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000181 if (p >= endp) {
182 if (error)
183 *error = "missing name termiantor in string table";
184 return 0;
185 }
Reid Spencerf9d7a512004-11-14 21:58:33 +0000186 } else {
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000187 if (error)
188 *error = "name index beyond string table";
189 return 0;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000190 }
191 }
192 break;
Reid Spencer9a29db42004-11-20 07:29:40 +0000193 case '_':
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000194 if (Hdr->name[1] == '_' &&
Reid Spencer9a29db42004-11-20 07:29:40 +0000195 (0 == memcmp(Hdr->name, ARFILE_BSD4_SYMTAB_NAME, 16))) {
196 pathname.assign(ARFILE_BSD4_SYMTAB_NAME);
197 flags |= ArchiveMember::BSD4SymbolTableFlag;
Reid Spencer84b9ced2004-11-23 22:35:39 +0000198 break;
Reid Spencer9a29db42004-11-20 07:29:40 +0000199 }
Reid Spencer84b9ced2004-11-23 22:35:39 +0000200 /* FALL THROUGH */
Reid Spencerf9d7a512004-11-14 21:58:33 +0000201
202 default:
Reid Spencer9a29db42004-11-20 07:29:40 +0000203 char* slash = (char*) memchr(Hdr->name, '/', 16);
Reid Spencerf9d7a512004-11-14 21:58:33 +0000204 if (slash == 0)
Reid Spencerdd95e8d2004-11-17 16:13:11 +0000205 slash = Hdr->name + 16;
Reid Spencer9a29db42004-11-20 07:29:40 +0000206 pathname.assign(Hdr->name, slash - Hdr->name);
Reid Spencerf9d7a512004-11-14 21:58:33 +0000207 break;
208 }
209
Gabor Greifdb5565a2007-07-06 20:28:40 +0000210 // Determine if this is a bitcode file
Reid Spencer9a29db42004-11-20 07:29:40 +0000211 switch (sys::IdentifyFileType(At, 4)) {
Chris Lattnere07c15c2007-05-06 06:18:07 +0000212 case sys::Bitcode_FileType:
Gabor Greife75ca3d2007-07-06 13:38:17 +0000213 flags |= ArchiveMember::BitcodeFlag;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000214 break;
215 default:
Gabor Greife75ca3d2007-07-06 13:38:17 +0000216 flags &= ~ArchiveMember::BitcodeFlag;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000217 break;
218 }
219
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000220 // Instantiate the ArchiveMember to be filled
221 ArchiveMember* member = new ArchiveMember(this);
222
Reid Spencerf9d7a512004-11-14 21:58:33 +0000223 // Fill in fields of the ArchiveMember
Reid Spencerf9d7a512004-11-14 21:58:33 +0000224 member->parent = this;
Reid Spencerdd04df02005-07-07 23:21:43 +0000225 member->path.set(pathname);
Reid Spencerf9d7a512004-11-14 21:58:33 +0000226 member->info.fileSize = MemberSize;
227 member->info.modTime.fromEpochTime(atoi(Hdr->date));
Reid Spencer56127622004-12-29 01:20:24 +0000228 unsigned int mode;
229 sscanf(Hdr->mode, "%o", &mode);
230 member->info.mode = mode;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000231 member->info.user = atoi(Hdr->uid);
232 member->info.group = atoi(Hdr->gid);
233 member->flags = flags;
234 member->data = At;
235
236 return member;
237}
238
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000239bool
240Archive::checkSignature(std::string* error) {
Reid Spencerf9d7a512004-11-14 21:58:33 +0000241 // Check the magic string at file's header
Chris Lattner7f6b4472008-04-01 04:26:46 +0000242 if (mapfile->getBufferSize() < 8 || memcmp(base, ARFILE_MAGIC, 8)) {
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000243 if (error)
244 *error = "invalid signature for an archive file";
245 return false;
246 }
247 return true;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000248}
249
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000250// This function loads the entire archive and fully populates its ilist with
Reid Spencerf9d7a512004-11-14 21:58:33 +0000251// the members of the archive file. This is typically used in preparation for
252// editing the contents of the archive.
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000253bool
254Archive::loadArchive(std::string* error) {
Reid Spencerf9d7a512004-11-14 21:58:33 +0000255
256 // Set up parsing
257 members.clear();
258 symTab.clear();
259 const char *At = base;
Chris Lattner7f6b4472008-04-01 04:26:46 +0000260 const char *End = mapfile->getBufferEnd();
Reid Spencerf9d7a512004-11-14 21:58:33 +0000261
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000262 if (!checkSignature(error))
263 return false;
264
Reid Spencerf9d7a512004-11-14 21:58:33 +0000265 At += 8; // Skip the magic string.
266
267 bool seenSymbolTable = false;
268 bool foundFirstFile = false;
269 while (At < End) {
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000270 // parse the member header
Reid Spencerf9d7a512004-11-14 21:58:33 +0000271 const char* Save = At;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000272 ArchiveMember* mbr = parseMemberHeader(At, End, error);
273 if (!mbr)
274 return false;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000275
276 // check if this is the foreign symbol table
Reid Spencer9a29db42004-11-20 07:29:40 +0000277 if (mbr->isSVR4SymbolTable() || mbr->isBSD4SymbolTable()) {
Reid Spencer4a980d12004-11-16 06:47:19 +0000278 // We just save this but don't do anything special
279 // with it. It doesn't count as the "first file".
Reid Spencer9a29db42004-11-20 07:29:40 +0000280 if (foreignST) {
281 // What? Multiple foreign symbol tables? Just chuck it
282 // and retain the last one found.
283 delete foreignST;
284 }
Reid Spencer4a980d12004-11-16 06:47:19 +0000285 foreignST = mbr;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000286 At += mbr->getSize();
Reid Spencer6405c9e2004-11-19 17:08:00 +0000287 if ((intptr_t(At) & 1) == 1)
Reid Spencerf9d7a512004-11-14 21:58:33 +0000288 At++;
289 } else if (mbr->isStringTable()) {
Reid Spencer4a980d12004-11-16 06:47:19 +0000290 // Simply suck the entire string table into a string
291 // variable. This will be used to get the names of the
292 // members that use the "/ddd" format for their names
293 // (SVR4 style long names).
Reid Spencer9a29db42004-11-20 07:29:40 +0000294 strtab.assign(At, mbr->getSize());
Reid Spencerf9d7a512004-11-14 21:58:33 +0000295 At += mbr->getSize();
Reid Spencer6405c9e2004-11-19 17:08:00 +0000296 if ((intptr_t(At) & 1) == 1)
Reid Spencerf9d7a512004-11-14 21:58:33 +0000297 At++;
298 delete mbr;
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000299 } else if (mbr->isLLVMSymbolTable()) {
Reid Spencer4a980d12004-11-16 06:47:19 +0000300 // This is the LLVM symbol table for the archive. If we've seen it
301 // already, its an error. Otherwise, parse the symbol table and move on.
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000302 if (seenSymbolTable) {
303 if (error)
304 *error = "invalid archive: multiple symbol tables";
305 return false;
306 }
307 if (!parseSymbolTable(mbr->getData(), mbr->getSize(), error))
308 return false;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000309 seenSymbolTable = true;
310 At += mbr->getSize();
Reid Spencer6405c9e2004-11-19 17:08:00 +0000311 if ((intptr_t(At) & 1) == 1)
Reid Spencerf9d7a512004-11-14 21:58:33 +0000312 At++;
Reid Spencer4a980d12004-11-16 06:47:19 +0000313 delete mbr; // We don't need this member in the list of members.
Reid Spencerf9d7a512004-11-14 21:58:33 +0000314 } else {
Reid Spencer4a980d12004-11-16 06:47:19 +0000315 // This is just a regular file. If its the first one, save its offset.
316 // Otherwise just push it on the list and move on to the next file.
Reid Spencerf9d7a512004-11-14 21:58:33 +0000317 if (!foundFirstFile) {
318 firstFileOffset = Save - base;
319 foundFirstFile = true;
320 }
321 members.push_back(mbr);
322 At += mbr->getSize();
Reid Spencer6405c9e2004-11-19 17:08:00 +0000323 if ((intptr_t(At) & 1) == 1)
Reid Spencerf9d7a512004-11-14 21:58:33 +0000324 At++;
325 }
326 }
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000327 return true;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000328}
329
330// Open and completely load the archive file.
331Archive*
Owen Anderson4434ed42009-07-01 23:13:44 +0000332Archive::OpenAndLoad(const sys::Path& file, LLVMContext& C,
Owen Anderson8b477ed2009-07-01 16:58:40 +0000333 std::string* ErrorMessage) {
334 std::auto_ptr<Archive> result ( new Archive(file, C));
Reid Spencer0ff2d312006-08-24 23:45:08 +0000335 if (result->mapToMemory(ErrorMessage))
336 return 0;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000337 if (!result->loadArchive(ErrorMessage))
Reid Spencer5af46882004-12-13 02:59:03 +0000338 return 0;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000339 return result.release();
Reid Spencerf9d7a512004-11-14 21:58:33 +0000340}
341
Gabor Greifa99be512007-07-05 17:07:56 +0000342// Get all the bitcode modules from the archive
Reid Spencerf9d7a512004-11-14 21:58:33 +0000343bool
Owen Anderson8b477ed2009-07-01 16:58:40 +0000344Archive::getAllModules(std::vector<Module*>& Modules,
345 std::string* ErrMessage) {
Reid Spencerf9d7a512004-11-14 21:58:33 +0000346
347 for (iterator I=begin(), E=end(); I != E; ++I) {
Gabor Greife75ca3d2007-07-06 13:38:17 +0000348 if (I->isBitcode()) {
Chris Lattner74382b72009-08-23 22:45:37 +0000349 std::string FullMemberName = archPath.str() +
350 "(" + I->getPath().str() + ")";
Chris Lattnerc1d56242007-05-06 09:28:33 +0000351 MemoryBuffer *Buffer =
Benjamin Kramer9d44e702010-04-19 16:15:31 +0000352 MemoryBuffer::getMemBufferCopy(StringRef(I->getData(), I->getSize()),
353 FullMemberName.c_str());
Chris Lattnere07c15c2007-05-06 06:18:07 +0000354
Owen Anderson8b477ed2009-07-01 16:58:40 +0000355 Module *M = ParseBitcodeFile(Buffer, Context, ErrMessage);
Chris Lattnerc1d56242007-05-06 09:28:33 +0000356 delete Buffer;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000357 if (!M)
358 return true;
359
360 Modules.push_back(M);
361 }
362 }
Brian Gaeke2c61d7b2003-11-16 23:08:48 +0000363 return false;
364}
Chris Lattner968cfd02003-04-19 21:45:34 +0000365
Reid Spencerf9d7a512004-11-14 21:58:33 +0000366// Load just the symbol table from the archive file
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000367bool
368Archive::loadSymbolTable(std::string* ErrorMsg) {
Chris Lattner968cfd02003-04-19 21:45:34 +0000369
Reid Spencerf9d7a512004-11-14 21:58:33 +0000370 // Set up parsing
371 members.clear();
372 symTab.clear();
373 const char *At = base;
Chris Lattner7f6b4472008-04-01 04:26:46 +0000374 const char *End = mapfile->getBufferEnd();
Chris Lattner968cfd02003-04-19 21:45:34 +0000375
Reid Spencerf9d7a512004-11-14 21:58:33 +0000376 // Make sure we're dealing with an archive
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000377 if (!checkSignature(ErrorMsg))
378 return false;
Chris Lattner968cfd02003-04-19 21:45:34 +0000379
Reid Spencerf9d7a512004-11-14 21:58:33 +0000380 At += 8; // Skip signature
381
382 // Parse the first file member header
383 const char* FirstFile = At;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000384 ArchiveMember* mbr = parseMemberHeader(At, End, ErrorMsg);
385 if (!mbr)
386 return false;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000387
Reid Spencer9a29db42004-11-20 07:29:40 +0000388 if (mbr->isSVR4SymbolTable() || mbr->isBSD4SymbolTable()) {
Reid Spencerf9d7a512004-11-14 21:58:33 +0000389 // Skip the foreign symbol table, we don't do anything with it
390 At += mbr->getSize();
Reid Spencer6405c9e2004-11-19 17:08:00 +0000391 if ((intptr_t(At) & 1) == 1)
Reid Spencerb3231132004-11-15 01:40:20 +0000392 At++;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000393 delete mbr;
394
Reid Spencerb3231132004-11-15 01:40:20 +0000395 // Read the next one
Reid Spencerf9d7a512004-11-14 21:58:33 +0000396 FirstFile = At;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000397 mbr = parseMemberHeader(At, End, ErrorMsg);
398 if (!mbr) {
399 delete mbr;
400 return false;
401 }
Reid Spencerb3231132004-11-15 01:40:20 +0000402 }
403
404 if (mbr->isStringTable()) {
405 // Process the string table entry
Reid Spencer9a29db42004-11-20 07:29:40 +0000406 strtab.assign((const char*)mbr->getData(), mbr->getSize());
Reid Spencerb3231132004-11-15 01:40:20 +0000407 At += mbr->getSize();
Reid Spencer6405c9e2004-11-19 17:08:00 +0000408 if ((intptr_t(At) & 1) == 1)
Reid Spencerb3231132004-11-15 01:40:20 +0000409 At++;
410 delete mbr;
411 // Get the next one
412 FirstFile = At;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000413 mbr = parseMemberHeader(At, End, ErrorMsg);
414 if (!mbr) {
415 delete mbr;
416 return false;
417 }
Chris Lattner968cfd02003-04-19 21:45:34 +0000418 }
419
Reid Spencerf9d7a512004-11-14 21:58:33 +0000420 // See if its the symbol table
421 if (mbr->isLLVMSymbolTable()) {
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000422 if (!parseSymbolTable(mbr->getData(), mbr->getSize(), ErrorMsg)) {
423 delete mbr;
424 return false;
425 }
426
Reid Spencer8dde18f2004-11-28 03:13:02 +0000427 At += mbr->getSize();
Reid Spencer6405c9e2004-11-19 17:08:00 +0000428 if ((intptr_t(At) & 1) == 1)
Reid Spencer8dde18f2004-11-28 03:13:02 +0000429 At++;
Reid Spencerffb4d622006-05-12 17:56:20 +0000430 delete mbr;
431 // Can't be any more symtab headers so just advance
Reid Spencer8dde18f2004-11-28 03:13:02 +0000432 FirstFile = At;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000433 } else {
434 // There's no symbol table in the file. We have to rebuild it from scratch
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000435 // because the intent of this method is to get the symbol table loaded so
436 // it can be searched efficiently.
Reid Spencerf9d7a512004-11-14 21:58:33 +0000437 // Add the member to the members list
438 members.push_back(mbr);
439 }
440
441 firstFileOffset = FirstFile - base;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000442 return true;
Chris Lattner968cfd02003-04-19 21:45:34 +0000443}
444
Reid Spencerf9d7a512004-11-14 21:58:33 +0000445// Open the archive and load just the symbol tables
Owen Anderson31895e72009-07-01 21:22:36 +0000446Archive* Archive::OpenAndLoadSymbols(const sys::Path& file,
Owen Anderson4434ed42009-07-01 23:13:44 +0000447 LLVMContext& C,
Owen Anderson8b477ed2009-07-01 16:58:40 +0000448 std::string* ErrorMessage) {
449 std::auto_ptr<Archive> result ( new Archive(file, C) );
Reid Spencer0ff2d312006-08-24 23:45:08 +0000450 if (result->mapToMemory(ErrorMessage))
451 return 0;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000452 if (!result->loadSymbolTable(ErrorMessage))
Reid Spencer5af46882004-12-13 02:59:03 +0000453 return 0;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000454 return result.release();
Reid Spencerf9d7a512004-11-14 21:58:33 +0000455}
Chris Lattner968cfd02003-04-19 21:45:34 +0000456
Jeffrey Yasskinf0356fe2010-01-27 20:34:15 +0000457// Look up one symbol in the symbol table and return the module that defines
458// that symbol.
459Module*
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000460Archive::findModuleDefiningSymbol(const std::string& symbol,
461 std::string* ErrMsg) {
Reid Spencerf9d7a512004-11-14 21:58:33 +0000462 SymTabType::iterator SI = symTab.find(symbol);
463 if (SI == symTab.end())
464 return 0;
465
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000466 // The symbol table was previously constructed assuming that the members were
Reid Spencerf9d7a512004-11-14 21:58:33 +0000467 // written without the symbol table header. Because VBR encoding is used, the
468 // values could not be adjusted to account for the offset of the symbol table
469 // because that could affect the size of the symbol table due to VBR encoding.
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000470 // We now have to account for this by adjusting the offset by the size of the
Reid Spencerf9d7a512004-11-14 21:58:33 +0000471 // symbol table and its header.
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000472 unsigned fileOffset =
Reid Spencerf9d7a512004-11-14 21:58:33 +0000473 SI->second + // offset in symbol-table-less file
474 firstFileOffset; // add offset to first "real" file in archive
475
476 // See if the module is already loaded
477 ModuleMap::iterator MI = modules.find(fileOffset);
478 if (MI != modules.end())
479 return MI->second.first;
480
481 // Module hasn't been loaded yet, we need to load it
482 const char* modptr = base + fileOffset;
Chris Lattner7f6b4472008-04-01 04:26:46 +0000483 ArchiveMember* mbr = parseMemberHeader(modptr, mapfile->getBufferEnd(),
484 ErrMsg);
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000485 if (!mbr)
Reid Spencer0b5a5042006-08-25 17:43:11 +0000486 return 0;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000487
Jeffrey Yasskinf0356fe2010-01-27 20:34:15 +0000488 // Now, load the bitcode module to get the Module.
Chris Lattner74382b72009-08-23 22:45:37 +0000489 std::string FullMemberName = archPath.str() + "(" +
490 mbr->getPath().str() + ")";
Benjamin Kramer9d44e702010-04-19 16:15:31 +0000491 MemoryBuffer *Buffer =
492 MemoryBuffer::getMemBufferCopy(StringRef(mbr->getData(), mbr->getSize()),
493 FullMemberName.c_str());
Chris Lattnerc1d56242007-05-06 09:28:33 +0000494
Jeffrey Yasskinf0356fe2010-01-27 20:34:15 +0000495 Module *m = getLazyBitcodeModule(Buffer, Context, ErrMsg);
496 if (!m)
Reid Spencer0b5a5042006-08-25 17:43:11 +0000497 return 0;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000498
Jeffrey Yasskinf0356fe2010-01-27 20:34:15 +0000499 modules.insert(std::make_pair(fileOffset, std::make_pair(m, mbr)));
Reid Spencerf9d7a512004-11-14 21:58:33 +0000500
Jeffrey Yasskinf0356fe2010-01-27 20:34:15 +0000501 return m;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000502}
503
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000504// Look up multiple symbols in the symbol table and return a set of
Jeffrey Yasskinf0356fe2010-01-27 20:34:15 +0000505// Modules that define those symbols.
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000506bool
Reid Spencer7783e8a2004-11-19 03:18:22 +0000507Archive::findModulesDefiningSymbols(std::set<std::string>& symbols,
Rafael Espindola603d6b52012-01-23 03:41:53 +0000508 SmallVectorImpl<Module*>& result,
Chris Lattnerf2e292c2007-02-07 21:41:02 +0000509 std::string* error) {
Reid Spencercd5561a2006-12-15 19:44:51 +0000510 if (!mapfile || !base) {
511 if (error)
512 *error = "Empty archive invalid for finding modules defining symbols";
513 return false;
514 }
515
Reid Spencer766b7932004-11-15 01:20:11 +0000516 if (symTab.empty()) {
517 // We don't have a symbol table, so we must build it now but lets also
518 // make sure that we populate the modules table as we do this to ensure
519 // that we don't load them twice when findModuleDefiningSymbol is called
520 // below.
521
522 // Get a pointer to the first file
Chris Lattner7f6b4472008-04-01 04:26:46 +0000523 const char* At = base + firstFileOffset;
524 const char* End = mapfile->getBufferEnd();
Reid Spencer766b7932004-11-15 01:20:11 +0000525
526 while ( At < End) {
527 // Compute the offset to be put in the symbol table
528 unsigned offset = At - base - firstFileOffset;
529
530 // Parse the file's header
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000531 ArchiveMember* mbr = parseMemberHeader(At, End, error);
532 if (!mbr)
533 return false;
Reid Spencer766b7932004-11-15 01:20:11 +0000534
535 // If it contains symbols
Gabor Greife75ca3d2007-07-06 13:38:17 +0000536 if (mbr->isBitcode()) {
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000537 // Get the symbols
Reid Spencer766b7932004-11-15 01:20:11 +0000538 std::vector<std::string> symbols;
Chris Lattner74382b72009-08-23 22:45:37 +0000539 std::string FullMemberName = archPath.str() + "(" +
540 mbr->getPath().str() + ")";
Jeffrey Yasskinf0356fe2010-01-27 20:34:15 +0000541 Module* M =
Benjamin Kramer9d44e702010-04-19 16:15:31 +0000542 GetBitcodeSymbols(At, mbr->getSize(), FullMemberName, Context,
543 symbols, error);
Reid Spencer766b7932004-11-15 01:20:11 +0000544
Jeffrey Yasskinf0356fe2010-01-27 20:34:15 +0000545 if (M) {
Reid Spencer766b7932004-11-15 01:20:11 +0000546 // Insert the module's symbols into the symbol table
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000547 for (std::vector<std::string>::iterator I = symbols.begin(),
Reid Spencer766b7932004-11-15 01:20:11 +0000548 E=symbols.end(); I != E; ++I ) {
Reid Spencer9a29db42004-11-20 07:29:40 +0000549 symTab.insert(std::make_pair(*I, offset));
Reid Spencer766b7932004-11-15 01:20:11 +0000550 }
Jeffrey Yasskinf0356fe2010-01-27 20:34:15 +0000551 // Insert the Module and the ArchiveMember into the table of
Reid Spencer766b7932004-11-15 01:20:11 +0000552 // modules.
Jeffrey Yasskinf0356fe2010-01-27 20:34:15 +0000553 modules.insert(std::make_pair(offset, std::make_pair(M, mbr)));
Reid Spencer766b7932004-11-15 01:20:11 +0000554 } else {
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000555 if (error)
Gabor Greifa99be512007-07-05 17:07:56 +0000556 *error = "Can't parse bitcode member: " +
Chris Lattner74382b72009-08-23 22:45:37 +0000557 mbr->getPath().str() + ": " + *error;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000558 delete mbr;
559 return false;
Reid Spencer766b7932004-11-15 01:20:11 +0000560 }
561 }
Reid Spencerb3231132004-11-15 01:40:20 +0000562
563 // Go to the next file location
564 At += mbr->getSize();
Reid Spencer6405c9e2004-11-19 17:08:00 +0000565 if ((intptr_t(At) & 1) == 1)
Reid Spencerb3231132004-11-15 01:40:20 +0000566 At++;
Reid Spencer766b7932004-11-15 01:20:11 +0000567 }
568 }
569
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000570 // At this point we have a valid symbol table (one way or another) so we
Reid Spencer766b7932004-11-15 01:20:11 +0000571 // just use it to quickly find the symbols requested.
572
Rafael Espindola603d6b52012-01-23 03:41:53 +0000573 SmallPtrSet<Module*, 16> Added;
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000574 for (std::set<std::string>::iterator I=symbols.begin(),
Rafael Espindola603d6b52012-01-23 03:41:53 +0000575 E=symbols.end(); I != E; ++I) {
Reid Spencer7783e8a2004-11-19 03:18:22 +0000576 // See if this symbol exists
Jeffrey Yasskinf0356fe2010-01-27 20:34:15 +0000577 Module* m = findModuleDefiningSymbol(*I,error);
Rafael Espindola603d6b52012-01-23 03:41:53 +0000578 if (!m)
579 continue;
580 bool NewMember = Added.insert(m);
581 if (!NewMember)
582 continue;
Reid Spencer7783e8a2004-11-19 03:18:22 +0000583
Rafael Espindola603d6b52012-01-23 03:41:53 +0000584 // The symbol exists, insert the Module into our result.
585 result.push_back(m);
586
587 // Remove the symbol now that its been resolved.
588 symbols.erase(I);
Reid Spencerf9d7a512004-11-14 21:58:33 +0000589 }
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000590 return true;
Chris Lattner968cfd02003-04-19 21:45:34 +0000591}
Reid Spencereaa06bb2005-02-26 22:00:32 +0000592
Gabor Greifa99be512007-07-05 17:07:56 +0000593bool Archive::isBitcodeArchive() {
Chris Lattner7c439922005-09-23 06:22:58 +0000594 // Make sure the symTab has been loaded. In most cases this should have been
595 // done when the archive was constructed, but still, this is just in case.
Dan Gohman30359592008-01-29 13:02:09 +0000596 if (symTab.empty())
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000597 if (!loadSymbolTable(0))
598 return false;
Reid Spencereaa06bb2005-02-26 22:00:32 +0000599
Chris Lattner7c439922005-09-23 06:22:58 +0000600 // Now that we know it's been loaded, return true
601 // if it has a size
602 if (symTab.size()) return true;
Reid Spencereaa06bb2005-02-26 22:00:32 +0000603
Gabor Greifa99be512007-07-05 17:07:56 +0000604 // We still can't be sure it isn't a bitcode archive
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000605 if (!loadArchive(0))
606 return false;
Reid Spencereaa06bb2005-02-26 22:00:32 +0000607
608 std::vector<Module *> Modules;
609 std::string ErrorMessage;
610
Gabor Greifa99be512007-07-05 17:07:56 +0000611 // Scan the archive, trying to load a bitcode member. We only load one to
Chris Lattner7c439922005-09-23 06:22:58 +0000612 // see if this works.
613 for (iterator I = begin(), E = end(); I != E; ++I) {
Gabor Greife75ca3d2007-07-06 13:38:17 +0000614 if (!I->isBitcode())
Chris Lattner7c439922005-09-23 06:22:58 +0000615 continue;
616
617 std::string FullMemberName =
Chris Lattner74382b72009-08-23 22:45:37 +0000618 archPath.str() + "(" + I->getPath().str() + ")";
Chris Lattnerc1d56242007-05-06 09:28:33 +0000619
620 MemoryBuffer *Buffer =
Benjamin Kramer9d44e702010-04-19 16:15:31 +0000621 MemoryBuffer::getMemBufferCopy(StringRef(I->getData(), I->getSize()),
622 FullMemberName.c_str());
Owen Anderson8b477ed2009-07-01 16:58:40 +0000623 Module *M = ParseBitcodeFile(Buffer, Context);
Chris Lattnerc1d56242007-05-06 09:28:33 +0000624 delete Buffer;
Chris Lattner7c439922005-09-23 06:22:58 +0000625 if (!M)
Gabor Greifa99be512007-07-05 17:07:56 +0000626 return false; // Couldn't parse bitcode, not a bitcode archive.
Chris Lattner7c439922005-09-23 06:22:58 +0000627 delete M;
628 return true;
629 }
630
631 return false;
Reid Spencereaa06bb2005-02-26 22:00:32 +0000632}