blob: b07e884b654796f0e01c72104eabaf917193d15d [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"
Chris Lattnere07c15c2007-05-06 06:18:07 +000015#include "llvm/Bitcode/ReaderWriter.h"
Chris Lattnere07c15c2007-05-06 06:18:07 +000016#include "llvm/Support/MemoryBuffer.h"
Chris Lattnerc1d56242007-05-06 09:28:33 +000017#include "llvm/Module.h"
Dan Gohmand68a0762009-01-05 17:59:02 +000018#include <cstdlib>
Duraid Madina87487462005-12-26 14:31:26 +000019#include <memory>
Chris Lattner3446ae82004-01-10 19:00:15 +000020using namespace llvm;
Brian Gaeked0fde302003-11-11 22:41:34 +000021
Reid Spencerf9d7a512004-11-14 21:58:33 +000022/// Read a variable-bit-rate encoded unsigned integer
Dan Gohman844731a2008-05-13 00:00:25 +000023static inline unsigned readInteger(const char*&At, const char*End) {
Reid Spencerf9d7a512004-11-14 21:58:33 +000024 unsigned Shift = 0;
25 unsigned Result = 0;
Misha Brukman2b37d7c2005-04-21 21:13:18 +000026
Reid Spencerf9d7a512004-11-14 21:58:33 +000027 do {
Misha Brukman2b37d7c2005-04-21 21:13:18 +000028 if (At == End)
Reid Spencer8d8a7ff2006-07-07 20:56:50 +000029 return Result;
Reid Spencerf9d7a512004-11-14 21:58:33 +000030 Result |= (unsigned)((*At++) & 0x7F) << Shift;
31 Shift += 7;
32 } while (At[-1] & 0x80);
33 return Result;
Chris Lattner968cfd02003-04-19 21:45:34 +000034}
35
Reid Spencerf9d7a512004-11-14 21:58:33 +000036// Completely parse the Archive's symbol table and populate symTab member var.
Reid Spencer8d8a7ff2006-07-07 20:56:50 +000037bool
38Archive::parseSymbolTable(const void* data, unsigned size, std::string* error) {
Reid Spencerf9d7a512004-11-14 21:58:33 +000039 const char* At = (const char*) data;
40 const char* End = At + size;
41 while (At < End) {
42 unsigned offset = readInteger(At, End);
Reid Spencer8d8a7ff2006-07-07 20:56:50 +000043 if (At == End) {
44 if (error)
45 *error = "Ran out of data reading vbr_uint for symtab offset!";
46 return false;
47 }
Reid Spencerf9d7a512004-11-14 21:58:33 +000048 unsigned length = readInteger(At, End);
Reid Spencer8d8a7ff2006-07-07 20:56:50 +000049 if (At == End) {
50 if (error)
51 *error = "Ran out of data reading vbr_uint for symtab length!";
52 return false;
53 }
54 if (At + length > End) {
55 if (error)
56 *error = "Malformed symbol table: length not consistent with size";
57 return false;
58 }
Reid Spencerf9d7a512004-11-14 21:58:33 +000059 // we don't care if it can't be inserted (duplicate entry)
Reid Spencer9a29db42004-11-20 07:29:40 +000060 symTab.insert(std::make_pair(std::string(At, length), offset));
Reid Spencerf9d7a512004-11-14 21:58:33 +000061 At += length;
62 }
63 symTabSize = size;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +000064 return true;
Chris Lattner968cfd02003-04-19 21:45:34 +000065}
66
Reid Spencerf9d7a512004-11-14 21:58:33 +000067// This member parses an ArchiveMemberHeader that is presumed to be pointed to
68// by At. The At pointer is updated to the byte just after the header, which
Misha Brukman2b37d7c2005-04-21 21:13:18 +000069// can be variable in size.
Reid Spencerf9d7a512004-11-14 21:58:33 +000070ArchiveMember*
Reid Spencer8d8a7ff2006-07-07 20:56:50 +000071Archive::parseMemberHeader(const char*& At, const char* End, std::string* error)
72{
Reid Spencercd5561a2006-12-15 19:44:51 +000073 if (At + sizeof(ArchiveMemberHeader) >= End) {
74 if (error)
75 *error = "Unexpected end of file";
76 return 0;
77 }
Reid Spencerf9d7a512004-11-14 21:58:33 +000078
79 // Cast archive member header
80 ArchiveMemberHeader* Hdr = (ArchiveMemberHeader*)At;
81 At += sizeof(ArchiveMemberHeader);
82
Misha Brukman2b37d7c2005-04-21 21:13:18 +000083 // Extract the size and determine if the file is
Reid Spencerf9d7a512004-11-14 21:58:33 +000084 // compressed or not (negative length).
85 int flags = 0;
86 int MemberSize = atoi(Hdr->size);
87 if (MemberSize < 0) {
88 flags |= ArchiveMember::CompressedFlag;
89 MemberSize = -MemberSize;
90 }
91
92 // Check the size of the member for sanity
Reid Spencer8d8a7ff2006-07-07 20:56:50 +000093 if (At + MemberSize > End) {
94 if (error)
95 *error = "invalid member length in archive file";
96 return 0;
97 }
Reid Spencerf9d7a512004-11-14 21:58:33 +000098
99 // Check the member signature
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000100 if (!Hdr->checkSignature()) {
101 if (error)
102 *error = "invalid file member signature";
103 return 0;
104 }
Reid Spencerf9d7a512004-11-14 21:58:33 +0000105
106 // Convert and check the member name
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000107 // The empty name ( '/' and 15 blanks) is for a foreign (non-LLVM) symbol
108 // table. The special name "//" and 14 blanks is for a string table, used
Reid Spencerf9d7a512004-11-14 21:58:33 +0000109 // for long file names. This library doesn't generate either of those but
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000110 // it will accept them. If the name starts with #1/ and the remainder is
111 // digits, then those digits specify the length of the name that is
112 // stored immediately following the header. The special name
Gabor Greifa99be512007-07-05 17:07:56 +0000113 // __LLVM_SYM_TAB__ identifies the symbol table for LLVM bitcode.
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000114 // Anything else is a regular, short filename that is terminated with
Reid Spencerf9d7a512004-11-14 21:58:33 +0000115 // a '/' and blanks.
116
117 std::string pathname;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000118 switch (Hdr->name[0]) {
119 case '#':
120 if (Hdr->name[1] == '1' && Hdr->name[2] == '/') {
121 if (isdigit(Hdr->name[3])) {
122 unsigned len = atoi(&Hdr->name[3]);
Reid Spencer9a29db42004-11-20 07:29:40 +0000123 pathname.assign(At, len);
Reid Spencerdd95e8d2004-11-17 16:13:11 +0000124 At += len;
125 MemberSize -= len;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000126 flags |= ArchiveMember::HasLongFilenameFlag;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000127 } else {
128 if (error)
129 *error = "invalid long filename";
130 return 0;
131 }
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000132 } else if (Hdr->name[1] == '_' &&
Reid Spencer9a29db42004-11-20 07:29:40 +0000133 (0 == memcmp(Hdr->name, ARFILE_LLVM_SYMTAB_NAME, 16))) {
Reid Spencerf9d7a512004-11-14 21:58:33 +0000134 // The member is using a long file name (>15 chars) format.
135 // This format is standard for 4.4BSD and Mac OSX operating
136 // systems. LLVM uses it similarly. In this format, the
137 // remainder of the name field (after #1/) specifies the
138 // length of the file name which occupy the first bytes of
139 // the member's data. The pathname already has the #1/ stripped.
140 pathname.assign(ARFILE_LLVM_SYMTAB_NAME);
141 flags |= ArchiveMember::LLVMSymbolTableFlag;
142 }
143 break;
144 case '/':
145 if (Hdr->name[1]== '/') {
Reid Spencer9a29db42004-11-20 07:29:40 +0000146 if (0 == memcmp(Hdr->name, ARFILE_STRTAB_NAME, 16)) {
Reid Spencerf9d7a512004-11-14 21:58:33 +0000147 pathname.assign(ARFILE_STRTAB_NAME);
148 flags |= ArchiveMember::StringTableFlag;
149 } else {
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000150 if (error)
151 *error = "invalid string table name";
152 return 0;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000153 }
154 } else if (Hdr->name[1] == ' ') {
Reid Spencer9a29db42004-11-20 07:29:40 +0000155 if (0 == memcmp(Hdr->name, ARFILE_SVR4_SYMTAB_NAME, 16)) {
156 pathname.assign(ARFILE_SVR4_SYMTAB_NAME);
157 flags |= ArchiveMember::SVR4SymbolTableFlag;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000158 } else {
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000159 if (error)
160 *error = "invalid SVR4 symbol table name";
161 return 0;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000162 }
163 } else if (isdigit(Hdr->name[1])) {
164 unsigned index = atoi(&Hdr->name[1]);
165 if (index < strtab.length()) {
166 const char* namep = strtab.c_str() + index;
167 const char* endp = strtab.c_str() + strtab.length();
168 const char* p = namep;
169 const char* last_p = p;
170 while (p < endp) {
171 if (*p == '\n' && *last_p == '/') {
Reid Spencer9a29db42004-11-20 07:29:40 +0000172 pathname.assign(namep, last_p - namep);
Reid Spencerf9d7a512004-11-14 21:58:33 +0000173 flags |= ArchiveMember::HasLongFilenameFlag;
174 break;
175 }
176 last_p = p;
177 p++;
178 }
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000179 if (p >= endp) {
180 if (error)
181 *error = "missing name termiantor in string table";
182 return 0;
183 }
Reid Spencerf9d7a512004-11-14 21:58:33 +0000184 } else {
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000185 if (error)
186 *error = "name index beyond string table";
187 return 0;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000188 }
189 }
190 break;
Reid Spencer9a29db42004-11-20 07:29:40 +0000191 case '_':
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000192 if (Hdr->name[1] == '_' &&
Reid Spencer9a29db42004-11-20 07:29:40 +0000193 (0 == memcmp(Hdr->name, ARFILE_BSD4_SYMTAB_NAME, 16))) {
194 pathname.assign(ARFILE_BSD4_SYMTAB_NAME);
195 flags |= ArchiveMember::BSD4SymbolTableFlag;
Reid Spencer84b9ced2004-11-23 22:35:39 +0000196 break;
Reid Spencer9a29db42004-11-20 07:29:40 +0000197 }
Reid Spencer84b9ced2004-11-23 22:35:39 +0000198 /* FALL THROUGH */
Reid Spencerf9d7a512004-11-14 21:58:33 +0000199
200 default:
Reid Spencer9a29db42004-11-20 07:29:40 +0000201 char* slash = (char*) memchr(Hdr->name, '/', 16);
Reid Spencerf9d7a512004-11-14 21:58:33 +0000202 if (slash == 0)
Reid Spencerdd95e8d2004-11-17 16:13:11 +0000203 slash = Hdr->name + 16;
Reid Spencer9a29db42004-11-20 07:29:40 +0000204 pathname.assign(Hdr->name, slash - Hdr->name);
Reid Spencerf9d7a512004-11-14 21:58:33 +0000205 break;
206 }
207
Gabor Greifdb5565a2007-07-06 20:28:40 +0000208 // Determine if this is a bitcode file
Reid Spencer9a29db42004-11-20 07:29:40 +0000209 switch (sys::IdentifyFileType(At, 4)) {
Chris Lattnere07c15c2007-05-06 06:18:07 +0000210 case sys::Bitcode_FileType:
Gabor Greife75ca3d2007-07-06 13:38:17 +0000211 flags |= ArchiveMember::BitcodeFlag;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000212 break;
213 default:
Gabor Greife75ca3d2007-07-06 13:38:17 +0000214 flags &= ~ArchiveMember::BitcodeFlag;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000215 break;
216 }
217
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000218 // Instantiate the ArchiveMember to be filled
219 ArchiveMember* member = new ArchiveMember(this);
220
Reid Spencerf9d7a512004-11-14 21:58:33 +0000221 // Fill in fields of the ArchiveMember
Reid Spencerf9d7a512004-11-14 21:58:33 +0000222 member->parent = this;
Reid Spencerdd04df02005-07-07 23:21:43 +0000223 member->path.set(pathname);
Reid Spencerf9d7a512004-11-14 21:58:33 +0000224 member->info.fileSize = MemberSize;
225 member->info.modTime.fromEpochTime(atoi(Hdr->date));
Reid Spencer56127622004-12-29 01:20:24 +0000226 unsigned int mode;
227 sscanf(Hdr->mode, "%o", &mode);
228 member->info.mode = mode;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000229 member->info.user = atoi(Hdr->uid);
230 member->info.group = atoi(Hdr->gid);
231 member->flags = flags;
232 member->data = At;
233
234 return member;
235}
236
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000237bool
238Archive::checkSignature(std::string* error) {
Reid Spencerf9d7a512004-11-14 21:58:33 +0000239 // Check the magic string at file's header
Chris Lattner7f6b4472008-04-01 04:26:46 +0000240 if (mapfile->getBufferSize() < 8 || memcmp(base, ARFILE_MAGIC, 8)) {
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000241 if (error)
242 *error = "invalid signature for an archive file";
243 return false;
244 }
245 return true;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000246}
247
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000248// This function loads the entire archive and fully populates its ilist with
Reid Spencerf9d7a512004-11-14 21:58:33 +0000249// the members of the archive file. This is typically used in preparation for
250// editing the contents of the archive.
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000251bool
252Archive::loadArchive(std::string* error) {
Reid Spencerf9d7a512004-11-14 21:58:33 +0000253
254 // Set up parsing
255 members.clear();
256 symTab.clear();
257 const char *At = base;
Chris Lattner7f6b4472008-04-01 04:26:46 +0000258 const char *End = mapfile->getBufferEnd();
Reid Spencerf9d7a512004-11-14 21:58:33 +0000259
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000260 if (!checkSignature(error))
261 return false;
262
Reid Spencerf9d7a512004-11-14 21:58:33 +0000263 At += 8; // Skip the magic string.
264
265 bool seenSymbolTable = false;
266 bool foundFirstFile = false;
267 while (At < End) {
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000268 // parse the member header
Reid Spencerf9d7a512004-11-14 21:58:33 +0000269 const char* Save = At;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000270 ArchiveMember* mbr = parseMemberHeader(At, End, error);
271 if (!mbr)
272 return false;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000273
274 // check if this is the foreign symbol table
Reid Spencer9a29db42004-11-20 07:29:40 +0000275 if (mbr->isSVR4SymbolTable() || mbr->isBSD4SymbolTable()) {
Reid Spencer4a980d12004-11-16 06:47:19 +0000276 // We just save this but don't do anything special
277 // with it. It doesn't count as the "first file".
Reid Spencer9a29db42004-11-20 07:29:40 +0000278 if (foreignST) {
279 // What? Multiple foreign symbol tables? Just chuck it
280 // and retain the last one found.
281 delete foreignST;
282 }
Reid Spencer4a980d12004-11-16 06:47:19 +0000283 foreignST = mbr;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000284 At += mbr->getSize();
Reid Spencer6405c9e2004-11-19 17:08:00 +0000285 if ((intptr_t(At) & 1) == 1)
Reid Spencerf9d7a512004-11-14 21:58:33 +0000286 At++;
287 } else if (mbr->isStringTable()) {
Reid Spencer4a980d12004-11-16 06:47:19 +0000288 // Simply suck the entire string table into a string
289 // variable. This will be used to get the names of the
290 // members that use the "/ddd" format for their names
291 // (SVR4 style long names).
Reid Spencer9a29db42004-11-20 07:29:40 +0000292 strtab.assign(At, mbr->getSize());
Reid Spencerf9d7a512004-11-14 21:58:33 +0000293 At += mbr->getSize();
Reid Spencer6405c9e2004-11-19 17:08:00 +0000294 if ((intptr_t(At) & 1) == 1)
Reid Spencerf9d7a512004-11-14 21:58:33 +0000295 At++;
296 delete mbr;
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000297 } else if (mbr->isLLVMSymbolTable()) {
Reid Spencer4a980d12004-11-16 06:47:19 +0000298 // This is the LLVM symbol table for the archive. If we've seen it
299 // already, its an error. Otherwise, parse the symbol table and move on.
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000300 if (seenSymbolTable) {
301 if (error)
302 *error = "invalid archive: multiple symbol tables";
303 return false;
304 }
305 if (!parseSymbolTable(mbr->getData(), mbr->getSize(), error))
306 return false;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000307 seenSymbolTable = true;
308 At += mbr->getSize();
Reid Spencer6405c9e2004-11-19 17:08:00 +0000309 if ((intptr_t(At) & 1) == 1)
Reid Spencerf9d7a512004-11-14 21:58:33 +0000310 At++;
Reid Spencer4a980d12004-11-16 06:47:19 +0000311 delete mbr; // We don't need this member in the list of members.
Reid Spencerf9d7a512004-11-14 21:58:33 +0000312 } else {
Reid Spencer4a980d12004-11-16 06:47:19 +0000313 // This is just a regular file. If its the first one, save its offset.
314 // Otherwise just push it on the list and move on to the next file.
Reid Spencerf9d7a512004-11-14 21:58:33 +0000315 if (!foundFirstFile) {
316 firstFileOffset = Save - base;
317 foundFirstFile = true;
318 }
319 members.push_back(mbr);
320 At += mbr->getSize();
Reid Spencer6405c9e2004-11-19 17:08:00 +0000321 if ((intptr_t(At) & 1) == 1)
Reid Spencerf9d7a512004-11-14 21:58:33 +0000322 At++;
323 }
324 }
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000325 return true;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000326}
327
328// Open and completely load the archive file.
329Archive*
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000330Archive::OpenAndLoad(const sys::Path& file, std::string* ErrorMessage)
331{
Reid Spencer0ff2d312006-08-24 23:45:08 +0000332 std::auto_ptr<Archive> result ( new Archive(file));
333 if (result->mapToMemory(ErrorMessage))
334 return 0;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000335 if (!result->loadArchive(ErrorMessage))
Reid Spencer5af46882004-12-13 02:59:03 +0000336 return 0;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000337 return result.release();
Reid Spencerf9d7a512004-11-14 21:58:33 +0000338}
339
Gabor Greifa99be512007-07-05 17:07:56 +0000340// Get all the bitcode modules from the archive
Reid Spencerf9d7a512004-11-14 21:58:33 +0000341bool
342Archive::getAllModules(std::vector<Module*>& Modules, std::string* ErrMessage) {
343
344 for (iterator I=begin(), E=end(); I != E; ++I) {
Gabor Greife75ca3d2007-07-06 13:38:17 +0000345 if (I->isBitcode()) {
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000346 std::string FullMemberName = archPath.toString() +
Reid Spencer1fce0912004-12-11 00:14:15 +0000347 "(" + I->getPath().toString() + ")";
Chris Lattnerc1d56242007-05-06 09:28:33 +0000348 MemoryBuffer *Buffer =
349 MemoryBuffer::getNewMemBuffer(I->getSize(), FullMemberName.c_str());
350 memcpy((char*)Buffer->getBufferStart(), I->getData(), I->getSize());
Chris Lattnere07c15c2007-05-06 06:18:07 +0000351
Chris Lattnerc1d56242007-05-06 09:28:33 +0000352 Module *M = ParseBitcodeFile(Buffer, ErrMessage);
353 delete Buffer;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000354 if (!M)
355 return true;
356
357 Modules.push_back(M);
358 }
359 }
Brian Gaeke2c61d7b2003-11-16 23:08:48 +0000360 return false;
361}
Chris Lattner968cfd02003-04-19 21:45:34 +0000362
Reid Spencerf9d7a512004-11-14 21:58:33 +0000363// Load just the symbol table from the archive file
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000364bool
365Archive::loadSymbolTable(std::string* ErrorMsg) {
Chris Lattner968cfd02003-04-19 21:45:34 +0000366
Reid Spencerf9d7a512004-11-14 21:58:33 +0000367 // Set up parsing
368 members.clear();
369 symTab.clear();
370 const char *At = base;
Chris Lattner7f6b4472008-04-01 04:26:46 +0000371 const char *End = mapfile->getBufferEnd();
Chris Lattner968cfd02003-04-19 21:45:34 +0000372
Reid Spencerf9d7a512004-11-14 21:58:33 +0000373 // Make sure we're dealing with an archive
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000374 if (!checkSignature(ErrorMsg))
375 return false;
Chris Lattner968cfd02003-04-19 21:45:34 +0000376
Reid Spencerf9d7a512004-11-14 21:58:33 +0000377 At += 8; // Skip signature
378
379 // Parse the first file member header
380 const char* FirstFile = At;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000381 ArchiveMember* mbr = parseMemberHeader(At, End, ErrorMsg);
382 if (!mbr)
383 return false;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000384
Reid Spencer9a29db42004-11-20 07:29:40 +0000385 if (mbr->isSVR4SymbolTable() || mbr->isBSD4SymbolTable()) {
Reid Spencerf9d7a512004-11-14 21:58:33 +0000386 // Skip the foreign symbol table, we don't do anything with it
387 At += mbr->getSize();
Reid Spencer6405c9e2004-11-19 17:08:00 +0000388 if ((intptr_t(At) & 1) == 1)
Reid Spencerb3231132004-11-15 01:40:20 +0000389 At++;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000390 delete mbr;
391
Reid Spencerb3231132004-11-15 01:40:20 +0000392 // Read the next one
Reid Spencerf9d7a512004-11-14 21:58:33 +0000393 FirstFile = At;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000394 mbr = parseMemberHeader(At, End, ErrorMsg);
395 if (!mbr) {
396 delete mbr;
397 return false;
398 }
Reid Spencerb3231132004-11-15 01:40:20 +0000399 }
400
401 if (mbr->isStringTable()) {
402 // Process the string table entry
Reid Spencer9a29db42004-11-20 07:29:40 +0000403 strtab.assign((const char*)mbr->getData(), mbr->getSize());
Reid Spencerb3231132004-11-15 01:40:20 +0000404 At += mbr->getSize();
Reid Spencer6405c9e2004-11-19 17:08:00 +0000405 if ((intptr_t(At) & 1) == 1)
Reid Spencerb3231132004-11-15 01:40:20 +0000406 At++;
407 delete mbr;
408 // Get the next one
409 FirstFile = At;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000410 mbr = parseMemberHeader(At, End, ErrorMsg);
411 if (!mbr) {
412 delete mbr;
413 return false;
414 }
Chris Lattner968cfd02003-04-19 21:45:34 +0000415 }
416
Reid Spencerf9d7a512004-11-14 21:58:33 +0000417 // See if its the symbol table
418 if (mbr->isLLVMSymbolTable()) {
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000419 if (!parseSymbolTable(mbr->getData(), mbr->getSize(), ErrorMsg)) {
420 delete mbr;
421 return false;
422 }
423
Reid Spencer8dde18f2004-11-28 03:13:02 +0000424 At += mbr->getSize();
Reid Spencer6405c9e2004-11-19 17:08:00 +0000425 if ((intptr_t(At) & 1) == 1)
Reid Spencer8dde18f2004-11-28 03:13:02 +0000426 At++;
Reid Spencerffb4d622006-05-12 17:56:20 +0000427 delete mbr;
428 // Can't be any more symtab headers so just advance
Reid Spencer8dde18f2004-11-28 03:13:02 +0000429 FirstFile = At;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000430 } else {
431 // There's no symbol table in the file. We have to rebuild it from scratch
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000432 // because the intent of this method is to get the symbol table loaded so
433 // it can be searched efficiently.
Reid Spencerf9d7a512004-11-14 21:58:33 +0000434 // Add the member to the members list
435 members.push_back(mbr);
436 }
437
438 firstFileOffset = FirstFile - base;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000439 return true;
Chris Lattner968cfd02003-04-19 21:45:34 +0000440}
441
Reid Spencerf9d7a512004-11-14 21:58:33 +0000442// Open the archive and load just the symbol tables
443Archive*
Reid Spencer5af46882004-12-13 02:59:03 +0000444Archive::OpenAndLoadSymbols(const sys::Path& file, std::string* ErrorMessage) {
Reid Spencer0ff2d312006-08-24 23:45:08 +0000445 std::auto_ptr<Archive> result ( new Archive(file) );
446 if (result->mapToMemory(ErrorMessage))
447 return 0;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000448 if (!result->loadSymbolTable(ErrorMessage))
Reid Spencer5af46882004-12-13 02:59:03 +0000449 return 0;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000450 return result.release();
Reid Spencerf9d7a512004-11-14 21:58:33 +0000451}
Chris Lattner968cfd02003-04-19 21:45:34 +0000452
Reid Spencerf9d7a512004-11-14 21:58:33 +0000453// Look up one symbol in the symbol table and return a ModuleProvider for the
454// module that defines that symbol.
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000455ModuleProvider*
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000456Archive::findModuleDefiningSymbol(const std::string& symbol,
457 std::string* ErrMsg) {
Reid Spencerf9d7a512004-11-14 21:58:33 +0000458 SymTabType::iterator SI = symTab.find(symbol);
459 if (SI == symTab.end())
460 return 0;
461
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000462 // The symbol table was previously constructed assuming that the members were
Reid Spencerf9d7a512004-11-14 21:58:33 +0000463 // written without the symbol table header. Because VBR encoding is used, the
464 // values could not be adjusted to account for the offset of the symbol table
465 // because that could affect the size of the symbol table due to VBR encoding.
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000466 // We now have to account for this by adjusting the offset by the size of the
Reid Spencerf9d7a512004-11-14 21:58:33 +0000467 // symbol table and its header.
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000468 unsigned fileOffset =
Reid Spencerf9d7a512004-11-14 21:58:33 +0000469 SI->second + // offset in symbol-table-less file
470 firstFileOffset; // add offset to first "real" file in archive
471
472 // See if the module is already loaded
473 ModuleMap::iterator MI = modules.find(fileOffset);
474 if (MI != modules.end())
475 return MI->second.first;
476
477 // Module hasn't been loaded yet, we need to load it
478 const char* modptr = base + fileOffset;
Chris Lattner7f6b4472008-04-01 04:26:46 +0000479 ArchiveMember* mbr = parseMemberHeader(modptr, mapfile->getBufferEnd(),
480 ErrMsg);
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000481 if (!mbr)
Reid Spencer0b5a5042006-08-25 17:43:11 +0000482 return 0;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000483
Gabor Greifa99be512007-07-05 17:07:56 +0000484 // Now, load the bitcode module to get the ModuleProvider
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000485 std::string FullMemberName = archPath.toString() + "(" +
Reid Spencer1fce0912004-12-11 00:14:15 +0000486 mbr->getPath().toString() + ")";
Chris Lattnerc1d56242007-05-06 09:28:33 +0000487 MemoryBuffer *Buffer =MemoryBuffer::getNewMemBuffer(mbr->getSize(),
488 FullMemberName.c_str());
489 memcpy((char*)Buffer->getBufferStart(), mbr->getData(), mbr->getSize());
490
491 ModuleProvider *mp = getBitcodeModuleProvider(Buffer, ErrMsg);
Reid Spencer0b5a5042006-08-25 17:43:11 +0000492 if (!mp)
493 return 0;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000494
Reid Spencer9a29db42004-11-20 07:29:40 +0000495 modules.insert(std::make_pair(fileOffset, std::make_pair(mp, mbr)));
Reid Spencerf9d7a512004-11-14 21:58:33 +0000496
497 return mp;
498}
499
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000500// Look up multiple symbols in the symbol table and return a set of
Reid Spencerf9d7a512004-11-14 21:58:33 +0000501// ModuleProviders that define those symbols.
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000502bool
Reid Spencer7783e8a2004-11-19 03:18:22 +0000503Archive::findModulesDefiningSymbols(std::set<std::string>& symbols,
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000504 std::set<ModuleProvider*>& result,
Chris Lattnerf2e292c2007-02-07 21:41:02 +0000505 std::string* error) {
Reid Spencercd5561a2006-12-15 19:44:51 +0000506 if (!mapfile || !base) {
507 if (error)
508 *error = "Empty archive invalid for finding modules defining symbols";
509 return false;
510 }
511
Reid Spencer766b7932004-11-15 01:20:11 +0000512 if (symTab.empty()) {
513 // We don't have a symbol table, so we must build it now but lets also
514 // make sure that we populate the modules table as we do this to ensure
515 // that we don't load them twice when findModuleDefiningSymbol is called
516 // below.
517
518 // Get a pointer to the first file
Chris Lattner7f6b4472008-04-01 04:26:46 +0000519 const char* At = base + firstFileOffset;
520 const char* End = mapfile->getBufferEnd();
Reid Spencer766b7932004-11-15 01:20:11 +0000521
522 while ( At < End) {
523 // Compute the offset to be put in the symbol table
524 unsigned offset = At - base - firstFileOffset;
525
526 // Parse the file's header
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000527 ArchiveMember* mbr = parseMemberHeader(At, End, error);
528 if (!mbr)
529 return false;
Reid Spencer766b7932004-11-15 01:20:11 +0000530
531 // If it contains symbols
Gabor Greife75ca3d2007-07-06 13:38:17 +0000532 if (mbr->isBitcode()) {
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000533 // Get the symbols
Reid Spencer766b7932004-11-15 01:20:11 +0000534 std::vector<std::string> symbols;
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000535 std::string FullMemberName = archPath.toString() + "(" +
Reid Spencer1fce0912004-12-11 00:14:15 +0000536 mbr->getPath().toString() + ")";
Chris Lattnerf2e292c2007-02-07 21:41:02 +0000537 ModuleProvider* MP =
Gabor Greifa99be512007-07-05 17:07:56 +0000538 GetBitcodeSymbols((const unsigned char*)At, mbr->getSize(),
539 FullMemberName, symbols, error);
Reid Spencer766b7932004-11-15 01:20:11 +0000540
541 if (MP) {
542 // Insert the module's symbols into the symbol table
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000543 for (std::vector<std::string>::iterator I = symbols.begin(),
Reid Spencer766b7932004-11-15 01:20:11 +0000544 E=symbols.end(); I != E; ++I ) {
Reid Spencer9a29db42004-11-20 07:29:40 +0000545 symTab.insert(std::make_pair(*I, offset));
Reid Spencer766b7932004-11-15 01:20:11 +0000546 }
547 // Insert the ModuleProvider and the ArchiveMember into the table of
548 // modules.
Reid Spencer9a29db42004-11-20 07:29:40 +0000549 modules.insert(std::make_pair(offset, std::make_pair(MP, mbr)));
Reid Spencer766b7932004-11-15 01:20:11 +0000550 } else {
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000551 if (error)
Gabor Greifa99be512007-07-05 17:07:56 +0000552 *error = "Can't parse bitcode member: " +
Reid Spencer0b5a5042006-08-25 17:43:11 +0000553 mbr->getPath().toString() + ": " + *error;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000554 delete mbr;
555 return false;
Reid Spencer766b7932004-11-15 01:20:11 +0000556 }
557 }
Reid Spencerb3231132004-11-15 01:40:20 +0000558
559 // Go to the next file location
560 At += mbr->getSize();
Reid Spencer6405c9e2004-11-19 17:08:00 +0000561 if ((intptr_t(At) & 1) == 1)
Reid Spencerb3231132004-11-15 01:40:20 +0000562 At++;
Reid Spencer766b7932004-11-15 01:20:11 +0000563 }
564 }
565
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000566 // At this point we have a valid symbol table (one way or another) so we
Reid Spencer766b7932004-11-15 01:20:11 +0000567 // just use it to quickly find the symbols requested.
568
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000569 for (std::set<std::string>::iterator I=symbols.begin(),
Reid Spencer7783e8a2004-11-19 03:18:22 +0000570 E=symbols.end(); I != E;) {
571 // See if this symbol exists
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000572 ModuleProvider* mp = findModuleDefiningSymbol(*I,error);
Reid Spencerf9d7a512004-11-14 21:58:33 +0000573 if (mp) {
Reid Spencer7783e8a2004-11-19 03:18:22 +0000574 // The symbol exists, insert the ModuleProvider into our result,
575 // duplicates wil be ignored
Reid Spencer766b7932004-11-15 01:20:11 +0000576 result.insert(mp);
Reid Spencer7783e8a2004-11-19 03:18:22 +0000577
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000578 // Remove the symbol now that its been resolved, being careful to
Reid Spencer57646ec2004-11-19 03:44:10 +0000579 // post-increment the iterator.
580 symbols.erase(I++);
Reid Spencer7783e8a2004-11-19 03:18:22 +0000581 } else {
582 ++I;
Chris Lattner968cfd02003-04-19 21:45:34 +0000583 }
Reid Spencerf9d7a512004-11-14 21:58:33 +0000584 }
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000585 return true;
Chris Lattner968cfd02003-04-19 21:45:34 +0000586}
Reid Spencereaa06bb2005-02-26 22:00:32 +0000587
Gabor Greifa99be512007-07-05 17:07:56 +0000588bool Archive::isBitcodeArchive() {
Chris Lattner7c439922005-09-23 06:22:58 +0000589 // Make sure the symTab has been loaded. In most cases this should have been
590 // done when the archive was constructed, but still, this is just in case.
Dan Gohman30359592008-01-29 13:02:09 +0000591 if (symTab.empty())
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000592 if (!loadSymbolTable(0))
593 return false;
Reid Spencereaa06bb2005-02-26 22:00:32 +0000594
Chris Lattner7c439922005-09-23 06:22:58 +0000595 // Now that we know it's been loaded, return true
596 // if it has a size
597 if (symTab.size()) return true;
Reid Spencereaa06bb2005-02-26 22:00:32 +0000598
Gabor Greifa99be512007-07-05 17:07:56 +0000599 // We still can't be sure it isn't a bitcode archive
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000600 if (!loadArchive(0))
601 return false;
Reid Spencereaa06bb2005-02-26 22:00:32 +0000602
603 std::vector<Module *> Modules;
604 std::string ErrorMessage;
605
Gabor Greifa99be512007-07-05 17:07:56 +0000606 // Scan the archive, trying to load a bitcode member. We only load one to
Chris Lattner7c439922005-09-23 06:22:58 +0000607 // see if this works.
608 for (iterator I = begin(), E = end(); I != E; ++I) {
Gabor Greife75ca3d2007-07-06 13:38:17 +0000609 if (!I->isBitcode())
Chris Lattner7c439922005-09-23 06:22:58 +0000610 continue;
611
612 std::string FullMemberName =
613 archPath.toString() + "(" + I->getPath().toString() + ")";
Chris Lattnerc1d56242007-05-06 09:28:33 +0000614
615 MemoryBuffer *Buffer =
616 MemoryBuffer::getNewMemBuffer(I->getSize(), FullMemberName.c_str());
617 memcpy((char*)Buffer->getBufferStart(), I->getData(), I->getSize());
618 Module *M = ParseBitcodeFile(Buffer);
619 delete Buffer;
Chris Lattner7c439922005-09-23 06:22:58 +0000620 if (!M)
Gabor Greifa99be512007-07-05 17:07:56 +0000621 return false; // Couldn't parse bitcode, not a bitcode archive.
Chris Lattner7c439922005-09-23 06:22:58 +0000622 delete M;
623 return true;
624 }
625
626 return false;
Reid Spencereaa06bb2005-02-26 22:00:32 +0000627}