blob: 209db77ee0093ac244c0b18c3f5969a7281e5d3c [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//
Misha Brukman2b37d7c2005-04-21 21:13:18 +00005// This file was developed by Reid Spencer and is distributed under the
Reid Spencerf9d7a512004-11-14 21:58:33 +00006// University of Illinois Open Source 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//
Reid Spencerf9d7a512004-11-14 21:58:33 +000010// Builds up standard unix archive files (.a) containing LLVM bytecode.
Chris Lattner968cfd02003-04-19 21:45:34 +000011//
12//===----------------------------------------------------------------------===//
13
Reid Spencerf9d7a512004-11-14 21:58:33 +000014#include "ArchiveInternals.h"
Chris Lattner968cfd02003-04-19 21:45:34 +000015#include "llvm/Bytecode/Reader.h"
Chris Lattnere07c15c2007-05-06 06:18:07 +000016#include "llvm/Bitcode/ReaderWriter.h"
Chris Lattnerf2e292c2007-02-07 21:41:02 +000017#include "llvm/Support/Compressor.h"
Chris Lattnere07c15c2007-05-06 06:18:07 +000018#include "llvm/Support/MemoryBuffer.h"
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
Chris Lattnere07c15c2007-05-06 06:18:07 +000022static bool Bitcode = false;
23
Reid Spencerf9d7a512004-11-14 21:58:33 +000024/// Read a variable-bit-rate encoded unsigned integer
Reid Spencer8d8a7ff2006-07-07 20:56:50 +000025inline unsigned readInteger(const char*&At, const char*End){
Reid Spencerf9d7a512004-11-14 21:58:33 +000026 unsigned Shift = 0;
27 unsigned Result = 0;
Misha Brukman2b37d7c2005-04-21 21:13:18 +000028
Reid Spencerf9d7a512004-11-14 21:58:33 +000029 do {
Misha Brukman2b37d7c2005-04-21 21:13:18 +000030 if (At == End)
Reid Spencer8d8a7ff2006-07-07 20:56:50 +000031 return Result;
Reid Spencerf9d7a512004-11-14 21:58:33 +000032 Result |= (unsigned)((*At++) & 0x7F) << Shift;
33 Shift += 7;
34 } while (At[-1] & 0x80);
35 return Result;
Chris Lattner968cfd02003-04-19 21:45:34 +000036}
37
Reid Spencerf9d7a512004-11-14 21:58:33 +000038// Completely parse the Archive's symbol table and populate symTab member var.
Reid Spencer8d8a7ff2006-07-07 20:56:50 +000039bool
40Archive::parseSymbolTable(const void* data, unsigned size, std::string* error) {
Reid Spencerf9d7a512004-11-14 21:58:33 +000041 const char* At = (const char*) data;
42 const char* End = At + size;
43 while (At < End) {
44 unsigned offset = readInteger(At, End);
Reid Spencer8d8a7ff2006-07-07 20:56:50 +000045 if (At == End) {
46 if (error)
47 *error = "Ran out of data reading vbr_uint for symtab offset!";
48 return false;
49 }
Reid Spencerf9d7a512004-11-14 21:58:33 +000050 unsigned length = readInteger(At, End);
Reid Spencer8d8a7ff2006-07-07 20:56:50 +000051 if (At == End) {
52 if (error)
53 *error = "Ran out of data reading vbr_uint for symtab length!";
54 return false;
55 }
56 if (At + length > End) {
57 if (error)
58 *error = "Malformed symbol table: length not consistent with size";
59 return false;
60 }
Reid Spencerf9d7a512004-11-14 21:58:33 +000061 // we don't care if it can't be inserted (duplicate entry)
Reid Spencer9a29db42004-11-20 07:29:40 +000062 symTab.insert(std::make_pair(std::string(At, length), offset));
Reid Spencerf9d7a512004-11-14 21:58:33 +000063 At += length;
64 }
65 symTabSize = size;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +000066 return true;
Chris Lattner968cfd02003-04-19 21:45:34 +000067}
68
Reid Spencerf9d7a512004-11-14 21:58:33 +000069// This member parses an ArchiveMemberHeader that is presumed to be pointed to
70// by At. The At pointer is updated to the byte just after the header, which
Misha Brukman2b37d7c2005-04-21 21:13:18 +000071// can be variable in size.
Reid Spencerf9d7a512004-11-14 21:58:33 +000072ArchiveMember*
Reid Spencer8d8a7ff2006-07-07 20:56:50 +000073Archive::parseMemberHeader(const char*& At, const char* End, std::string* error)
74{
Reid Spencercd5561a2006-12-15 19:44:51 +000075 if (At + sizeof(ArchiveMemberHeader) >= End) {
76 if (error)
77 *error = "Unexpected end of file";
78 return 0;
79 }
Reid Spencerf9d7a512004-11-14 21:58:33 +000080
81 // Cast archive member header
82 ArchiveMemberHeader* Hdr = (ArchiveMemberHeader*)At;
83 At += sizeof(ArchiveMemberHeader);
84
Misha Brukman2b37d7c2005-04-21 21:13:18 +000085 // Extract the size and determine if the file is
Reid Spencerf9d7a512004-11-14 21:58:33 +000086 // compressed or not (negative length).
87 int flags = 0;
88 int MemberSize = atoi(Hdr->size);
89 if (MemberSize < 0) {
90 flags |= ArchiveMember::CompressedFlag;
91 MemberSize = -MemberSize;
92 }
93
94 // Check the size of the member for sanity
Reid Spencer8d8a7ff2006-07-07 20:56:50 +000095 if (At + MemberSize > End) {
96 if (error)
97 *error = "invalid member length in archive file";
98 return 0;
99 }
Reid Spencerf9d7a512004-11-14 21:58:33 +0000100
101 // Check the member signature
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000102 if (!Hdr->checkSignature()) {
103 if (error)
104 *error = "invalid file member signature";
105 return 0;
106 }
Reid Spencerf9d7a512004-11-14 21:58:33 +0000107
108 // Convert and check the member name
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000109 // The empty name ( '/' and 15 blanks) is for a foreign (non-LLVM) symbol
110 // table. The special name "//" and 14 blanks is for a string table, used
Reid Spencerf9d7a512004-11-14 21:58:33 +0000111 // for long file names. This library doesn't generate either of those but
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000112 // it will accept them. If the name starts with #1/ and the remainder is
113 // digits, then those digits specify the length of the name that is
114 // stored immediately following the header. The special name
115 // __LLVM_SYM_TAB__ identifies the symbol table for LLVM bytecode.
116 // Anything else is a regular, short filename that is terminated with
Reid Spencerf9d7a512004-11-14 21:58:33 +0000117 // a '/' and blanks.
118
119 std::string pathname;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000120 switch (Hdr->name[0]) {
121 case '#':
122 if (Hdr->name[1] == '1' && Hdr->name[2] == '/') {
123 if (isdigit(Hdr->name[3])) {
124 unsigned len = atoi(&Hdr->name[3]);
Reid Spencer9a29db42004-11-20 07:29:40 +0000125 pathname.assign(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
210 // Determine if this is a bytecode 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:
Reid Spencer20c34892007-04-04 06:31:04 +0000213 case sys::Bytecode_FileType:
Reid Spencerf9d7a512004-11-14 21:58:33 +0000214 flags |= ArchiveMember::BytecodeFlag;
215 break;
Reid Spencer20c34892007-04-04 06:31:04 +0000216 case sys::CompressedBytecode_FileType:
Reid Spencerf9d7a512004-11-14 21:58:33 +0000217 flags |= ArchiveMember::CompressedBytecodeFlag;
218 flags &= ~ArchiveMember::CompressedFlag;
219 break;
220 default:
221 flags &= ~(ArchiveMember::BytecodeFlag|
222 ArchiveMember::CompressedBytecodeFlag);
223 break;
224 }
225
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000226 // Instantiate the ArchiveMember to be filled
227 ArchiveMember* member = new ArchiveMember(this);
228
Reid Spencerf9d7a512004-11-14 21:58:33 +0000229 // Fill in fields of the ArchiveMember
230 member->next = 0;
231 member->prev = 0;
232 member->parent = this;
Reid Spencerdd04df02005-07-07 23:21:43 +0000233 member->path.set(pathname);
Reid Spencerf9d7a512004-11-14 21:58:33 +0000234 member->info.fileSize = MemberSize;
235 member->info.modTime.fromEpochTime(atoi(Hdr->date));
Reid Spencer56127622004-12-29 01:20:24 +0000236 unsigned int mode;
237 sscanf(Hdr->mode, "%o", &mode);
238 member->info.mode = mode;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000239 member->info.user = atoi(Hdr->uid);
240 member->info.group = atoi(Hdr->gid);
241 member->flags = flags;
242 member->data = At;
243
244 return member;
245}
246
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000247bool
248Archive::checkSignature(std::string* error) {
Reid Spencerf9d7a512004-11-14 21:58:33 +0000249 // Check the magic string at file's header
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000250 if (mapfile->size() < 8 || memcmp(base, ARFILE_MAGIC, 8)) {
251 if (error)
252 *error = "invalid signature for an archive file";
253 return false;
254 }
255 return true;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000256}
257
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000258// This function loads the entire archive and fully populates its ilist with
Reid Spencerf9d7a512004-11-14 21:58:33 +0000259// the members of the archive file. This is typically used in preparation for
260// editing the contents of the archive.
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000261bool
262Archive::loadArchive(std::string* error) {
Reid Spencerf9d7a512004-11-14 21:58:33 +0000263
264 // Set up parsing
265 members.clear();
266 symTab.clear();
267 const char *At = base;
268 const char *End = base + mapfile->size();
269
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000270 if (!checkSignature(error))
271 return false;
272
Reid Spencerf9d7a512004-11-14 21:58:33 +0000273 At += 8; // Skip the magic string.
274
275 bool seenSymbolTable = false;
276 bool foundFirstFile = false;
277 while (At < End) {
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000278 // parse the member header
Reid Spencerf9d7a512004-11-14 21:58:33 +0000279 const char* Save = At;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000280 ArchiveMember* mbr = parseMemberHeader(At, End, error);
281 if (!mbr)
282 return false;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000283
284 // check if this is the foreign symbol table
Reid Spencer9a29db42004-11-20 07:29:40 +0000285 if (mbr->isSVR4SymbolTable() || mbr->isBSD4SymbolTable()) {
Reid Spencer4a980d12004-11-16 06:47:19 +0000286 // We just save this but don't do anything special
287 // with it. It doesn't count as the "first file".
Reid Spencer9a29db42004-11-20 07:29:40 +0000288 if (foreignST) {
289 // What? Multiple foreign symbol tables? Just chuck it
290 // and retain the last one found.
291 delete foreignST;
292 }
Reid Spencer4a980d12004-11-16 06:47:19 +0000293 foreignST = mbr;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000294 At += mbr->getSize();
Reid Spencer6405c9e2004-11-19 17:08:00 +0000295 if ((intptr_t(At) & 1) == 1)
Reid Spencerf9d7a512004-11-14 21:58:33 +0000296 At++;
297 } else if (mbr->isStringTable()) {
Reid Spencer4a980d12004-11-16 06:47:19 +0000298 // Simply suck the entire string table into a string
299 // variable. This will be used to get the names of the
300 // members that use the "/ddd" format for their names
301 // (SVR4 style long names).
Reid Spencer9a29db42004-11-20 07:29:40 +0000302 strtab.assign(At, mbr->getSize());
Reid Spencerf9d7a512004-11-14 21:58:33 +0000303 At += mbr->getSize();
Reid Spencer6405c9e2004-11-19 17:08:00 +0000304 if ((intptr_t(At) & 1) == 1)
Reid Spencerf9d7a512004-11-14 21:58:33 +0000305 At++;
306 delete mbr;
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000307 } else if (mbr->isLLVMSymbolTable()) {
Reid Spencer4a980d12004-11-16 06:47:19 +0000308 // This is the LLVM symbol table for the archive. If we've seen it
309 // already, its an error. Otherwise, parse the symbol table and move on.
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000310 if (seenSymbolTable) {
311 if (error)
312 *error = "invalid archive: multiple symbol tables";
313 return false;
314 }
315 if (!parseSymbolTable(mbr->getData(), mbr->getSize(), error))
316 return false;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000317 seenSymbolTable = true;
318 At += mbr->getSize();
Reid Spencer6405c9e2004-11-19 17:08:00 +0000319 if ((intptr_t(At) & 1) == 1)
Reid Spencerf9d7a512004-11-14 21:58:33 +0000320 At++;
Reid Spencer4a980d12004-11-16 06:47:19 +0000321 delete mbr; // We don't need this member in the list of members.
Reid Spencerf9d7a512004-11-14 21:58:33 +0000322 } else {
Reid Spencer4a980d12004-11-16 06:47:19 +0000323 // This is just a regular file. If its the first one, save its offset.
324 // Otherwise just push it on the list and move on to the next file.
Reid Spencerf9d7a512004-11-14 21:58:33 +0000325 if (!foundFirstFile) {
326 firstFileOffset = Save - base;
327 foundFirstFile = true;
328 }
329 members.push_back(mbr);
330 At += mbr->getSize();
Reid Spencer6405c9e2004-11-19 17:08:00 +0000331 if ((intptr_t(At) & 1) == 1)
Reid Spencerf9d7a512004-11-14 21:58:33 +0000332 At++;
333 }
334 }
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000335 return true;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000336}
337
338// Open and completely load the archive file.
339Archive*
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000340Archive::OpenAndLoad(const sys::Path& file, std::string* ErrorMessage)
341{
Reid Spencer0ff2d312006-08-24 23:45:08 +0000342 std::auto_ptr<Archive> result ( new Archive(file));
343 if (result->mapToMemory(ErrorMessage))
344 return 0;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000345 if (!result->loadArchive(ErrorMessage))
Reid Spencer5af46882004-12-13 02:59:03 +0000346 return 0;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000347 return result.release();
Reid Spencerf9d7a512004-11-14 21:58:33 +0000348}
349
350// Get all the bytecode modules from the archive
351bool
352Archive::getAllModules(std::vector<Module*>& Modules, std::string* ErrMessage) {
353
354 for (iterator I=begin(), E=end(); I != E; ++I) {
355 if (I->isBytecode() || I->isCompressedBytecode()) {
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000356 std::string FullMemberName = archPath.toString() +
Reid Spencer1fce0912004-12-11 00:14:15 +0000357 "(" + I->getPath().toString() + ")";
Chris Lattnere07c15c2007-05-06 06:18:07 +0000358 Module *M;
359
360 if (Bitcode) {
361 MemoryBuffer *Buffer =
362 MemoryBuffer::getNewMemBuffer(I->getSize(), FullMemberName.c_str());
363 memcpy((char*)Buffer->getBufferStart(), I->getData(), I->getSize());
364
365 M = ParseBitcodeFile(Buffer, ErrMessage);
366 delete Buffer;
367 } else {
368 M = ParseBytecodeBuffer((const unsigned char*)I->getData(),
369 I->getSize(), FullMemberName,
370 Compressor::decompressToNewBuffer,
371 ErrMessage);
372 }
Reid Spencerf9d7a512004-11-14 21:58:33 +0000373 if (!M)
374 return true;
375
376 Modules.push_back(M);
377 }
378 }
Brian Gaeke2c61d7b2003-11-16 23:08:48 +0000379 return false;
380}
Chris Lattner968cfd02003-04-19 21:45:34 +0000381
Reid Spencerf9d7a512004-11-14 21:58:33 +0000382// Load just the symbol table from the archive file
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000383bool
384Archive::loadSymbolTable(std::string* ErrorMsg) {
Chris Lattner968cfd02003-04-19 21:45:34 +0000385
Reid Spencerf9d7a512004-11-14 21:58:33 +0000386 // Set up parsing
387 members.clear();
388 symTab.clear();
389 const char *At = base;
390 const char *End = base + mapfile->size();
Chris Lattner968cfd02003-04-19 21:45:34 +0000391
Reid Spencerf9d7a512004-11-14 21:58:33 +0000392 // Make sure we're dealing with an archive
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000393 if (!checkSignature(ErrorMsg))
394 return false;
Chris Lattner968cfd02003-04-19 21:45:34 +0000395
Reid Spencerf9d7a512004-11-14 21:58:33 +0000396 At += 8; // Skip signature
397
398 // Parse the first file member header
399 const char* FirstFile = At;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000400 ArchiveMember* mbr = parseMemberHeader(At, End, ErrorMsg);
401 if (!mbr)
402 return false;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000403
Reid Spencer9a29db42004-11-20 07:29:40 +0000404 if (mbr->isSVR4SymbolTable() || mbr->isBSD4SymbolTable()) {
Reid Spencerf9d7a512004-11-14 21:58:33 +0000405 // Skip the foreign symbol table, we don't do anything with it
406 At += mbr->getSize();
Reid Spencer6405c9e2004-11-19 17:08:00 +0000407 if ((intptr_t(At) & 1) == 1)
Reid Spencerb3231132004-11-15 01:40:20 +0000408 At++;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000409 delete mbr;
410
Reid Spencerb3231132004-11-15 01:40:20 +0000411 // Read the next one
Reid Spencerf9d7a512004-11-14 21:58:33 +0000412 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 }
Reid Spencerb3231132004-11-15 01:40:20 +0000418 }
419
420 if (mbr->isStringTable()) {
421 // Process the string table entry
Reid Spencer9a29db42004-11-20 07:29:40 +0000422 strtab.assign((const char*)mbr->getData(), mbr->getSize());
Reid Spencerb3231132004-11-15 01:40:20 +0000423 At += mbr->getSize();
Reid Spencer6405c9e2004-11-19 17:08:00 +0000424 if ((intptr_t(At) & 1) == 1)
Reid Spencerb3231132004-11-15 01:40:20 +0000425 At++;
426 delete mbr;
427 // Get the next one
428 FirstFile = At;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000429 mbr = parseMemberHeader(At, End, ErrorMsg);
430 if (!mbr) {
431 delete mbr;
432 return false;
433 }
Chris Lattner968cfd02003-04-19 21:45:34 +0000434 }
435
Reid Spencerf9d7a512004-11-14 21:58:33 +0000436 // See if its the symbol table
437 if (mbr->isLLVMSymbolTable()) {
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000438 if (!parseSymbolTable(mbr->getData(), mbr->getSize(), ErrorMsg)) {
439 delete mbr;
440 return false;
441 }
442
Reid Spencer8dde18f2004-11-28 03:13:02 +0000443 At += mbr->getSize();
Reid Spencer6405c9e2004-11-19 17:08:00 +0000444 if ((intptr_t(At) & 1) == 1)
Reid Spencer8dde18f2004-11-28 03:13:02 +0000445 At++;
Reid Spencerffb4d622006-05-12 17:56:20 +0000446 delete mbr;
447 // Can't be any more symtab headers so just advance
Reid Spencer8dde18f2004-11-28 03:13:02 +0000448 FirstFile = At;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000449 } else {
450 // There's no symbol table in the file. We have to rebuild it from scratch
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000451 // because the intent of this method is to get the symbol table loaded so
452 // it can be searched efficiently.
Reid Spencerf9d7a512004-11-14 21:58:33 +0000453 // Add the member to the members list
454 members.push_back(mbr);
455 }
456
457 firstFileOffset = FirstFile - base;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000458 return true;
Chris Lattner968cfd02003-04-19 21:45:34 +0000459}
460
Reid Spencerf9d7a512004-11-14 21:58:33 +0000461// Open the archive and load just the symbol tables
462Archive*
Reid Spencer5af46882004-12-13 02:59:03 +0000463Archive::OpenAndLoadSymbols(const sys::Path& file, std::string* ErrorMessage) {
Reid Spencer0ff2d312006-08-24 23:45:08 +0000464 std::auto_ptr<Archive> result ( new Archive(file) );
465 if (result->mapToMemory(ErrorMessage))
466 return 0;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000467 if (!result->loadSymbolTable(ErrorMessage))
Reid Spencer5af46882004-12-13 02:59:03 +0000468 return 0;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000469 return result.release();
Reid Spencerf9d7a512004-11-14 21:58:33 +0000470}
Chris Lattner968cfd02003-04-19 21:45:34 +0000471
Reid Spencerf9d7a512004-11-14 21:58:33 +0000472// Look up one symbol in the symbol table and return a ModuleProvider for the
473// module that defines that symbol.
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000474ModuleProvider*
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000475Archive::findModuleDefiningSymbol(const std::string& symbol,
476 std::string* ErrMsg) {
Reid Spencerf9d7a512004-11-14 21:58:33 +0000477 SymTabType::iterator SI = symTab.find(symbol);
478 if (SI == symTab.end())
479 return 0;
480
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000481 // The symbol table was previously constructed assuming that the members were
Reid Spencerf9d7a512004-11-14 21:58:33 +0000482 // written without the symbol table header. Because VBR encoding is used, the
483 // values could not be adjusted to account for the offset of the symbol table
484 // because that could affect the size of the symbol table due to VBR encoding.
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000485 // We now have to account for this by adjusting the offset by the size of the
Reid Spencerf9d7a512004-11-14 21:58:33 +0000486 // symbol table and its header.
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000487 unsigned fileOffset =
Reid Spencerf9d7a512004-11-14 21:58:33 +0000488 SI->second + // offset in symbol-table-less file
489 firstFileOffset; // add offset to first "real" file in archive
490
491 // See if the module is already loaded
492 ModuleMap::iterator MI = modules.find(fileOffset);
493 if (MI != modules.end())
494 return MI->second.first;
495
496 // Module hasn't been loaded yet, we need to load it
497 const char* modptr = base + fileOffset;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000498 ArchiveMember* mbr = parseMemberHeader(modptr, base + mapfile->size(),ErrMsg);
499 if (!mbr)
Reid Spencer0b5a5042006-08-25 17:43:11 +0000500 return 0;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000501
502 // Now, load the bytecode module to get the ModuleProvider
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000503 std::string FullMemberName = archPath.toString() + "(" +
Reid Spencer1fce0912004-12-11 00:14:15 +0000504 mbr->getPath().toString() + ")";
Chris Lattnere07c15c2007-05-06 06:18:07 +0000505 ModuleProvider* mp;
506 if (Bitcode) {
507 MemoryBuffer *Buffer =MemoryBuffer::getNewMemBuffer(mbr->getSize(),
508 FullMemberName.c_str());
509 memcpy((char*)Buffer->getBufferStart(), mbr->getData(), mbr->getSize());
510
511 mp = getBitcodeModuleProvider(Buffer, ErrMsg);
512 } else
513 mp = getBytecodeBufferModuleProvider(
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000514 (const unsigned char*) mbr->getData(), mbr->getSize(),
Chris Lattnerf2e292c2007-02-07 21:41:02 +0000515 FullMemberName, Decompressor, ErrMsg, 0);
Reid Spencer0b5a5042006-08-25 17:43:11 +0000516 if (!mp)
517 return 0;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000518
Reid Spencer9a29db42004-11-20 07:29:40 +0000519 modules.insert(std::make_pair(fileOffset, std::make_pair(mp, mbr)));
Reid Spencerf9d7a512004-11-14 21:58:33 +0000520
521 return mp;
522}
523
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000524// Look up multiple symbols in the symbol table and return a set of
Reid Spencerf9d7a512004-11-14 21:58:33 +0000525// ModuleProviders that define those symbols.
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000526bool
Reid Spencer7783e8a2004-11-19 03:18:22 +0000527Archive::findModulesDefiningSymbols(std::set<std::string>& symbols,
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000528 std::set<ModuleProvider*>& result,
Chris Lattnerf2e292c2007-02-07 21:41:02 +0000529 std::string* error) {
Reid Spencercd5561a2006-12-15 19:44:51 +0000530 if (!mapfile || !base) {
531 if (error)
532 *error = "Empty archive invalid for finding modules defining symbols";
533 return false;
534 }
535
Reid Spencer766b7932004-11-15 01:20:11 +0000536 if (symTab.empty()) {
537 // We don't have a symbol table, so we must build it now but lets also
538 // make sure that we populate the modules table as we do this to ensure
539 // that we don't load them twice when findModuleDefiningSymbol is called
540 // below.
541
542 // Get a pointer to the first file
543 const char* At = ((const char*)base) + firstFileOffset;
544 const char* End = ((const char*)base) + mapfile->size();
545
546 while ( At < End) {
547 // Compute the offset to be put in the symbol table
548 unsigned offset = At - base - firstFileOffset;
549
550 // Parse the file's header
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000551 ArchiveMember* mbr = parseMemberHeader(At, End, error);
552 if (!mbr)
553 return false;
Reid Spencer766b7932004-11-15 01:20:11 +0000554
555 // If it contains symbols
556 if (mbr->isBytecode() || mbr->isCompressedBytecode()) {
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000557 // Get the symbols
Reid Spencer766b7932004-11-15 01:20:11 +0000558 std::vector<std::string> symbols;
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000559 std::string FullMemberName = archPath.toString() + "(" +
Reid Spencer1fce0912004-12-11 00:14:15 +0000560 mbr->getPath().toString() + ")";
Chris Lattnerf2e292c2007-02-07 21:41:02 +0000561 ModuleProvider* MP =
562 GetBytecodeSymbols((const unsigned char*)At, mbr->getSize(),
563 FullMemberName, symbols,
564 Compressor::decompressToNewBuffer, error);
Reid Spencer766b7932004-11-15 01:20:11 +0000565
566 if (MP) {
567 // Insert the module's symbols into the symbol table
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000568 for (std::vector<std::string>::iterator I = symbols.begin(),
Reid Spencer766b7932004-11-15 01:20:11 +0000569 E=symbols.end(); I != E; ++I ) {
Reid Spencer9a29db42004-11-20 07:29:40 +0000570 symTab.insert(std::make_pair(*I, offset));
Reid Spencer766b7932004-11-15 01:20:11 +0000571 }
572 // Insert the ModuleProvider and the ArchiveMember into the table of
573 // modules.
Reid Spencer9a29db42004-11-20 07:29:40 +0000574 modules.insert(std::make_pair(offset, std::make_pair(MP, mbr)));
Reid Spencer766b7932004-11-15 01:20:11 +0000575 } else {
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000576 if (error)
577 *error = "Can't parse bytecode member: " +
Reid Spencer0b5a5042006-08-25 17:43:11 +0000578 mbr->getPath().toString() + ": " + *error;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000579 delete mbr;
580 return false;
Reid Spencer766b7932004-11-15 01:20:11 +0000581 }
582 }
Reid Spencerb3231132004-11-15 01:40:20 +0000583
584 // Go to the next file location
585 At += mbr->getSize();
Reid Spencer6405c9e2004-11-19 17:08:00 +0000586 if ((intptr_t(At) & 1) == 1)
Reid Spencerb3231132004-11-15 01:40:20 +0000587 At++;
Reid Spencer766b7932004-11-15 01:20:11 +0000588 }
589 }
590
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000591 // At this point we have a valid symbol table (one way or another) so we
Reid Spencer766b7932004-11-15 01:20:11 +0000592 // just use it to quickly find the symbols requested.
593
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000594 for (std::set<std::string>::iterator I=symbols.begin(),
Reid Spencer7783e8a2004-11-19 03:18:22 +0000595 E=symbols.end(); I != E;) {
596 // See if this symbol exists
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000597 ModuleProvider* mp = findModuleDefiningSymbol(*I,error);
Reid Spencerf9d7a512004-11-14 21:58:33 +0000598 if (mp) {
Reid Spencer7783e8a2004-11-19 03:18:22 +0000599 // The symbol exists, insert the ModuleProvider into our result,
600 // duplicates wil be ignored
Reid Spencer766b7932004-11-15 01:20:11 +0000601 result.insert(mp);
Reid Spencer7783e8a2004-11-19 03:18:22 +0000602
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000603 // Remove the symbol now that its been resolved, being careful to
Reid Spencer57646ec2004-11-19 03:44:10 +0000604 // post-increment the iterator.
605 symbols.erase(I++);
Reid Spencer7783e8a2004-11-19 03:18:22 +0000606 } else {
607 ++I;
Chris Lattner968cfd02003-04-19 21:45:34 +0000608 }
Reid Spencerf9d7a512004-11-14 21:58:33 +0000609 }
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000610 return true;
Chris Lattner968cfd02003-04-19 21:45:34 +0000611}
Reid Spencereaa06bb2005-02-26 22:00:32 +0000612
Chris Lattner7c439922005-09-23 06:22:58 +0000613bool Archive::isBytecodeArchive() {
614 // Make sure the symTab has been loaded. In most cases this should have been
615 // done when the archive was constructed, but still, this is just in case.
616 if (!symTab.size())
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000617 if (!loadSymbolTable(0))
618 return false;
Reid Spencereaa06bb2005-02-26 22:00:32 +0000619
Chris Lattner7c439922005-09-23 06:22:58 +0000620 // Now that we know it's been loaded, return true
621 // if it has a size
622 if (symTab.size()) return true;
Reid Spencereaa06bb2005-02-26 22:00:32 +0000623
624 //We still can't be sure it isn't a bytecode archive
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000625 if (!loadArchive(0))
626 return false;
Reid Spencereaa06bb2005-02-26 22:00:32 +0000627
628 std::vector<Module *> Modules;
629 std::string ErrorMessage;
630
Chris Lattner7c439922005-09-23 06:22:58 +0000631 // Scan the archive, trying to load a bytecode member. We only load one to
632 // see if this works.
633 for (iterator I = begin(), E = end(); I != E; ++I) {
634 if (!I->isBytecode() && !I->isCompressedBytecode())
635 continue;
636
637 std::string FullMemberName =
638 archPath.toString() + "(" + I->getPath().toString() + ")";
Chris Lattnere07c15c2007-05-06 06:18:07 +0000639 Module *M;
640
641 if (Bitcode) {
642 MemoryBuffer *Buffer =
643 MemoryBuffer::getNewMemBuffer(I->getSize(), FullMemberName.c_str());
644 memcpy((char*)Buffer->getBufferStart(), I->getData(), I->getSize());
645 M = ParseBitcodeFile(Buffer);
646 delete Buffer;
647 } else {
648 M = ParseBytecodeBuffer((const unsigned char*)I->getData(),
649 I->getSize(), FullMemberName);
650 }
Chris Lattner7c439922005-09-23 06:22:58 +0000651 if (!M)
652 return false; // Couldn't parse bytecode, not a bytecode archive.
653 delete M;
654 return true;
655 }
656
657 return false;
Reid Spencereaa06bb2005-02-26 22:00:32 +0000658}