blob: 8d607f0df7f8d672320f8f4725d6ec16812e60d2 [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"
Duraid Madina87487462005-12-26 14:31:26 +000018#include <memory>
Chris Lattner3446ae82004-01-10 19:00:15 +000019using namespace llvm;
Brian Gaeked0fde302003-11-11 22:41:34 +000020
Reid Spencerf9d7a512004-11-14 21:58:33 +000021/// Read a variable-bit-rate encoded unsigned integer
Dan Gohman844731a2008-05-13 00:00:25 +000022static inline unsigned readInteger(const char*&At, const char*End) {
Reid Spencerf9d7a512004-11-14 21:58:33 +000023 unsigned Shift = 0;
24 unsigned Result = 0;
Misha Brukman2b37d7c2005-04-21 21:13:18 +000025
Reid Spencerf9d7a512004-11-14 21:58:33 +000026 do {
Misha Brukman2b37d7c2005-04-21 21:13:18 +000027 if (At == End)
Reid Spencer8d8a7ff2006-07-07 20:56:50 +000028 return Result;
Reid Spencerf9d7a512004-11-14 21:58:33 +000029 Result |= (unsigned)((*At++) & 0x7F) << Shift;
30 Shift += 7;
31 } while (At[-1] & 0x80);
32 return Result;
Chris Lattner968cfd02003-04-19 21:45:34 +000033}
34
Reid Spencerf9d7a512004-11-14 21:58:33 +000035// Completely parse the Archive's symbol table and populate symTab member var.
Reid Spencer8d8a7ff2006-07-07 20:56:50 +000036bool
37Archive::parseSymbolTable(const void* data, unsigned size, std::string* error) {
Reid Spencerf9d7a512004-11-14 21:58:33 +000038 const char* At = (const char*) data;
39 const char* End = At + size;
40 while (At < End) {
41 unsigned offset = readInteger(At, End);
Reid Spencer8d8a7ff2006-07-07 20:56:50 +000042 if (At == End) {
43 if (error)
44 *error = "Ran out of data reading vbr_uint for symtab offset!";
45 return false;
46 }
Reid Spencerf9d7a512004-11-14 21:58:33 +000047 unsigned length = readInteger(At, End);
Reid Spencer8d8a7ff2006-07-07 20:56:50 +000048 if (At == End) {
49 if (error)
50 *error = "Ran out of data reading vbr_uint for symtab length!";
51 return false;
52 }
53 if (At + length > End) {
54 if (error)
55 *error = "Malformed symbol table: length not consistent with size";
56 return false;
57 }
Reid Spencerf9d7a512004-11-14 21:58:33 +000058 // we don't care if it can't be inserted (duplicate entry)
Reid Spencer9a29db42004-11-20 07:29:40 +000059 symTab.insert(std::make_pair(std::string(At, length), offset));
Reid Spencerf9d7a512004-11-14 21:58:33 +000060 At += length;
61 }
62 symTabSize = size;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +000063 return true;
Chris Lattner968cfd02003-04-19 21:45:34 +000064}
65
Reid Spencerf9d7a512004-11-14 21:58:33 +000066// This member parses an ArchiveMemberHeader that is presumed to be pointed to
67// by At. The At pointer is updated to the byte just after the header, which
Misha Brukman2b37d7c2005-04-21 21:13:18 +000068// can be variable in size.
Reid Spencerf9d7a512004-11-14 21:58:33 +000069ArchiveMember*
Reid Spencer8d8a7ff2006-07-07 20:56:50 +000070Archive::parseMemberHeader(const char*& At, const char* End, std::string* error)
71{
Reid Spencercd5561a2006-12-15 19:44:51 +000072 if (At + sizeof(ArchiveMemberHeader) >= End) {
73 if (error)
74 *error = "Unexpected end of file";
75 return 0;
76 }
Reid Spencerf9d7a512004-11-14 21:58:33 +000077
78 // Cast archive member header
79 ArchiveMemberHeader* Hdr = (ArchiveMemberHeader*)At;
80 At += sizeof(ArchiveMemberHeader);
81
Misha Brukman2b37d7c2005-04-21 21:13:18 +000082 // Extract the size and determine if the file is
Reid Spencerf9d7a512004-11-14 21:58:33 +000083 // compressed or not (negative length).
84 int flags = 0;
85 int MemberSize = atoi(Hdr->size);
86 if (MemberSize < 0) {
87 flags |= ArchiveMember::CompressedFlag;
88 MemberSize = -MemberSize;
89 }
90
91 // Check the size of the member for sanity
Reid Spencer8d8a7ff2006-07-07 20:56:50 +000092 if (At + MemberSize > End) {
93 if (error)
94 *error = "invalid member length in archive file";
95 return 0;
96 }
Reid Spencerf9d7a512004-11-14 21:58:33 +000097
98 // Check the member signature
Reid Spencer8d8a7ff2006-07-07 20:56:50 +000099 if (!Hdr->checkSignature()) {
100 if (error)
101 *error = "invalid file member signature";
102 return 0;
103 }
Reid Spencerf9d7a512004-11-14 21:58:33 +0000104
105 // Convert and check the member name
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000106 // The empty name ( '/' and 15 blanks) is for a foreign (non-LLVM) symbol
107 // table. The special name "//" and 14 blanks is for a string table, used
Reid Spencerf9d7a512004-11-14 21:58:33 +0000108 // for long file names. This library doesn't generate either of those but
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000109 // it will accept them. If the name starts with #1/ and the remainder is
110 // digits, then those digits specify the length of the name that is
111 // stored immediately following the header. The special name
Gabor Greifa99be512007-07-05 17:07:56 +0000112 // __LLVM_SYM_TAB__ identifies the symbol table for LLVM bitcode.
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000113 // Anything else is a regular, short filename that is terminated with
Reid Spencerf9d7a512004-11-14 21:58:33 +0000114 // a '/' and blanks.
115
116 std::string pathname;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000117 switch (Hdr->name[0]) {
118 case '#':
119 if (Hdr->name[1] == '1' && Hdr->name[2] == '/') {
120 if (isdigit(Hdr->name[3])) {
121 unsigned len = atoi(&Hdr->name[3]);
Reid Spencer9a29db42004-11-20 07:29:40 +0000122 pathname.assign(At, len);
Reid Spencerdd95e8d2004-11-17 16:13:11 +0000123 At += len;
124 MemberSize -= len;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000125 flags |= ArchiveMember::HasLongFilenameFlag;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000126 } else {
127 if (error)
128 *error = "invalid long filename";
129 return 0;
130 }
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000131 } else if (Hdr->name[1] == '_' &&
Reid Spencer9a29db42004-11-20 07:29:40 +0000132 (0 == memcmp(Hdr->name, ARFILE_LLVM_SYMTAB_NAME, 16))) {
Reid Spencerf9d7a512004-11-14 21:58:33 +0000133 // The member is using a long file name (>15 chars) format.
134 // This format is standard for 4.4BSD and Mac OSX operating
135 // systems. LLVM uses it similarly. In this format, the
136 // remainder of the name field (after #1/) specifies the
137 // length of the file name which occupy the first bytes of
138 // the member's data. The pathname already has the #1/ stripped.
139 pathname.assign(ARFILE_LLVM_SYMTAB_NAME);
140 flags |= ArchiveMember::LLVMSymbolTableFlag;
141 }
142 break;
143 case '/':
144 if (Hdr->name[1]== '/') {
Reid Spencer9a29db42004-11-20 07:29:40 +0000145 if (0 == memcmp(Hdr->name, ARFILE_STRTAB_NAME, 16)) {
Reid Spencerf9d7a512004-11-14 21:58:33 +0000146 pathname.assign(ARFILE_STRTAB_NAME);
147 flags |= ArchiveMember::StringTableFlag;
148 } else {
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000149 if (error)
150 *error = "invalid string table name";
151 return 0;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000152 }
153 } else if (Hdr->name[1] == ' ') {
Reid Spencer9a29db42004-11-20 07:29:40 +0000154 if (0 == memcmp(Hdr->name, ARFILE_SVR4_SYMTAB_NAME, 16)) {
155 pathname.assign(ARFILE_SVR4_SYMTAB_NAME);
156 flags |= ArchiveMember::SVR4SymbolTableFlag;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000157 } else {
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000158 if (error)
159 *error = "invalid SVR4 symbol table name";
160 return 0;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000161 }
162 } else if (isdigit(Hdr->name[1])) {
163 unsigned index = atoi(&Hdr->name[1]);
164 if (index < strtab.length()) {
165 const char* namep = strtab.c_str() + index;
166 const char* endp = strtab.c_str() + strtab.length();
167 const char* p = namep;
168 const char* last_p = p;
169 while (p < endp) {
170 if (*p == '\n' && *last_p == '/') {
Reid Spencer9a29db42004-11-20 07:29:40 +0000171 pathname.assign(namep, last_p - namep);
Reid Spencerf9d7a512004-11-14 21:58:33 +0000172 flags |= ArchiveMember::HasLongFilenameFlag;
173 break;
174 }
175 last_p = p;
176 p++;
177 }
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000178 if (p >= endp) {
179 if (error)
180 *error = "missing name termiantor in string table";
181 return 0;
182 }
Reid Spencerf9d7a512004-11-14 21:58:33 +0000183 } else {
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000184 if (error)
185 *error = "name index beyond string table";
186 return 0;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000187 }
188 }
189 break;
Reid Spencer9a29db42004-11-20 07:29:40 +0000190 case '_':
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000191 if (Hdr->name[1] == '_' &&
Reid Spencer9a29db42004-11-20 07:29:40 +0000192 (0 == memcmp(Hdr->name, ARFILE_BSD4_SYMTAB_NAME, 16))) {
193 pathname.assign(ARFILE_BSD4_SYMTAB_NAME);
194 flags |= ArchiveMember::BSD4SymbolTableFlag;
Reid Spencer84b9ced2004-11-23 22:35:39 +0000195 break;
Reid Spencer9a29db42004-11-20 07:29:40 +0000196 }
Reid Spencer84b9ced2004-11-23 22:35:39 +0000197 /* FALL THROUGH */
Reid Spencerf9d7a512004-11-14 21:58:33 +0000198
199 default:
Reid Spencer9a29db42004-11-20 07:29:40 +0000200 char* slash = (char*) memchr(Hdr->name, '/', 16);
Reid Spencerf9d7a512004-11-14 21:58:33 +0000201 if (slash == 0)
Reid Spencerdd95e8d2004-11-17 16:13:11 +0000202 slash = Hdr->name + 16;
Reid Spencer9a29db42004-11-20 07:29:40 +0000203 pathname.assign(Hdr->name, slash - Hdr->name);
Reid Spencerf9d7a512004-11-14 21:58:33 +0000204 break;
205 }
206
Gabor Greifdb5565a2007-07-06 20:28:40 +0000207 // Determine if this is a bitcode file
Reid Spencer9a29db42004-11-20 07:29:40 +0000208 switch (sys::IdentifyFileType(At, 4)) {
Chris Lattnere07c15c2007-05-06 06:18:07 +0000209 case sys::Bitcode_FileType:
Gabor Greife75ca3d2007-07-06 13:38:17 +0000210 flags |= ArchiveMember::BitcodeFlag;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000211 break;
212 default:
Gabor Greife75ca3d2007-07-06 13:38:17 +0000213 flags &= ~ArchiveMember::BitcodeFlag;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000214 break;
215 }
216
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000217 // Instantiate the ArchiveMember to be filled
218 ArchiveMember* member = new ArchiveMember(this);
219
Reid Spencerf9d7a512004-11-14 21:58:33 +0000220 // Fill in fields of the ArchiveMember
Reid Spencerf9d7a512004-11-14 21:58:33 +0000221 member->parent = this;
Reid Spencerdd04df02005-07-07 23:21:43 +0000222 member->path.set(pathname);
Reid Spencerf9d7a512004-11-14 21:58:33 +0000223 member->info.fileSize = MemberSize;
224 member->info.modTime.fromEpochTime(atoi(Hdr->date));
Reid Spencer56127622004-12-29 01:20:24 +0000225 unsigned int mode;
226 sscanf(Hdr->mode, "%o", &mode);
227 member->info.mode = mode;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000228 member->info.user = atoi(Hdr->uid);
229 member->info.group = atoi(Hdr->gid);
230 member->flags = flags;
231 member->data = At;
232
233 return member;
234}
235
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000236bool
237Archive::checkSignature(std::string* error) {
Reid Spencerf9d7a512004-11-14 21:58:33 +0000238 // Check the magic string at file's header
Chris Lattner7f6b4472008-04-01 04:26:46 +0000239 if (mapfile->getBufferSize() < 8 || memcmp(base, ARFILE_MAGIC, 8)) {
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000240 if (error)
241 *error = "invalid signature for an archive file";
242 return false;
243 }
244 return true;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000245}
246
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000247// This function loads the entire archive and fully populates its ilist with
Reid Spencerf9d7a512004-11-14 21:58:33 +0000248// the members of the archive file. This is typically used in preparation for
249// editing the contents of the archive.
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000250bool
251Archive::loadArchive(std::string* error) {
Reid Spencerf9d7a512004-11-14 21:58:33 +0000252
253 // Set up parsing
254 members.clear();
255 symTab.clear();
256 const char *At = base;
Chris Lattner7f6b4472008-04-01 04:26:46 +0000257 const char *End = mapfile->getBufferEnd();
Reid Spencerf9d7a512004-11-14 21:58:33 +0000258
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000259 if (!checkSignature(error))
260 return false;
261
Reid Spencerf9d7a512004-11-14 21:58:33 +0000262 At += 8; // Skip the magic string.
263
264 bool seenSymbolTable = false;
265 bool foundFirstFile = false;
266 while (At < End) {
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000267 // parse the member header
Reid Spencerf9d7a512004-11-14 21:58:33 +0000268 const char* Save = At;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000269 ArchiveMember* mbr = parseMemberHeader(At, End, error);
270 if (!mbr)
271 return false;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000272
273 // check if this is the foreign symbol table
Reid Spencer9a29db42004-11-20 07:29:40 +0000274 if (mbr->isSVR4SymbolTable() || mbr->isBSD4SymbolTable()) {
Reid Spencer4a980d12004-11-16 06:47:19 +0000275 // We just save this but don't do anything special
276 // with it. It doesn't count as the "first file".
Reid Spencer9a29db42004-11-20 07:29:40 +0000277 if (foreignST) {
278 // What? Multiple foreign symbol tables? Just chuck it
279 // and retain the last one found.
280 delete foreignST;
281 }
Reid Spencer4a980d12004-11-16 06:47:19 +0000282 foreignST = mbr;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000283 At += mbr->getSize();
Reid Spencer6405c9e2004-11-19 17:08:00 +0000284 if ((intptr_t(At) & 1) == 1)
Reid Spencerf9d7a512004-11-14 21:58:33 +0000285 At++;
286 } else if (mbr->isStringTable()) {
Reid Spencer4a980d12004-11-16 06:47:19 +0000287 // Simply suck the entire string table into a string
288 // variable. This will be used to get the names of the
289 // members that use the "/ddd" format for their names
290 // (SVR4 style long names).
Reid Spencer9a29db42004-11-20 07:29:40 +0000291 strtab.assign(At, mbr->getSize());
Reid Spencerf9d7a512004-11-14 21:58:33 +0000292 At += mbr->getSize();
Reid Spencer6405c9e2004-11-19 17:08:00 +0000293 if ((intptr_t(At) & 1) == 1)
Reid Spencerf9d7a512004-11-14 21:58:33 +0000294 At++;
295 delete mbr;
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000296 } else if (mbr->isLLVMSymbolTable()) {
Reid Spencer4a980d12004-11-16 06:47:19 +0000297 // This is the LLVM symbol table for the archive. If we've seen it
298 // already, its an error. Otherwise, parse the symbol table and move on.
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000299 if (seenSymbolTable) {
300 if (error)
301 *error = "invalid archive: multiple symbol tables";
302 return false;
303 }
304 if (!parseSymbolTable(mbr->getData(), mbr->getSize(), error))
305 return false;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000306 seenSymbolTable = true;
307 At += mbr->getSize();
Reid Spencer6405c9e2004-11-19 17:08:00 +0000308 if ((intptr_t(At) & 1) == 1)
Reid Spencerf9d7a512004-11-14 21:58:33 +0000309 At++;
Reid Spencer4a980d12004-11-16 06:47:19 +0000310 delete mbr; // We don't need this member in the list of members.
Reid Spencerf9d7a512004-11-14 21:58:33 +0000311 } else {
Reid Spencer4a980d12004-11-16 06:47:19 +0000312 // This is just a regular file. If its the first one, save its offset.
313 // Otherwise just push it on the list and move on to the next file.
Reid Spencerf9d7a512004-11-14 21:58:33 +0000314 if (!foundFirstFile) {
315 firstFileOffset = Save - base;
316 foundFirstFile = true;
317 }
318 members.push_back(mbr);
319 At += mbr->getSize();
Reid Spencer6405c9e2004-11-19 17:08:00 +0000320 if ((intptr_t(At) & 1) == 1)
Reid Spencerf9d7a512004-11-14 21:58:33 +0000321 At++;
322 }
323 }
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000324 return true;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000325}
326
327// Open and completely load the archive file.
328Archive*
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000329Archive::OpenAndLoad(const sys::Path& file, std::string* ErrorMessage)
330{
Reid Spencer0ff2d312006-08-24 23:45:08 +0000331 std::auto_ptr<Archive> result ( new Archive(file));
332 if (result->mapToMemory(ErrorMessage))
333 return 0;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000334 if (!result->loadArchive(ErrorMessage))
Reid Spencer5af46882004-12-13 02:59:03 +0000335 return 0;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000336 return result.release();
Reid Spencerf9d7a512004-11-14 21:58:33 +0000337}
338
Gabor Greifa99be512007-07-05 17:07:56 +0000339// Get all the bitcode modules from the archive
Reid Spencerf9d7a512004-11-14 21:58:33 +0000340bool
341Archive::getAllModules(std::vector<Module*>& Modules, std::string* ErrMessage) {
342
343 for (iterator I=begin(), E=end(); I != E; ++I) {
Gabor Greife75ca3d2007-07-06 13:38:17 +0000344 if (I->isBitcode()) {
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000345 std::string FullMemberName = archPath.toString() +
Reid Spencer1fce0912004-12-11 00:14:15 +0000346 "(" + I->getPath().toString() + ")";
Chris Lattnerc1d56242007-05-06 09:28:33 +0000347 MemoryBuffer *Buffer =
348 MemoryBuffer::getNewMemBuffer(I->getSize(), FullMemberName.c_str());
349 memcpy((char*)Buffer->getBufferStart(), I->getData(), I->getSize());
Chris Lattnere07c15c2007-05-06 06:18:07 +0000350
Chris Lattnerc1d56242007-05-06 09:28:33 +0000351 Module *M = ParseBitcodeFile(Buffer, ErrMessage);
352 delete Buffer;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000353 if (!M)
354 return true;
355
356 Modules.push_back(M);
357 }
358 }
Brian Gaeke2c61d7b2003-11-16 23:08:48 +0000359 return false;
360}
Chris Lattner968cfd02003-04-19 21:45:34 +0000361
Reid Spencerf9d7a512004-11-14 21:58:33 +0000362// Load just the symbol table from the archive file
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000363bool
364Archive::loadSymbolTable(std::string* ErrorMsg) {
Chris Lattner968cfd02003-04-19 21:45:34 +0000365
Reid Spencerf9d7a512004-11-14 21:58:33 +0000366 // Set up parsing
367 members.clear();
368 symTab.clear();
369 const char *At = base;
Chris Lattner7f6b4472008-04-01 04:26:46 +0000370 const char *End = mapfile->getBufferEnd();
Chris Lattner968cfd02003-04-19 21:45:34 +0000371
Reid Spencerf9d7a512004-11-14 21:58:33 +0000372 // Make sure we're dealing with an archive
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000373 if (!checkSignature(ErrorMsg))
374 return false;
Chris Lattner968cfd02003-04-19 21:45:34 +0000375
Reid Spencerf9d7a512004-11-14 21:58:33 +0000376 At += 8; // Skip signature
377
378 // Parse the first file member header
379 const char* FirstFile = At;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000380 ArchiveMember* mbr = parseMemberHeader(At, End, ErrorMsg);
381 if (!mbr)
382 return false;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000383
Reid Spencer9a29db42004-11-20 07:29:40 +0000384 if (mbr->isSVR4SymbolTable() || mbr->isBSD4SymbolTable()) {
Reid Spencerf9d7a512004-11-14 21:58:33 +0000385 // Skip the foreign symbol table, we don't do anything with it
386 At += mbr->getSize();
Reid Spencer6405c9e2004-11-19 17:08:00 +0000387 if ((intptr_t(At) & 1) == 1)
Reid Spencerb3231132004-11-15 01:40:20 +0000388 At++;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000389 delete mbr;
390
Reid Spencerb3231132004-11-15 01:40:20 +0000391 // Read the next one
Reid Spencerf9d7a512004-11-14 21:58:33 +0000392 FirstFile = At;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000393 mbr = parseMemberHeader(At, End, ErrorMsg);
394 if (!mbr) {
395 delete mbr;
396 return false;
397 }
Reid Spencerb3231132004-11-15 01:40:20 +0000398 }
399
400 if (mbr->isStringTable()) {
401 // Process the string table entry
Reid Spencer9a29db42004-11-20 07:29:40 +0000402 strtab.assign((const char*)mbr->getData(), mbr->getSize());
Reid Spencerb3231132004-11-15 01:40:20 +0000403 At += mbr->getSize();
Reid Spencer6405c9e2004-11-19 17:08:00 +0000404 if ((intptr_t(At) & 1) == 1)
Reid Spencerb3231132004-11-15 01:40:20 +0000405 At++;
406 delete mbr;
407 // Get the next one
408 FirstFile = At;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000409 mbr = parseMemberHeader(At, End, ErrorMsg);
410 if (!mbr) {
411 delete mbr;
412 return false;
413 }
Chris Lattner968cfd02003-04-19 21:45:34 +0000414 }
415
Reid Spencerf9d7a512004-11-14 21:58:33 +0000416 // See if its the symbol table
417 if (mbr->isLLVMSymbolTable()) {
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000418 if (!parseSymbolTable(mbr->getData(), mbr->getSize(), ErrorMsg)) {
419 delete mbr;
420 return false;
421 }
422
Reid Spencer8dde18f2004-11-28 03:13:02 +0000423 At += mbr->getSize();
Reid Spencer6405c9e2004-11-19 17:08:00 +0000424 if ((intptr_t(At) & 1) == 1)
Reid Spencer8dde18f2004-11-28 03:13:02 +0000425 At++;
Reid Spencerffb4d622006-05-12 17:56:20 +0000426 delete mbr;
427 // Can't be any more symtab headers so just advance
Reid Spencer8dde18f2004-11-28 03:13:02 +0000428 FirstFile = At;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000429 } else {
430 // There's no symbol table in the file. We have to rebuild it from scratch
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000431 // because the intent of this method is to get the symbol table loaded so
432 // it can be searched efficiently.
Reid Spencerf9d7a512004-11-14 21:58:33 +0000433 // Add the member to the members list
434 members.push_back(mbr);
435 }
436
437 firstFileOffset = FirstFile - base;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000438 return true;
Chris Lattner968cfd02003-04-19 21:45:34 +0000439}
440
Reid Spencerf9d7a512004-11-14 21:58:33 +0000441// Open the archive and load just the symbol tables
442Archive*
Reid Spencer5af46882004-12-13 02:59:03 +0000443Archive::OpenAndLoadSymbols(const sys::Path& file, std::string* ErrorMessage) {
Reid Spencer0ff2d312006-08-24 23:45:08 +0000444 std::auto_ptr<Archive> result ( new Archive(file) );
445 if (result->mapToMemory(ErrorMessage))
446 return 0;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000447 if (!result->loadSymbolTable(ErrorMessage))
Reid Spencer5af46882004-12-13 02:59:03 +0000448 return 0;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000449 return result.release();
Reid Spencerf9d7a512004-11-14 21:58:33 +0000450}
Chris Lattner968cfd02003-04-19 21:45:34 +0000451
Reid Spencerf9d7a512004-11-14 21:58:33 +0000452// Look up one symbol in the symbol table and return a ModuleProvider for the
453// module that defines that symbol.
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000454ModuleProvider*
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000455Archive::findModuleDefiningSymbol(const std::string& symbol,
456 std::string* ErrMsg) {
Reid Spencerf9d7a512004-11-14 21:58:33 +0000457 SymTabType::iterator SI = symTab.find(symbol);
458 if (SI == symTab.end())
459 return 0;
460
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000461 // The symbol table was previously constructed assuming that the members were
Reid Spencerf9d7a512004-11-14 21:58:33 +0000462 // written without the symbol table header. Because VBR encoding is used, the
463 // values could not be adjusted to account for the offset of the symbol table
464 // because that could affect the size of the symbol table due to VBR encoding.
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000465 // We now have to account for this by adjusting the offset by the size of the
Reid Spencerf9d7a512004-11-14 21:58:33 +0000466 // symbol table and its header.
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000467 unsigned fileOffset =
Reid Spencerf9d7a512004-11-14 21:58:33 +0000468 SI->second + // offset in symbol-table-less file
469 firstFileOffset; // add offset to first "real" file in archive
470
471 // See if the module is already loaded
472 ModuleMap::iterator MI = modules.find(fileOffset);
473 if (MI != modules.end())
474 return MI->second.first;
475
476 // Module hasn't been loaded yet, we need to load it
477 const char* modptr = base + fileOffset;
Chris Lattner7f6b4472008-04-01 04:26:46 +0000478 ArchiveMember* mbr = parseMemberHeader(modptr, mapfile->getBufferEnd(),
479 ErrMsg);
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000480 if (!mbr)
Reid Spencer0b5a5042006-08-25 17:43:11 +0000481 return 0;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000482
Gabor Greifa99be512007-07-05 17:07:56 +0000483 // Now, load the bitcode module to get the ModuleProvider
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000484 std::string FullMemberName = archPath.toString() + "(" +
Reid Spencer1fce0912004-12-11 00:14:15 +0000485 mbr->getPath().toString() + ")";
Chris Lattnerc1d56242007-05-06 09:28:33 +0000486 MemoryBuffer *Buffer =MemoryBuffer::getNewMemBuffer(mbr->getSize(),
487 FullMemberName.c_str());
488 memcpy((char*)Buffer->getBufferStart(), mbr->getData(), mbr->getSize());
489
490 ModuleProvider *mp = getBitcodeModuleProvider(Buffer, ErrMsg);
Reid Spencer0b5a5042006-08-25 17:43:11 +0000491 if (!mp)
492 return 0;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000493
Reid Spencer9a29db42004-11-20 07:29:40 +0000494 modules.insert(std::make_pair(fileOffset, std::make_pair(mp, mbr)));
Reid Spencerf9d7a512004-11-14 21:58:33 +0000495
496 return mp;
497}
498
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000499// Look up multiple symbols in the symbol table and return a set of
Reid Spencerf9d7a512004-11-14 21:58:33 +0000500// ModuleProviders that define those symbols.
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000501bool
Reid Spencer7783e8a2004-11-19 03:18:22 +0000502Archive::findModulesDefiningSymbols(std::set<std::string>& symbols,
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000503 std::set<ModuleProvider*>& result,
Chris Lattnerf2e292c2007-02-07 21:41:02 +0000504 std::string* error) {
Reid Spencercd5561a2006-12-15 19:44:51 +0000505 if (!mapfile || !base) {
506 if (error)
507 *error = "Empty archive invalid for finding modules defining symbols";
508 return false;
509 }
510
Reid Spencer766b7932004-11-15 01:20:11 +0000511 if (symTab.empty()) {
512 // We don't have a symbol table, so we must build it now but lets also
513 // make sure that we populate the modules table as we do this to ensure
514 // that we don't load them twice when findModuleDefiningSymbol is called
515 // below.
516
517 // Get a pointer to the first file
Chris Lattner7f6b4472008-04-01 04:26:46 +0000518 const char* At = base + firstFileOffset;
519 const char* End = mapfile->getBufferEnd();
Reid Spencer766b7932004-11-15 01:20:11 +0000520
521 while ( At < End) {
522 // Compute the offset to be put in the symbol table
523 unsigned offset = At - base - firstFileOffset;
524
525 // Parse the file's header
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000526 ArchiveMember* mbr = parseMemberHeader(At, End, error);
527 if (!mbr)
528 return false;
Reid Spencer766b7932004-11-15 01:20:11 +0000529
530 // If it contains symbols
Gabor Greife75ca3d2007-07-06 13:38:17 +0000531 if (mbr->isBitcode()) {
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000532 // Get the symbols
Reid Spencer766b7932004-11-15 01:20:11 +0000533 std::vector<std::string> symbols;
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000534 std::string FullMemberName = archPath.toString() + "(" +
Reid Spencer1fce0912004-12-11 00:14:15 +0000535 mbr->getPath().toString() + ")";
Chris Lattnerf2e292c2007-02-07 21:41:02 +0000536 ModuleProvider* MP =
Gabor Greifa99be512007-07-05 17:07:56 +0000537 GetBitcodeSymbols((const unsigned char*)At, mbr->getSize(),
538 FullMemberName, symbols, error);
Reid Spencer766b7932004-11-15 01:20:11 +0000539
540 if (MP) {
541 // Insert the module's symbols into the symbol table
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000542 for (std::vector<std::string>::iterator I = symbols.begin(),
Reid Spencer766b7932004-11-15 01:20:11 +0000543 E=symbols.end(); I != E; ++I ) {
Reid Spencer9a29db42004-11-20 07:29:40 +0000544 symTab.insert(std::make_pair(*I, offset));
Reid Spencer766b7932004-11-15 01:20:11 +0000545 }
546 // Insert the ModuleProvider and the ArchiveMember into the table of
547 // modules.
Reid Spencer9a29db42004-11-20 07:29:40 +0000548 modules.insert(std::make_pair(offset, std::make_pair(MP, mbr)));
Reid Spencer766b7932004-11-15 01:20:11 +0000549 } else {
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000550 if (error)
Gabor Greifa99be512007-07-05 17:07:56 +0000551 *error = "Can't parse bitcode member: " +
Reid Spencer0b5a5042006-08-25 17:43:11 +0000552 mbr->getPath().toString() + ": " + *error;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000553 delete mbr;
554 return false;
Reid Spencer766b7932004-11-15 01:20:11 +0000555 }
556 }
Reid Spencerb3231132004-11-15 01:40:20 +0000557
558 // Go to the next file location
559 At += mbr->getSize();
Reid Spencer6405c9e2004-11-19 17:08:00 +0000560 if ((intptr_t(At) & 1) == 1)
Reid Spencerb3231132004-11-15 01:40:20 +0000561 At++;
Reid Spencer766b7932004-11-15 01:20:11 +0000562 }
563 }
564
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000565 // At this point we have a valid symbol table (one way or another) so we
Reid Spencer766b7932004-11-15 01:20:11 +0000566 // just use it to quickly find the symbols requested.
567
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000568 for (std::set<std::string>::iterator I=symbols.begin(),
Reid Spencer7783e8a2004-11-19 03:18:22 +0000569 E=symbols.end(); I != E;) {
570 // See if this symbol exists
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000571 ModuleProvider* mp = findModuleDefiningSymbol(*I,error);
Reid Spencerf9d7a512004-11-14 21:58:33 +0000572 if (mp) {
Reid Spencer7783e8a2004-11-19 03:18:22 +0000573 // The symbol exists, insert the ModuleProvider into our result,
574 // duplicates wil be ignored
Reid Spencer766b7932004-11-15 01:20:11 +0000575 result.insert(mp);
Reid Spencer7783e8a2004-11-19 03:18:22 +0000576
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000577 // Remove the symbol now that its been resolved, being careful to
Reid Spencer57646ec2004-11-19 03:44:10 +0000578 // post-increment the iterator.
579 symbols.erase(I++);
Reid Spencer7783e8a2004-11-19 03:18:22 +0000580 } else {
581 ++I;
Chris Lattner968cfd02003-04-19 21:45:34 +0000582 }
Reid Spencerf9d7a512004-11-14 21:58:33 +0000583 }
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000584 return true;
Chris Lattner968cfd02003-04-19 21:45:34 +0000585}
Reid Spencereaa06bb2005-02-26 22:00:32 +0000586
Gabor Greifa99be512007-07-05 17:07:56 +0000587bool Archive::isBitcodeArchive() {
Chris Lattner7c439922005-09-23 06:22:58 +0000588 // Make sure the symTab has been loaded. In most cases this should have been
589 // done when the archive was constructed, but still, this is just in case.
Dan Gohman30359592008-01-29 13:02:09 +0000590 if (symTab.empty())
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000591 if (!loadSymbolTable(0))
592 return false;
Reid Spencereaa06bb2005-02-26 22:00:32 +0000593
Chris Lattner7c439922005-09-23 06:22:58 +0000594 // Now that we know it's been loaded, return true
595 // if it has a size
596 if (symTab.size()) return true;
Reid Spencereaa06bb2005-02-26 22:00:32 +0000597
Gabor Greifa99be512007-07-05 17:07:56 +0000598 // We still can't be sure it isn't a bitcode archive
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000599 if (!loadArchive(0))
600 return false;
Reid Spencereaa06bb2005-02-26 22:00:32 +0000601
602 std::vector<Module *> Modules;
603 std::string ErrorMessage;
604
Gabor Greifa99be512007-07-05 17:07:56 +0000605 // Scan the archive, trying to load a bitcode member. We only load one to
Chris Lattner7c439922005-09-23 06:22:58 +0000606 // see if this works.
607 for (iterator I = begin(), E = end(); I != E; ++I) {
Gabor Greife75ca3d2007-07-06 13:38:17 +0000608 if (!I->isBitcode())
Chris Lattner7c439922005-09-23 06:22:58 +0000609 continue;
610
611 std::string FullMemberName =
612 archPath.toString() + "(" + I->getPath().toString() + ")";
Chris Lattnerc1d56242007-05-06 09:28:33 +0000613
614 MemoryBuffer *Buffer =
615 MemoryBuffer::getNewMemBuffer(I->getSize(), FullMemberName.c_str());
616 memcpy((char*)Buffer->getBufferStart(), I->getData(), I->getSize());
617 Module *M = ParseBitcodeFile(Buffer);
618 delete Buffer;
Chris Lattner7c439922005-09-23 06:22:58 +0000619 if (!M)
Gabor Greifa99be512007-07-05 17:07:56 +0000620 return false; // Couldn't parse bitcode, not a bitcode archive.
Chris Lattner7c439922005-09-23 06:22:58 +0000621 delete M;
622 return true;
623 }
624
625 return false;
Reid Spencereaa06bb2005-02-26 22:00:32 +0000626}