blob: 68873e2768d33cb1ca5ec3d682b1ecc4d4d1f1c6 [file] [log] [blame]
Reid Spencerf9d7a512004-11-14 21:58:33 +00001//===-- ArchiveReader.cpp - Read LLVM archive files -------------*- C++ -*-===//
Misha Brukman2b37d7c2005-04-21 21:13:18 +00002//
John Criswellb576c942003-10-20 19:43:21 +00003// The LLVM Compiler Infrastructure
4//
Chris Lattner4ee451d2007-12-29 20:36:04 +00005// This file is distributed under the University of Illinois Open Source
6// License. See LICENSE.TXT for details.
Misha Brukman2b37d7c2005-04-21 21:13:18 +00007//
John Criswellb576c942003-10-20 19:43:21 +00008//===----------------------------------------------------------------------===//
Chris Lattner968cfd02003-04-19 21:45:34 +00009//
Gabor Greifa99be512007-07-05 17:07:56 +000010// Builds up standard unix archive files (.a) containing LLVM bitcode.
Chris Lattner968cfd02003-04-19 21:45:34 +000011//
12//===----------------------------------------------------------------------===//
13
Reid Spencerf9d7a512004-11-14 21:58:33 +000014#include "ArchiveInternals.h"
Rafael Espindola603d6b52012-01-23 03:41:53 +000015#include "llvm/ADT/SmallPtrSet.h"
Chris Lattnere07c15c2007-05-06 06:18:07 +000016#include "llvm/Bitcode/ReaderWriter.h"
Chris Lattnere07c15c2007-05-06 06:18:07 +000017#include "llvm/Support/MemoryBuffer.h"
Chris Lattnerc1d56242007-05-06 09:28:33 +000018#include "llvm/Module.h"
Benjamin Kramer901b8582012-03-23 11:35:30 +000019#include <cstdio>
Dan Gohmand68a0762009-01-05 17:59:02 +000020#include <cstdlib>
Duraid Madina87487462005-12-26 14:31:26 +000021#include <memory>
Chris Lattner3446ae82004-01-10 19:00:15 +000022using namespace llvm;
Brian Gaeked0fde302003-11-11 22:41:34 +000023
Reid Spencerf9d7a512004-11-14 21:58:33 +000024/// Read a variable-bit-rate encoded unsigned integer
Dan Gohman844731a2008-05-13 00:00:25 +000025static inline 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
Gabor Greifa99be512007-07-05 17:07:56 +0000115 // __LLVM_SYM_TAB__ identifies the symbol table for LLVM bitcode.
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000116 // 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]);
Chris Lattner46333062010-02-04 06:19:43 +0000125 const char *nulp = (const char *)memchr(At, '\0', len);
Chris Lattner73382422010-04-25 04:44:26 +0000126 pathname.assign(At, nulp != 0 ? (uintptr_t)(nulp - At) : len);
Reid Spencerdd95e8d2004-11-17 16:13:11 +0000127 At += len;
128 MemberSize -= len;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000129 flags |= ArchiveMember::HasLongFilenameFlag;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000130 } else {
131 if (error)
132 *error = "invalid long filename";
133 return 0;
134 }
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000135 } else if (Hdr->name[1] == '_' &&
Reid Spencer9a29db42004-11-20 07:29:40 +0000136 (0 == memcmp(Hdr->name, ARFILE_LLVM_SYMTAB_NAME, 16))) {
Reid Spencerf9d7a512004-11-14 21:58:33 +0000137 // The member is using a long file name (>15 chars) format.
138 // This format is standard for 4.4BSD and Mac OSX operating
139 // systems. LLVM uses it similarly. In this format, the
140 // remainder of the name field (after #1/) specifies the
141 // length of the file name which occupy the first bytes of
142 // the member's data. The pathname already has the #1/ stripped.
143 pathname.assign(ARFILE_LLVM_SYMTAB_NAME);
144 flags |= ArchiveMember::LLVMSymbolTableFlag;
145 }
146 break;
147 case '/':
148 if (Hdr->name[1]== '/') {
Reid Spencer9a29db42004-11-20 07:29:40 +0000149 if (0 == memcmp(Hdr->name, ARFILE_STRTAB_NAME, 16)) {
Reid Spencerf9d7a512004-11-14 21:58:33 +0000150 pathname.assign(ARFILE_STRTAB_NAME);
151 flags |= ArchiveMember::StringTableFlag;
152 } else {
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000153 if (error)
154 *error = "invalid string table name";
155 return 0;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000156 }
157 } else if (Hdr->name[1] == ' ') {
Reid Spencer9a29db42004-11-20 07:29:40 +0000158 if (0 == memcmp(Hdr->name, ARFILE_SVR4_SYMTAB_NAME, 16)) {
159 pathname.assign(ARFILE_SVR4_SYMTAB_NAME);
160 flags |= ArchiveMember::SVR4SymbolTableFlag;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000161 } else {
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000162 if (error)
163 *error = "invalid SVR4 symbol table name";
164 return 0;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000165 }
166 } else if (isdigit(Hdr->name[1])) {
167 unsigned index = atoi(&Hdr->name[1]);
168 if (index < strtab.length()) {
169 const char* namep = strtab.c_str() + index;
170 const char* endp = strtab.c_str() + strtab.length();
171 const char* p = namep;
172 const char* last_p = p;
173 while (p < endp) {
174 if (*p == '\n' && *last_p == '/') {
Reid Spencer9a29db42004-11-20 07:29:40 +0000175 pathname.assign(namep, last_p - namep);
Reid Spencerf9d7a512004-11-14 21:58:33 +0000176 flags |= ArchiveMember::HasLongFilenameFlag;
177 break;
178 }
179 last_p = p;
180 p++;
181 }
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000182 if (p >= endp) {
183 if (error)
184 *error = "missing name termiantor in string table";
185 return 0;
186 }
Reid Spencerf9d7a512004-11-14 21:58:33 +0000187 } else {
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000188 if (error)
189 *error = "name index beyond string table";
190 return 0;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000191 }
192 }
193 break;
Reid Spencer9a29db42004-11-20 07:29:40 +0000194 case '_':
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000195 if (Hdr->name[1] == '_' &&
Reid Spencer9a29db42004-11-20 07:29:40 +0000196 (0 == memcmp(Hdr->name, ARFILE_BSD4_SYMTAB_NAME, 16))) {
197 pathname.assign(ARFILE_BSD4_SYMTAB_NAME);
198 flags |= ArchiveMember::BSD4SymbolTableFlag;
Reid Spencer84b9ced2004-11-23 22:35:39 +0000199 break;
Reid Spencer9a29db42004-11-20 07:29:40 +0000200 }
Reid Spencer84b9ced2004-11-23 22:35:39 +0000201 /* FALL THROUGH */
Reid Spencerf9d7a512004-11-14 21:58:33 +0000202
203 default:
Reid Spencer9a29db42004-11-20 07:29:40 +0000204 char* slash = (char*) memchr(Hdr->name, '/', 16);
Reid Spencerf9d7a512004-11-14 21:58:33 +0000205 if (slash == 0)
Reid Spencerdd95e8d2004-11-17 16:13:11 +0000206 slash = Hdr->name + 16;
Reid Spencer9a29db42004-11-20 07:29:40 +0000207 pathname.assign(Hdr->name, slash - Hdr->name);
Reid Spencerf9d7a512004-11-14 21:58:33 +0000208 break;
209 }
210
Gabor Greifdb5565a2007-07-06 20:28:40 +0000211 // Determine if this is a bitcode file
Reid Spencer9a29db42004-11-20 07:29:40 +0000212 switch (sys::IdentifyFileType(At, 4)) {
Chris Lattnere07c15c2007-05-06 06:18:07 +0000213 case sys::Bitcode_FileType:
Gabor Greife75ca3d2007-07-06 13:38:17 +0000214 flags |= ArchiveMember::BitcodeFlag;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000215 break;
216 default:
Gabor Greife75ca3d2007-07-06 13:38:17 +0000217 flags &= ~ArchiveMember::BitcodeFlag;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000218 break;
219 }
220
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000221 // Instantiate the ArchiveMember to be filled
222 ArchiveMember* member = new ArchiveMember(this);
223
Reid Spencerf9d7a512004-11-14 21:58:33 +0000224 // Fill in fields of the ArchiveMember
Reid Spencerf9d7a512004-11-14 21:58:33 +0000225 member->parent = this;
Reid Spencerdd04df02005-07-07 23:21:43 +0000226 member->path.set(pathname);
Reid Spencerf9d7a512004-11-14 21:58:33 +0000227 member->info.fileSize = MemberSize;
228 member->info.modTime.fromEpochTime(atoi(Hdr->date));
Reid Spencer56127622004-12-29 01:20:24 +0000229 unsigned int mode;
230 sscanf(Hdr->mode, "%o", &mode);
231 member->info.mode = mode;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000232 member->info.user = atoi(Hdr->uid);
233 member->info.group = atoi(Hdr->gid);
234 member->flags = flags;
235 member->data = At;
236
237 return member;
238}
239
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000240bool
241Archive::checkSignature(std::string* error) {
Reid Spencerf9d7a512004-11-14 21:58:33 +0000242 // Check the magic string at file's header
Chris Lattner7f6b4472008-04-01 04:26:46 +0000243 if (mapfile->getBufferSize() < 8 || memcmp(base, ARFILE_MAGIC, 8)) {
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000244 if (error)
245 *error = "invalid signature for an archive file";
246 return false;
247 }
248 return true;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000249}
250
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000251// This function loads the entire archive and fully populates its ilist with
Reid Spencerf9d7a512004-11-14 21:58:33 +0000252// the members of the archive file. This is typically used in preparation for
253// editing the contents of the archive.
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000254bool
255Archive::loadArchive(std::string* error) {
Reid Spencerf9d7a512004-11-14 21:58:33 +0000256
257 // Set up parsing
258 members.clear();
259 symTab.clear();
260 const char *At = base;
Chris Lattner7f6b4472008-04-01 04:26:46 +0000261 const char *End = mapfile->getBufferEnd();
Reid Spencerf9d7a512004-11-14 21:58:33 +0000262
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000263 if (!checkSignature(error))
264 return false;
265
Reid Spencerf9d7a512004-11-14 21:58:33 +0000266 At += 8; // Skip the magic string.
267
268 bool seenSymbolTable = false;
269 bool foundFirstFile = false;
270 while (At < End) {
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000271 // parse the member header
Reid Spencerf9d7a512004-11-14 21:58:33 +0000272 const char* Save = At;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000273 ArchiveMember* mbr = parseMemberHeader(At, End, error);
274 if (!mbr)
275 return false;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000276
277 // check if this is the foreign symbol table
Reid Spencer9a29db42004-11-20 07:29:40 +0000278 if (mbr->isSVR4SymbolTable() || mbr->isBSD4SymbolTable()) {
Reid Spencer4a980d12004-11-16 06:47:19 +0000279 // We just save this but don't do anything special
280 // with it. It doesn't count as the "first file".
Reid Spencer9a29db42004-11-20 07:29:40 +0000281 if (foreignST) {
282 // What? Multiple foreign symbol tables? Just chuck it
283 // and retain the last one found.
284 delete foreignST;
285 }
Reid Spencer4a980d12004-11-16 06:47:19 +0000286 foreignST = mbr;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000287 At += mbr->getSize();
Reid Spencer6405c9e2004-11-19 17:08:00 +0000288 if ((intptr_t(At) & 1) == 1)
Reid Spencerf9d7a512004-11-14 21:58:33 +0000289 At++;
290 } else if (mbr->isStringTable()) {
Reid Spencer4a980d12004-11-16 06:47:19 +0000291 // Simply suck the entire string table into a string
292 // variable. This will be used to get the names of the
293 // members that use the "/ddd" format for their names
294 // (SVR4 style long names).
Reid Spencer9a29db42004-11-20 07:29:40 +0000295 strtab.assign(At, mbr->getSize());
Reid Spencerf9d7a512004-11-14 21:58:33 +0000296 At += mbr->getSize();
Reid Spencer6405c9e2004-11-19 17:08:00 +0000297 if ((intptr_t(At) & 1) == 1)
Reid Spencerf9d7a512004-11-14 21:58:33 +0000298 At++;
299 delete mbr;
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000300 } else if (mbr->isLLVMSymbolTable()) {
Reid Spencer4a980d12004-11-16 06:47:19 +0000301 // This is the LLVM symbol table for the archive. If we've seen it
302 // already, its an error. Otherwise, parse the symbol table and move on.
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000303 if (seenSymbolTable) {
304 if (error)
305 *error = "invalid archive: multiple symbol tables";
306 return false;
307 }
308 if (!parseSymbolTable(mbr->getData(), mbr->getSize(), error))
309 return false;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000310 seenSymbolTable = true;
311 At += mbr->getSize();
Reid Spencer6405c9e2004-11-19 17:08:00 +0000312 if ((intptr_t(At) & 1) == 1)
Reid Spencerf9d7a512004-11-14 21:58:33 +0000313 At++;
Reid Spencer4a980d12004-11-16 06:47:19 +0000314 delete mbr; // We don't need this member in the list of members.
Reid Spencerf9d7a512004-11-14 21:58:33 +0000315 } else {
Reid Spencer4a980d12004-11-16 06:47:19 +0000316 // This is just a regular file. If its the first one, save its offset.
317 // Otherwise just push it on the list and move on to the next file.
Reid Spencerf9d7a512004-11-14 21:58:33 +0000318 if (!foundFirstFile) {
319 firstFileOffset = Save - base;
320 foundFirstFile = true;
321 }
322 members.push_back(mbr);
323 At += mbr->getSize();
Reid Spencer6405c9e2004-11-19 17:08:00 +0000324 if ((intptr_t(At) & 1) == 1)
Reid Spencerf9d7a512004-11-14 21:58:33 +0000325 At++;
326 }
327 }
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000328 return true;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000329}
330
331// Open and completely load the archive file.
332Archive*
Owen Anderson4434ed42009-07-01 23:13:44 +0000333Archive::OpenAndLoad(const sys::Path& file, LLVMContext& C,
Owen Anderson8b477ed2009-07-01 16:58:40 +0000334 std::string* ErrorMessage) {
335 std::auto_ptr<Archive> result ( new Archive(file, C));
Reid Spencer0ff2d312006-08-24 23:45:08 +0000336 if (result->mapToMemory(ErrorMessage))
337 return 0;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000338 if (!result->loadArchive(ErrorMessage))
Reid Spencer5af46882004-12-13 02:59:03 +0000339 return 0;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000340 return result.release();
Reid Spencerf9d7a512004-11-14 21:58:33 +0000341}
342
Gabor Greifa99be512007-07-05 17:07:56 +0000343// Get all the bitcode modules from the archive
Reid Spencerf9d7a512004-11-14 21:58:33 +0000344bool
Owen Anderson8b477ed2009-07-01 16:58:40 +0000345Archive::getAllModules(std::vector<Module*>& Modules,
346 std::string* ErrMessage) {
Reid Spencerf9d7a512004-11-14 21:58:33 +0000347
348 for (iterator I=begin(), E=end(); I != E; ++I) {
Gabor Greife75ca3d2007-07-06 13:38:17 +0000349 if (I->isBitcode()) {
Chris Lattner74382b72009-08-23 22:45:37 +0000350 std::string FullMemberName = archPath.str() +
351 "(" + I->getPath().str() + ")";
Chris Lattnerc1d56242007-05-06 09:28:33 +0000352 MemoryBuffer *Buffer =
Benjamin Kramer9d44e702010-04-19 16:15:31 +0000353 MemoryBuffer::getMemBufferCopy(StringRef(I->getData(), I->getSize()),
354 FullMemberName.c_str());
Chris Lattnere07c15c2007-05-06 06:18:07 +0000355
Owen Anderson8b477ed2009-07-01 16:58:40 +0000356 Module *M = ParseBitcodeFile(Buffer, Context, ErrMessage);
Chris Lattnerc1d56242007-05-06 09:28:33 +0000357 delete Buffer;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000358 if (!M)
359 return true;
360
361 Modules.push_back(M);
362 }
363 }
Brian Gaeke2c61d7b2003-11-16 23:08:48 +0000364 return false;
365}
Chris Lattner968cfd02003-04-19 21:45:34 +0000366
Reid Spencerf9d7a512004-11-14 21:58:33 +0000367// Load just the symbol table from the archive file
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000368bool
369Archive::loadSymbolTable(std::string* ErrorMsg) {
Chris Lattner968cfd02003-04-19 21:45:34 +0000370
Reid Spencerf9d7a512004-11-14 21:58:33 +0000371 // Set up parsing
372 members.clear();
373 symTab.clear();
374 const char *At = base;
Chris Lattner7f6b4472008-04-01 04:26:46 +0000375 const char *End = mapfile->getBufferEnd();
Chris Lattner968cfd02003-04-19 21:45:34 +0000376
Reid Spencerf9d7a512004-11-14 21:58:33 +0000377 // Make sure we're dealing with an archive
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000378 if (!checkSignature(ErrorMsg))
379 return false;
Chris Lattner968cfd02003-04-19 21:45:34 +0000380
Reid Spencerf9d7a512004-11-14 21:58:33 +0000381 At += 8; // Skip signature
382
383 // Parse the first file member header
384 const char* FirstFile = At;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000385 ArchiveMember* mbr = parseMemberHeader(At, End, ErrorMsg);
386 if (!mbr)
387 return false;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000388
Reid Spencer9a29db42004-11-20 07:29:40 +0000389 if (mbr->isSVR4SymbolTable() || mbr->isBSD4SymbolTable()) {
Reid Spencerf9d7a512004-11-14 21:58:33 +0000390 // Skip the foreign symbol table, we don't do anything with it
391 At += mbr->getSize();
Reid Spencer6405c9e2004-11-19 17:08:00 +0000392 if ((intptr_t(At) & 1) == 1)
Reid Spencerb3231132004-11-15 01:40:20 +0000393 At++;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000394 delete mbr;
395
Reid Spencerb3231132004-11-15 01:40:20 +0000396 // Read the next one
Reid Spencerf9d7a512004-11-14 21:58:33 +0000397 FirstFile = At;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000398 mbr = parseMemberHeader(At, End, ErrorMsg);
399 if (!mbr) {
400 delete mbr;
401 return false;
402 }
Reid Spencerb3231132004-11-15 01:40:20 +0000403 }
404
405 if (mbr->isStringTable()) {
406 // Process the string table entry
Reid Spencer9a29db42004-11-20 07:29:40 +0000407 strtab.assign((const char*)mbr->getData(), mbr->getSize());
Reid Spencerb3231132004-11-15 01:40:20 +0000408 At += mbr->getSize();
Reid Spencer6405c9e2004-11-19 17:08:00 +0000409 if ((intptr_t(At) & 1) == 1)
Reid Spencerb3231132004-11-15 01:40:20 +0000410 At++;
411 delete mbr;
412 // Get the next one
413 FirstFile = At;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000414 mbr = parseMemberHeader(At, End, ErrorMsg);
415 if (!mbr) {
416 delete mbr;
417 return false;
418 }
Chris Lattner968cfd02003-04-19 21:45:34 +0000419 }
420
Reid Spencerf9d7a512004-11-14 21:58:33 +0000421 // See if its the symbol table
422 if (mbr->isLLVMSymbolTable()) {
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000423 if (!parseSymbolTable(mbr->getData(), mbr->getSize(), ErrorMsg)) {
424 delete mbr;
425 return false;
426 }
427
Reid Spencer8dde18f2004-11-28 03:13:02 +0000428 At += mbr->getSize();
Reid Spencer6405c9e2004-11-19 17:08:00 +0000429 if ((intptr_t(At) & 1) == 1)
Reid Spencer8dde18f2004-11-28 03:13:02 +0000430 At++;
Reid Spencerffb4d622006-05-12 17:56:20 +0000431 delete mbr;
432 // Can't be any more symtab headers so just advance
Reid Spencer8dde18f2004-11-28 03:13:02 +0000433 FirstFile = At;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000434 } else {
435 // There's no symbol table in the file. We have to rebuild it from scratch
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000436 // because the intent of this method is to get the symbol table loaded so
437 // it can be searched efficiently.
Reid Spencerf9d7a512004-11-14 21:58:33 +0000438 // Add the member to the members list
439 members.push_back(mbr);
440 }
441
442 firstFileOffset = FirstFile - base;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000443 return true;
Chris Lattner968cfd02003-04-19 21:45:34 +0000444}
445
Reid Spencerf9d7a512004-11-14 21:58:33 +0000446// Open the archive and load just the symbol tables
Owen Anderson31895e72009-07-01 21:22:36 +0000447Archive* Archive::OpenAndLoadSymbols(const sys::Path& file,
Owen Anderson4434ed42009-07-01 23:13:44 +0000448 LLVMContext& C,
Owen Anderson8b477ed2009-07-01 16:58:40 +0000449 std::string* ErrorMessage) {
450 std::auto_ptr<Archive> result ( new Archive(file, C) );
Reid Spencer0ff2d312006-08-24 23:45:08 +0000451 if (result->mapToMemory(ErrorMessage))
452 return 0;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000453 if (!result->loadSymbolTable(ErrorMessage))
Reid Spencer5af46882004-12-13 02:59:03 +0000454 return 0;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000455 return result.release();
Reid Spencerf9d7a512004-11-14 21:58:33 +0000456}
Chris Lattner968cfd02003-04-19 21:45:34 +0000457
Jeffrey Yasskinf0356fe2010-01-27 20:34:15 +0000458// Look up one symbol in the symbol table and return the module that defines
459// that symbol.
460Module*
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000461Archive::findModuleDefiningSymbol(const std::string& symbol,
462 std::string* ErrMsg) {
Reid Spencerf9d7a512004-11-14 21:58:33 +0000463 SymTabType::iterator SI = symTab.find(symbol);
464 if (SI == symTab.end())
465 return 0;
466
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000467 // The symbol table was previously constructed assuming that the members were
Reid Spencerf9d7a512004-11-14 21:58:33 +0000468 // written without the symbol table header. Because VBR encoding is used, the
469 // values could not be adjusted to account for the offset of the symbol table
470 // because that could affect the size of the symbol table due to VBR encoding.
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000471 // We now have to account for this by adjusting the offset by the size of the
Reid Spencerf9d7a512004-11-14 21:58:33 +0000472 // symbol table and its header.
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000473 unsigned fileOffset =
Reid Spencerf9d7a512004-11-14 21:58:33 +0000474 SI->second + // offset in symbol-table-less file
475 firstFileOffset; // add offset to first "real" file in archive
476
477 // See if the module is already loaded
478 ModuleMap::iterator MI = modules.find(fileOffset);
479 if (MI != modules.end())
480 return MI->second.first;
481
482 // Module hasn't been loaded yet, we need to load it
483 const char* modptr = base + fileOffset;
Chris Lattner7f6b4472008-04-01 04:26:46 +0000484 ArchiveMember* mbr = parseMemberHeader(modptr, mapfile->getBufferEnd(),
485 ErrMsg);
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000486 if (!mbr)
Reid Spencer0b5a5042006-08-25 17:43:11 +0000487 return 0;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000488
Jeffrey Yasskinf0356fe2010-01-27 20:34:15 +0000489 // Now, load the bitcode module to get the Module.
Chris Lattner74382b72009-08-23 22:45:37 +0000490 std::string FullMemberName = archPath.str() + "(" +
491 mbr->getPath().str() + ")";
Benjamin Kramer9d44e702010-04-19 16:15:31 +0000492 MemoryBuffer *Buffer =
493 MemoryBuffer::getMemBufferCopy(StringRef(mbr->getData(), mbr->getSize()),
494 FullMemberName.c_str());
Chris Lattnerc1d56242007-05-06 09:28:33 +0000495
Jeffrey Yasskinf0356fe2010-01-27 20:34:15 +0000496 Module *m = getLazyBitcodeModule(Buffer, Context, ErrMsg);
497 if (!m)
Reid Spencer0b5a5042006-08-25 17:43:11 +0000498 return 0;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000499
Jeffrey Yasskinf0356fe2010-01-27 20:34:15 +0000500 modules.insert(std::make_pair(fileOffset, std::make_pair(m, mbr)));
Reid Spencerf9d7a512004-11-14 21:58:33 +0000501
Jeffrey Yasskinf0356fe2010-01-27 20:34:15 +0000502 return m;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000503}
504
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000505// Look up multiple symbols in the symbol table and return a set of
Jeffrey Yasskinf0356fe2010-01-27 20:34:15 +0000506// Modules that define those symbols.
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000507bool
Reid Spencer7783e8a2004-11-19 03:18:22 +0000508Archive::findModulesDefiningSymbols(std::set<std::string>& symbols,
Rafael Espindola603d6b52012-01-23 03:41:53 +0000509 SmallVectorImpl<Module*>& result,
Chris Lattnerf2e292c2007-02-07 21:41:02 +0000510 std::string* error) {
Reid Spencercd5561a2006-12-15 19:44:51 +0000511 if (!mapfile || !base) {
512 if (error)
513 *error = "Empty archive invalid for finding modules defining symbols";
514 return false;
515 }
516
Reid Spencer766b7932004-11-15 01:20:11 +0000517 if (symTab.empty()) {
518 // We don't have a symbol table, so we must build it now but lets also
519 // make sure that we populate the modules table as we do this to ensure
520 // that we don't load them twice when findModuleDefiningSymbol is called
521 // below.
522
523 // Get a pointer to the first file
Chris Lattner7f6b4472008-04-01 04:26:46 +0000524 const char* At = base + firstFileOffset;
525 const char* End = mapfile->getBufferEnd();
Reid Spencer766b7932004-11-15 01:20:11 +0000526
527 while ( At < End) {
528 // Compute the offset to be put in the symbol table
529 unsigned offset = At - base - firstFileOffset;
530
531 // Parse the file's header
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000532 ArchiveMember* mbr = parseMemberHeader(At, End, error);
533 if (!mbr)
534 return false;
Reid Spencer766b7932004-11-15 01:20:11 +0000535
536 // If it contains symbols
Gabor Greife75ca3d2007-07-06 13:38:17 +0000537 if (mbr->isBitcode()) {
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000538 // Get the symbols
Reid Spencer766b7932004-11-15 01:20:11 +0000539 std::vector<std::string> symbols;
Chris Lattner74382b72009-08-23 22:45:37 +0000540 std::string FullMemberName = archPath.str() + "(" +
541 mbr->getPath().str() + ")";
Jeffrey Yasskinf0356fe2010-01-27 20:34:15 +0000542 Module* M =
Benjamin Kramer9d44e702010-04-19 16:15:31 +0000543 GetBitcodeSymbols(At, mbr->getSize(), FullMemberName, Context,
544 symbols, error);
Reid Spencer766b7932004-11-15 01:20:11 +0000545
Jeffrey Yasskinf0356fe2010-01-27 20:34:15 +0000546 if (M) {
Reid Spencer766b7932004-11-15 01:20:11 +0000547 // Insert the module's symbols into the symbol table
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000548 for (std::vector<std::string>::iterator I = symbols.begin(),
Reid Spencer766b7932004-11-15 01:20:11 +0000549 E=symbols.end(); I != E; ++I ) {
Reid Spencer9a29db42004-11-20 07:29:40 +0000550 symTab.insert(std::make_pair(*I, offset));
Reid Spencer766b7932004-11-15 01:20:11 +0000551 }
Jeffrey Yasskinf0356fe2010-01-27 20:34:15 +0000552 // Insert the Module and the ArchiveMember into the table of
Reid Spencer766b7932004-11-15 01:20:11 +0000553 // modules.
Jeffrey Yasskinf0356fe2010-01-27 20:34:15 +0000554 modules.insert(std::make_pair(offset, std::make_pair(M, mbr)));
Reid Spencer766b7932004-11-15 01:20:11 +0000555 } else {
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000556 if (error)
Gabor Greifa99be512007-07-05 17:07:56 +0000557 *error = "Can't parse bitcode member: " +
Chris Lattner74382b72009-08-23 22:45:37 +0000558 mbr->getPath().str() + ": " + *error;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000559 delete mbr;
560 return false;
Reid Spencer766b7932004-11-15 01:20:11 +0000561 }
562 }
Reid Spencerb3231132004-11-15 01:40:20 +0000563
564 // Go to the next file location
565 At += mbr->getSize();
Reid Spencer6405c9e2004-11-19 17:08:00 +0000566 if ((intptr_t(At) & 1) == 1)
Reid Spencerb3231132004-11-15 01:40:20 +0000567 At++;
Reid Spencer766b7932004-11-15 01:20:11 +0000568 }
569 }
570
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000571 // At this point we have a valid symbol table (one way or another) so we
Reid Spencer766b7932004-11-15 01:20:11 +0000572 // just use it to quickly find the symbols requested.
573
Rafael Espindola603d6b52012-01-23 03:41:53 +0000574 SmallPtrSet<Module*, 16> Added;
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000575 for (std::set<std::string>::iterator I=symbols.begin(),
Rafael Espindola4254df32012-01-23 05:07:16 +0000576 Next = I,
577 E=symbols.end(); I != E; I = Next) {
578 // Increment Next before we invalidate it.
579 ++Next;
580
Reid Spencer7783e8a2004-11-19 03:18:22 +0000581 // See if this symbol exists
Jeffrey Yasskinf0356fe2010-01-27 20:34:15 +0000582 Module* m = findModuleDefiningSymbol(*I,error);
Rafael Espindola603d6b52012-01-23 03:41:53 +0000583 if (!m)
584 continue;
585 bool NewMember = Added.insert(m);
586 if (!NewMember)
587 continue;
Reid Spencer7783e8a2004-11-19 03:18:22 +0000588
Rafael Espindola603d6b52012-01-23 03:41:53 +0000589 // The symbol exists, insert the Module into our result.
590 result.push_back(m);
591
592 // Remove the symbol now that its been resolved.
593 symbols.erase(I);
Reid Spencerf9d7a512004-11-14 21:58:33 +0000594 }
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000595 return true;
Chris Lattner968cfd02003-04-19 21:45:34 +0000596}
Reid Spencereaa06bb2005-02-26 22:00:32 +0000597
Gabor Greifa99be512007-07-05 17:07:56 +0000598bool Archive::isBitcodeArchive() {
Chris Lattner7c439922005-09-23 06:22:58 +0000599 // Make sure the symTab has been loaded. In most cases this should have been
600 // done when the archive was constructed, but still, this is just in case.
Dan Gohman30359592008-01-29 13:02:09 +0000601 if (symTab.empty())
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000602 if (!loadSymbolTable(0))
603 return false;
Reid Spencereaa06bb2005-02-26 22:00:32 +0000604
Chris Lattner7c439922005-09-23 06:22:58 +0000605 // Now that we know it's been loaded, return true
606 // if it has a size
607 if (symTab.size()) return true;
Reid Spencereaa06bb2005-02-26 22:00:32 +0000608
Gabor Greifa99be512007-07-05 17:07:56 +0000609 // We still can't be sure it isn't a bitcode archive
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000610 if (!loadArchive(0))
611 return false;
Reid Spencereaa06bb2005-02-26 22:00:32 +0000612
613 std::vector<Module *> Modules;
614 std::string ErrorMessage;
615
Gabor Greifa99be512007-07-05 17:07:56 +0000616 // Scan the archive, trying to load a bitcode member. We only load one to
Chris Lattner7c439922005-09-23 06:22:58 +0000617 // see if this works.
618 for (iterator I = begin(), E = end(); I != E; ++I) {
Gabor Greife75ca3d2007-07-06 13:38:17 +0000619 if (!I->isBitcode())
Chris Lattner7c439922005-09-23 06:22:58 +0000620 continue;
621
622 std::string FullMemberName =
Chris Lattner74382b72009-08-23 22:45:37 +0000623 archPath.str() + "(" + I->getPath().str() + ")";
Chris Lattnerc1d56242007-05-06 09:28:33 +0000624
625 MemoryBuffer *Buffer =
Benjamin Kramer9d44e702010-04-19 16:15:31 +0000626 MemoryBuffer::getMemBufferCopy(StringRef(I->getData(), I->getSize()),
627 FullMemberName.c_str());
Owen Anderson8b477ed2009-07-01 16:58:40 +0000628 Module *M = ParseBitcodeFile(Buffer, Context);
Chris Lattnerc1d56242007-05-06 09:28:33 +0000629 delete Buffer;
Chris Lattner7c439922005-09-23 06:22:58 +0000630 if (!M)
Gabor Greifa99be512007-07-05 17:07:56 +0000631 return false; // Couldn't parse bitcode, not a bitcode archive.
Chris Lattner7c439922005-09-23 06:22:58 +0000632 delete M;
633 return true;
634 }
635
636 return false;
Reid Spencereaa06bb2005-02-26 22:00:32 +0000637}