blob: 5052495c0d62ebd30d3e405f83c02f09a81f39a6 [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
Roman Divacky651e8002012-09-05 22:09:23 +000082 const ArchiveMemberHeader* Hdr = (const ArchiveMemberHeader*)At;
Reid Spencerf9d7a512004-11-14 21:58:33 +000083 At += sizeof(ArchiveMemberHeader);
84
Reid Spencerf9d7a512004-11-14 21:58:33 +000085 int flags = 0;
86 int MemberSize = atoi(Hdr->size);
Rafael Espindola94bc2462012-08-10 01:57:52 +000087 assert(MemberSize >= 0);
Reid Spencerf9d7a512004-11-14 21:58:33 +000088
89 // Check the size of the member for sanity
Reid Spencer8d8a7ff2006-07-07 20:56:50 +000090 if (At + MemberSize > End) {
91 if (error)
92 *error = "invalid member length in archive file";
93 return 0;
94 }
Reid Spencerf9d7a512004-11-14 21:58:33 +000095
96 // Check the member signature
Reid Spencer8d8a7ff2006-07-07 20:56:50 +000097 if (!Hdr->checkSignature()) {
98 if (error)
99 *error = "invalid file member signature";
100 return 0;
101 }
Reid Spencerf9d7a512004-11-14 21:58:33 +0000102
103 // Convert and check the member name
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000104 // The empty name ( '/' and 15 blanks) is for a foreign (non-LLVM) symbol
105 // table. The special name "//" and 14 blanks is for a string table, used
Reid Spencerf9d7a512004-11-14 21:58:33 +0000106 // for long file names. This library doesn't generate either of those but
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000107 // it will accept them. If the name starts with #1/ and the remainder is
108 // digits, then those digits specify the length of the name that is
109 // stored immediately following the header. The special name
Gabor Greifa99be512007-07-05 17:07:56 +0000110 // __LLVM_SYM_TAB__ identifies the symbol table for LLVM bitcode.
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000111 // Anything else is a regular, short filename that is terminated with
Reid Spencerf9d7a512004-11-14 21:58:33 +0000112 // a '/' and blanks.
113
114 std::string pathname;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000115 switch (Hdr->name[0]) {
116 case '#':
117 if (Hdr->name[1] == '1' && Hdr->name[2] == '/') {
118 if (isdigit(Hdr->name[3])) {
119 unsigned len = atoi(&Hdr->name[3]);
Chris Lattner46333062010-02-04 06:19:43 +0000120 const char *nulp = (const char *)memchr(At, '\0', len);
Chris Lattner73382422010-04-25 04:44:26 +0000121 pathname.assign(At, nulp != 0 ? (uintptr_t)(nulp - At) : len);
Reid Spencerdd95e8d2004-11-17 16:13:11 +0000122 At += len;
123 MemberSize -= len;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000124 flags |= ArchiveMember::HasLongFilenameFlag;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000125 } else {
126 if (error)
127 *error = "invalid long filename";
128 return 0;
129 }
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000130 } else if (Hdr->name[1] == '_' &&
Reid Spencer9a29db42004-11-20 07:29:40 +0000131 (0 == memcmp(Hdr->name, ARFILE_LLVM_SYMTAB_NAME, 16))) {
Reid Spencerf9d7a512004-11-14 21:58:33 +0000132 // The member is using a long file name (>15 chars) format.
133 // This format is standard for 4.4BSD and Mac OSX operating
134 // systems. LLVM uses it similarly. In this format, the
135 // remainder of the name field (after #1/) specifies the
136 // length of the file name which occupy the first bytes of
137 // the member's data. The pathname already has the #1/ stripped.
138 pathname.assign(ARFILE_LLVM_SYMTAB_NAME);
139 flags |= ArchiveMember::LLVMSymbolTableFlag;
140 }
141 break;
142 case '/':
143 if (Hdr->name[1]== '/') {
Reid Spencer9a29db42004-11-20 07:29:40 +0000144 if (0 == memcmp(Hdr->name, ARFILE_STRTAB_NAME, 16)) {
Reid Spencerf9d7a512004-11-14 21:58:33 +0000145 pathname.assign(ARFILE_STRTAB_NAME);
146 flags |= ArchiveMember::StringTableFlag;
147 } else {
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000148 if (error)
149 *error = "invalid string table name";
150 return 0;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000151 }
152 } else if (Hdr->name[1] == ' ') {
Reid Spencer9a29db42004-11-20 07:29:40 +0000153 if (0 == memcmp(Hdr->name, ARFILE_SVR4_SYMTAB_NAME, 16)) {
154 pathname.assign(ARFILE_SVR4_SYMTAB_NAME);
155 flags |= ArchiveMember::SVR4SymbolTableFlag;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000156 } else {
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000157 if (error)
158 *error = "invalid SVR4 symbol table name";
159 return 0;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000160 }
161 } else if (isdigit(Hdr->name[1])) {
162 unsigned index = atoi(&Hdr->name[1]);
163 if (index < strtab.length()) {
164 const char* namep = strtab.c_str() + index;
165 const char* endp = strtab.c_str() + strtab.length();
166 const char* p = namep;
167 const char* last_p = p;
168 while (p < endp) {
169 if (*p == '\n' && *last_p == '/') {
Reid Spencer9a29db42004-11-20 07:29:40 +0000170 pathname.assign(namep, last_p - namep);
Reid Spencerf9d7a512004-11-14 21:58:33 +0000171 flags |= ArchiveMember::HasLongFilenameFlag;
172 break;
173 }
174 last_p = p;
175 p++;
176 }
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000177 if (p >= endp) {
178 if (error)
179 *error = "missing name termiantor in string table";
180 return 0;
181 }
Reid Spencerf9d7a512004-11-14 21:58:33 +0000182 } else {
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000183 if (error)
184 *error = "name index beyond string table";
185 return 0;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000186 }
187 }
188 break;
Reid Spencer9a29db42004-11-20 07:29:40 +0000189 case '_':
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000190 if (Hdr->name[1] == '_' &&
Reid Spencer9a29db42004-11-20 07:29:40 +0000191 (0 == memcmp(Hdr->name, ARFILE_BSD4_SYMTAB_NAME, 16))) {
192 pathname.assign(ARFILE_BSD4_SYMTAB_NAME);
193 flags |= ArchiveMember::BSD4SymbolTableFlag;
Reid Spencer84b9ced2004-11-23 22:35:39 +0000194 break;
Reid Spencer9a29db42004-11-20 07:29:40 +0000195 }
Reid Spencer84b9ced2004-11-23 22:35:39 +0000196 /* FALL THROUGH */
Reid Spencerf9d7a512004-11-14 21:58:33 +0000197
198 default:
Roman Divacky651e8002012-09-05 22:09:23 +0000199 const char* slash = (const char*) memchr(Hdr->name, '/', 16);
Reid Spencerf9d7a512004-11-14 21:58:33 +0000200 if (slash == 0)
Reid Spencerdd95e8d2004-11-17 16:13:11 +0000201 slash = Hdr->name + 16;
Reid Spencer9a29db42004-11-20 07:29:40 +0000202 pathname.assign(Hdr->name, slash - Hdr->name);
Reid Spencerf9d7a512004-11-14 21:58:33 +0000203 break;
204 }
205
Gabor Greifdb5565a2007-07-06 20:28:40 +0000206 // Determine if this is a bitcode file
Reid Spencer9a29db42004-11-20 07:29:40 +0000207 switch (sys::IdentifyFileType(At, 4)) {
Chris Lattnere07c15c2007-05-06 06:18:07 +0000208 case sys::Bitcode_FileType:
Gabor Greife75ca3d2007-07-06 13:38:17 +0000209 flags |= ArchiveMember::BitcodeFlag;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000210 break;
211 default:
Gabor Greife75ca3d2007-07-06 13:38:17 +0000212 flags &= ~ArchiveMember::BitcodeFlag;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000213 break;
214 }
215
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000216 // Instantiate the ArchiveMember to be filled
217 ArchiveMember* member = new ArchiveMember(this);
218
Reid Spencerf9d7a512004-11-14 21:58:33 +0000219 // Fill in fields of the ArchiveMember
Reid Spencerf9d7a512004-11-14 21:58:33 +0000220 member->parent = this;
Reid Spencerdd04df02005-07-07 23:21:43 +0000221 member->path.set(pathname);
Reid Spencerf9d7a512004-11-14 21:58:33 +0000222 member->info.fileSize = MemberSize;
223 member->info.modTime.fromEpochTime(atoi(Hdr->date));
Reid Spencer56127622004-12-29 01:20:24 +0000224 unsigned int mode;
225 sscanf(Hdr->mode, "%o", &mode);
226 member->info.mode = mode;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000227 member->info.user = atoi(Hdr->uid);
228 member->info.group = atoi(Hdr->gid);
229 member->flags = flags;
230 member->data = At;
231
232 return member;
233}
234
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000235bool
236Archive::checkSignature(std::string* error) {
Reid Spencerf9d7a512004-11-14 21:58:33 +0000237 // Check the magic string at file's header
Chris Lattner7f6b4472008-04-01 04:26:46 +0000238 if (mapfile->getBufferSize() < 8 || memcmp(base, ARFILE_MAGIC, 8)) {
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000239 if (error)
240 *error = "invalid signature for an archive file";
241 return false;
242 }
243 return true;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000244}
245
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000246// This function loads the entire archive and fully populates its ilist with
Reid Spencerf9d7a512004-11-14 21:58:33 +0000247// the members of the archive file. This is typically used in preparation for
248// editing the contents of the archive.
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000249bool
250Archive::loadArchive(std::string* error) {
Reid Spencerf9d7a512004-11-14 21:58:33 +0000251
252 // Set up parsing
253 members.clear();
254 symTab.clear();
255 const char *At = base;
Chris Lattner7f6b4472008-04-01 04:26:46 +0000256 const char *End = mapfile->getBufferEnd();
Reid Spencerf9d7a512004-11-14 21:58:33 +0000257
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000258 if (!checkSignature(error))
259 return false;
260
Reid Spencerf9d7a512004-11-14 21:58:33 +0000261 At += 8; // Skip the magic string.
262
263 bool seenSymbolTable = false;
264 bool foundFirstFile = false;
265 while (At < End) {
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000266 // parse the member header
Reid Spencerf9d7a512004-11-14 21:58:33 +0000267 const char* Save = At;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000268 ArchiveMember* mbr = parseMemberHeader(At, End, error);
269 if (!mbr)
270 return false;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000271
272 // check if this is the foreign symbol table
Reid Spencer9a29db42004-11-20 07:29:40 +0000273 if (mbr->isSVR4SymbolTable() || mbr->isBSD4SymbolTable()) {
Reid Spencer4a980d12004-11-16 06:47:19 +0000274 // We just save this but don't do anything special
275 // with it. It doesn't count as the "first file".
Reid Spencer9a29db42004-11-20 07:29:40 +0000276 if (foreignST) {
277 // What? Multiple foreign symbol tables? Just chuck it
278 // and retain the last one found.
279 delete foreignST;
280 }
Reid Spencer4a980d12004-11-16 06:47:19 +0000281 foreignST = mbr;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000282 At += mbr->getSize();
Reid Spencer6405c9e2004-11-19 17:08:00 +0000283 if ((intptr_t(At) & 1) == 1)
Reid Spencerf9d7a512004-11-14 21:58:33 +0000284 At++;
285 } else if (mbr->isStringTable()) {
Reid Spencer4a980d12004-11-16 06:47:19 +0000286 // Simply suck the entire string table into a string
287 // variable. This will be used to get the names of the
288 // members that use the "/ddd" format for their names
289 // (SVR4 style long names).
Reid Spencer9a29db42004-11-20 07:29:40 +0000290 strtab.assign(At, mbr->getSize());
Reid Spencerf9d7a512004-11-14 21:58:33 +0000291 At += mbr->getSize();
Reid Spencer6405c9e2004-11-19 17:08:00 +0000292 if ((intptr_t(At) & 1) == 1)
Reid Spencerf9d7a512004-11-14 21:58:33 +0000293 At++;
294 delete mbr;
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000295 } else if (mbr->isLLVMSymbolTable()) {
Reid Spencer4a980d12004-11-16 06:47:19 +0000296 // This is the LLVM symbol table for the archive. If we've seen it
297 // already, its an error. Otherwise, parse the symbol table and move on.
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000298 if (seenSymbolTable) {
299 if (error)
300 *error = "invalid archive: multiple symbol tables";
301 return false;
302 }
303 if (!parseSymbolTable(mbr->getData(), mbr->getSize(), error))
304 return false;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000305 seenSymbolTable = true;
306 At += mbr->getSize();
Reid Spencer6405c9e2004-11-19 17:08:00 +0000307 if ((intptr_t(At) & 1) == 1)
Reid Spencerf9d7a512004-11-14 21:58:33 +0000308 At++;
Reid Spencer4a980d12004-11-16 06:47:19 +0000309 delete mbr; // We don't need this member in the list of members.
Reid Spencerf9d7a512004-11-14 21:58:33 +0000310 } else {
Reid Spencer4a980d12004-11-16 06:47:19 +0000311 // This is just a regular file. If its the first one, save its offset.
312 // Otherwise just push it on the list and move on to the next file.
Reid Spencerf9d7a512004-11-14 21:58:33 +0000313 if (!foundFirstFile) {
314 firstFileOffset = Save - base;
315 foundFirstFile = true;
316 }
317 members.push_back(mbr);
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++;
321 }
322 }
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000323 return true;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000324}
325
326// Open and completely load the archive file.
327Archive*
Owen Anderson4434ed42009-07-01 23:13:44 +0000328Archive::OpenAndLoad(const sys::Path& file, LLVMContext& C,
Owen Anderson8b477ed2009-07-01 16:58:40 +0000329 std::string* ErrorMessage) {
330 std::auto_ptr<Archive> result ( new Archive(file, C));
Reid Spencer0ff2d312006-08-24 23:45:08 +0000331 if (result->mapToMemory(ErrorMessage))
332 return 0;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000333 if (!result->loadArchive(ErrorMessage))
Reid Spencer5af46882004-12-13 02:59:03 +0000334 return 0;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000335 return result.release();
Reid Spencerf9d7a512004-11-14 21:58:33 +0000336}
337
Gabor Greifa99be512007-07-05 17:07:56 +0000338// Get all the bitcode modules from the archive
Reid Spencerf9d7a512004-11-14 21:58:33 +0000339bool
Owen Anderson8b477ed2009-07-01 16:58:40 +0000340Archive::getAllModules(std::vector<Module*>& Modules,
341 std::string* ErrMessage) {
Reid Spencerf9d7a512004-11-14 21:58:33 +0000342
343 for (iterator I=begin(), E=end(); I != E; ++I) {
Gabor Greife75ca3d2007-07-06 13:38:17 +0000344 if (I->isBitcode()) {
Chris Lattner74382b72009-08-23 22:45:37 +0000345 std::string FullMemberName = archPath.str() +
346 "(" + I->getPath().str() + ")";
Chris Lattnerc1d56242007-05-06 09:28:33 +0000347 MemoryBuffer *Buffer =
Benjamin Kramer9d44e702010-04-19 16:15:31 +0000348 MemoryBuffer::getMemBufferCopy(StringRef(I->getData(), I->getSize()),
349 FullMemberName.c_str());
Chris Lattnere07c15c2007-05-06 06:18:07 +0000350
Owen Anderson8b477ed2009-07-01 16:58:40 +0000351 Module *M = ParseBitcodeFile(Buffer, Context, ErrMessage);
Chris Lattnerc1d56242007-05-06 09:28:33 +0000352 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
Owen Anderson31895e72009-07-01 21:22:36 +0000442Archive* Archive::OpenAndLoadSymbols(const sys::Path& file,
Owen Anderson4434ed42009-07-01 23:13:44 +0000443 LLVMContext& C,
Owen Anderson8b477ed2009-07-01 16:58:40 +0000444 std::string* ErrorMessage) {
445 std::auto_ptr<Archive> result ( new Archive(file, C) );
Reid Spencer0ff2d312006-08-24 23:45:08 +0000446 if (result->mapToMemory(ErrorMessage))
447 return 0;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000448 if (!result->loadSymbolTable(ErrorMessage))
Reid Spencer5af46882004-12-13 02:59:03 +0000449 return 0;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000450 return result.release();
Reid Spencerf9d7a512004-11-14 21:58:33 +0000451}
Chris Lattner968cfd02003-04-19 21:45:34 +0000452
Jeffrey Yasskinf0356fe2010-01-27 20:34:15 +0000453// Look up one symbol in the symbol table and return the module that defines
454// that symbol.
455Module*
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000456Archive::findModuleDefiningSymbol(const std::string& symbol,
457 std::string* ErrMsg) {
Reid Spencerf9d7a512004-11-14 21:58:33 +0000458 SymTabType::iterator SI = symTab.find(symbol);
459 if (SI == symTab.end())
460 return 0;
461
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000462 // The symbol table was previously constructed assuming that the members were
Reid Spencerf9d7a512004-11-14 21:58:33 +0000463 // written without the symbol table header. Because VBR encoding is used, the
464 // values could not be adjusted to account for the offset of the symbol table
465 // because that could affect the size of the symbol table due to VBR encoding.
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000466 // We now have to account for this by adjusting the offset by the size of the
Reid Spencerf9d7a512004-11-14 21:58:33 +0000467 // symbol table and its header.
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000468 unsigned fileOffset =
Reid Spencerf9d7a512004-11-14 21:58:33 +0000469 SI->second + // offset in symbol-table-less file
470 firstFileOffset; // add offset to first "real" file in archive
471
472 // See if the module is already loaded
473 ModuleMap::iterator MI = modules.find(fileOffset);
474 if (MI != modules.end())
475 return MI->second.first;
476
477 // Module hasn't been loaded yet, we need to load it
478 const char* modptr = base + fileOffset;
Chris Lattner7f6b4472008-04-01 04:26:46 +0000479 ArchiveMember* mbr = parseMemberHeader(modptr, mapfile->getBufferEnd(),
480 ErrMsg);
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000481 if (!mbr)
Reid Spencer0b5a5042006-08-25 17:43:11 +0000482 return 0;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000483
Jeffrey Yasskinf0356fe2010-01-27 20:34:15 +0000484 // Now, load the bitcode module to get the Module.
Chris Lattner74382b72009-08-23 22:45:37 +0000485 std::string FullMemberName = archPath.str() + "(" +
486 mbr->getPath().str() + ")";
Benjamin Kramer9d44e702010-04-19 16:15:31 +0000487 MemoryBuffer *Buffer =
488 MemoryBuffer::getMemBufferCopy(StringRef(mbr->getData(), mbr->getSize()),
489 FullMemberName.c_str());
Chris Lattnerc1d56242007-05-06 09:28:33 +0000490
Jeffrey Yasskinf0356fe2010-01-27 20:34:15 +0000491 Module *m = getLazyBitcodeModule(Buffer, Context, ErrMsg);
492 if (!m)
Reid Spencer0b5a5042006-08-25 17:43:11 +0000493 return 0;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000494
Jeffrey Yasskinf0356fe2010-01-27 20:34:15 +0000495 modules.insert(std::make_pair(fileOffset, std::make_pair(m, mbr)));
Reid Spencerf9d7a512004-11-14 21:58:33 +0000496
Jeffrey Yasskinf0356fe2010-01-27 20:34:15 +0000497 return m;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000498}
499
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000500// Look up multiple symbols in the symbol table and return a set of
Jeffrey Yasskinf0356fe2010-01-27 20:34:15 +0000501// Modules that define those symbols.
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000502bool
Reid Spencer7783e8a2004-11-19 03:18:22 +0000503Archive::findModulesDefiningSymbols(std::set<std::string>& symbols,
Rafael Espindola603d6b52012-01-23 03:41:53 +0000504 SmallVectorImpl<Module*>& result,
Chris Lattnerf2e292c2007-02-07 21:41:02 +0000505 std::string* error) {
Reid Spencercd5561a2006-12-15 19:44:51 +0000506 if (!mapfile || !base) {
507 if (error)
508 *error = "Empty archive invalid for finding modules defining symbols";
509 return false;
510 }
511
Reid Spencer766b7932004-11-15 01:20:11 +0000512 if (symTab.empty()) {
513 // We don't have a symbol table, so we must build it now but lets also
514 // make sure that we populate the modules table as we do this to ensure
515 // that we don't load them twice when findModuleDefiningSymbol is called
516 // below.
517
518 // Get a pointer to the first file
Chris Lattner7f6b4472008-04-01 04:26:46 +0000519 const char* At = base + firstFileOffset;
520 const char* End = mapfile->getBufferEnd();
Reid Spencer766b7932004-11-15 01:20:11 +0000521
522 while ( At < End) {
523 // Compute the offset to be put in the symbol table
524 unsigned offset = At - base - firstFileOffset;
525
526 // Parse the file's header
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000527 ArchiveMember* mbr = parseMemberHeader(At, End, error);
528 if (!mbr)
529 return false;
Reid Spencer766b7932004-11-15 01:20:11 +0000530
531 // If it contains symbols
Gabor Greife75ca3d2007-07-06 13:38:17 +0000532 if (mbr->isBitcode()) {
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000533 // Get the symbols
Reid Spencer766b7932004-11-15 01:20:11 +0000534 std::vector<std::string> symbols;
Chris Lattner74382b72009-08-23 22:45:37 +0000535 std::string FullMemberName = archPath.str() + "(" +
536 mbr->getPath().str() + ")";
Jeffrey Yasskinf0356fe2010-01-27 20:34:15 +0000537 Module* M =
Benjamin Kramer9d44e702010-04-19 16:15:31 +0000538 GetBitcodeSymbols(At, mbr->getSize(), FullMemberName, Context,
539 symbols, error);
Reid Spencer766b7932004-11-15 01:20:11 +0000540
Jeffrey Yasskinf0356fe2010-01-27 20:34:15 +0000541 if (M) {
Reid Spencer766b7932004-11-15 01:20:11 +0000542 // Insert the module's symbols into the symbol table
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000543 for (std::vector<std::string>::iterator I = symbols.begin(),
Reid Spencer766b7932004-11-15 01:20:11 +0000544 E=symbols.end(); I != E; ++I ) {
Reid Spencer9a29db42004-11-20 07:29:40 +0000545 symTab.insert(std::make_pair(*I, offset));
Reid Spencer766b7932004-11-15 01:20:11 +0000546 }
Jeffrey Yasskinf0356fe2010-01-27 20:34:15 +0000547 // Insert the Module and the ArchiveMember into the table of
Reid Spencer766b7932004-11-15 01:20:11 +0000548 // modules.
Jeffrey Yasskinf0356fe2010-01-27 20:34:15 +0000549 modules.insert(std::make_pair(offset, std::make_pair(M, mbr)));
Reid Spencer766b7932004-11-15 01:20:11 +0000550 } else {
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000551 if (error)
Gabor Greifa99be512007-07-05 17:07:56 +0000552 *error = "Can't parse bitcode member: " +
Chris Lattner74382b72009-08-23 22:45:37 +0000553 mbr->getPath().str() + ": " + *error;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000554 delete mbr;
555 return false;
Reid Spencer766b7932004-11-15 01:20:11 +0000556 }
557 }
Reid Spencerb3231132004-11-15 01:40:20 +0000558
559 // Go to the next file location
560 At += mbr->getSize();
Reid Spencer6405c9e2004-11-19 17:08:00 +0000561 if ((intptr_t(At) & 1) == 1)
Reid Spencerb3231132004-11-15 01:40:20 +0000562 At++;
Reid Spencer766b7932004-11-15 01:20:11 +0000563 }
564 }
565
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000566 // At this point we have a valid symbol table (one way or another) so we
Reid Spencer766b7932004-11-15 01:20:11 +0000567 // just use it to quickly find the symbols requested.
568
Rafael Espindola603d6b52012-01-23 03:41:53 +0000569 SmallPtrSet<Module*, 16> Added;
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000570 for (std::set<std::string>::iterator I=symbols.begin(),
Rafael Espindola4254df32012-01-23 05:07:16 +0000571 Next = I,
572 E=symbols.end(); I != E; I = Next) {
573 // Increment Next before we invalidate it.
574 ++Next;
575
Reid Spencer7783e8a2004-11-19 03:18:22 +0000576 // See if this symbol exists
Jeffrey Yasskinf0356fe2010-01-27 20:34:15 +0000577 Module* m = findModuleDefiningSymbol(*I,error);
Rafael Espindola603d6b52012-01-23 03:41:53 +0000578 if (!m)
579 continue;
580 bool NewMember = Added.insert(m);
581 if (!NewMember)
582 continue;
Reid Spencer7783e8a2004-11-19 03:18:22 +0000583
Rafael Espindola603d6b52012-01-23 03:41:53 +0000584 // The symbol exists, insert the Module into our result.
585 result.push_back(m);
586
587 // Remove the symbol now that its been resolved.
588 symbols.erase(I);
Reid Spencerf9d7a512004-11-14 21:58:33 +0000589 }
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000590 return true;
Chris Lattner968cfd02003-04-19 21:45:34 +0000591}
Reid Spencereaa06bb2005-02-26 22:00:32 +0000592
Gabor Greifa99be512007-07-05 17:07:56 +0000593bool Archive::isBitcodeArchive() {
Chris Lattner7c439922005-09-23 06:22:58 +0000594 // Make sure the symTab has been loaded. In most cases this should have been
595 // done when the archive was constructed, but still, this is just in case.
Dan Gohman30359592008-01-29 13:02:09 +0000596 if (symTab.empty())
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000597 if (!loadSymbolTable(0))
598 return false;
Reid Spencereaa06bb2005-02-26 22:00:32 +0000599
Chris Lattner7c439922005-09-23 06:22:58 +0000600 // Now that we know it's been loaded, return true
601 // if it has a size
602 if (symTab.size()) return true;
Reid Spencereaa06bb2005-02-26 22:00:32 +0000603
Gabor Greifa99be512007-07-05 17:07:56 +0000604 // We still can't be sure it isn't a bitcode archive
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000605 if (!loadArchive(0))
606 return false;
Reid Spencereaa06bb2005-02-26 22:00:32 +0000607
608 std::vector<Module *> Modules;
609 std::string ErrorMessage;
610
Gabor Greifa99be512007-07-05 17:07:56 +0000611 // Scan the archive, trying to load a bitcode member. We only load one to
Chris Lattner7c439922005-09-23 06:22:58 +0000612 // see if this works.
613 for (iterator I = begin(), E = end(); I != E; ++I) {
Gabor Greife75ca3d2007-07-06 13:38:17 +0000614 if (!I->isBitcode())
Chris Lattner7c439922005-09-23 06:22:58 +0000615 continue;
616
617 std::string FullMemberName =
Chris Lattner74382b72009-08-23 22:45:37 +0000618 archPath.str() + "(" + I->getPath().str() + ")";
Chris Lattnerc1d56242007-05-06 09:28:33 +0000619
620 MemoryBuffer *Buffer =
Benjamin Kramer9d44e702010-04-19 16:15:31 +0000621 MemoryBuffer::getMemBufferCopy(StringRef(I->getData(), I->getSize()),
622 FullMemberName.c_str());
Owen Anderson8b477ed2009-07-01 16:58:40 +0000623 Module *M = ParseBitcodeFile(Buffer, Context);
Chris Lattnerc1d56242007-05-06 09:28:33 +0000624 delete Buffer;
Chris Lattner7c439922005-09-23 06:22:58 +0000625 if (!M)
Gabor Greifa99be512007-07-05 17:07:56 +0000626 return false; // Couldn't parse bitcode, not a bitcode archive.
Chris Lattner7c439922005-09-23 06:22:58 +0000627 delete M;
628 return true;
629 }
630
631 return false;
Reid Spencereaa06bb2005-02-26 22:00:32 +0000632}