blob: 13a92a95ca20961d3ba79f85277502cd79542f4a [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
Reid Spencer8d8a7ff2006-07-07 20:56:50 +000022inline 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
221 member->next = 0;
222 member->prev = 0;
223 member->parent = this;
Reid Spencerdd04df02005-07-07 23:21:43 +0000224 member->path.set(pathname);
Reid Spencerf9d7a512004-11-14 21:58:33 +0000225 member->info.fileSize = MemberSize;
226 member->info.modTime.fromEpochTime(atoi(Hdr->date));
Reid Spencer56127622004-12-29 01:20:24 +0000227 unsigned int mode;
228 sscanf(Hdr->mode, "%o", &mode);
229 member->info.mode = mode;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000230 member->info.user = atoi(Hdr->uid);
231 member->info.group = atoi(Hdr->gid);
232 member->flags = flags;
233 member->data = At;
234
235 return member;
236}
237
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000238bool
239Archive::checkSignature(std::string* error) {
Reid Spencerf9d7a512004-11-14 21:58:33 +0000240 // Check the magic string at file's header
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000241 if (mapfile->size() < 8 || memcmp(base, ARFILE_MAGIC, 8)) {
242 if (error)
243 *error = "invalid signature for an archive file";
244 return false;
245 }
246 return true;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000247}
248
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000249// This function loads the entire archive and fully populates its ilist with
Reid Spencerf9d7a512004-11-14 21:58:33 +0000250// the members of the archive file. This is typically used in preparation for
251// editing the contents of the archive.
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000252bool
253Archive::loadArchive(std::string* error) {
Reid Spencerf9d7a512004-11-14 21:58:33 +0000254
255 // Set up parsing
256 members.clear();
257 symTab.clear();
258 const char *At = base;
259 const char *End = base + mapfile->size();
260
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000261 if (!checkSignature(error))
262 return false;
263
Reid Spencerf9d7a512004-11-14 21:58:33 +0000264 At += 8; // Skip the magic string.
265
266 bool seenSymbolTable = false;
267 bool foundFirstFile = false;
268 while (At < End) {
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000269 // parse the member header
Reid Spencerf9d7a512004-11-14 21:58:33 +0000270 const char* Save = At;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000271 ArchiveMember* mbr = parseMemberHeader(At, End, error);
272 if (!mbr)
273 return false;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000274
275 // check if this is the foreign symbol table
Reid Spencer9a29db42004-11-20 07:29:40 +0000276 if (mbr->isSVR4SymbolTable() || mbr->isBSD4SymbolTable()) {
Reid Spencer4a980d12004-11-16 06:47:19 +0000277 // We just save this but don't do anything special
278 // with it. It doesn't count as the "first file".
Reid Spencer9a29db42004-11-20 07:29:40 +0000279 if (foreignST) {
280 // What? Multiple foreign symbol tables? Just chuck it
281 // and retain the last one found.
282 delete foreignST;
283 }
Reid Spencer4a980d12004-11-16 06:47:19 +0000284 foreignST = mbr;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000285 At += mbr->getSize();
Reid Spencer6405c9e2004-11-19 17:08:00 +0000286 if ((intptr_t(At) & 1) == 1)
Reid Spencerf9d7a512004-11-14 21:58:33 +0000287 At++;
288 } else if (mbr->isStringTable()) {
Reid Spencer4a980d12004-11-16 06:47:19 +0000289 // Simply suck the entire string table into a string
290 // variable. This will be used to get the names of the
291 // members that use the "/ddd" format for their names
292 // (SVR4 style long names).
Reid Spencer9a29db42004-11-20 07:29:40 +0000293 strtab.assign(At, mbr->getSize());
Reid Spencerf9d7a512004-11-14 21:58:33 +0000294 At += mbr->getSize();
Reid Spencer6405c9e2004-11-19 17:08:00 +0000295 if ((intptr_t(At) & 1) == 1)
Reid Spencerf9d7a512004-11-14 21:58:33 +0000296 At++;
297 delete mbr;
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000298 } else if (mbr->isLLVMSymbolTable()) {
Reid Spencer4a980d12004-11-16 06:47:19 +0000299 // This is the LLVM symbol table for the archive. If we've seen it
300 // already, its an error. Otherwise, parse the symbol table and move on.
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000301 if (seenSymbolTable) {
302 if (error)
303 *error = "invalid archive: multiple symbol tables";
304 return false;
305 }
306 if (!parseSymbolTable(mbr->getData(), mbr->getSize(), error))
307 return false;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000308 seenSymbolTable = true;
309 At += mbr->getSize();
Reid Spencer6405c9e2004-11-19 17:08:00 +0000310 if ((intptr_t(At) & 1) == 1)
Reid Spencerf9d7a512004-11-14 21:58:33 +0000311 At++;
Reid Spencer4a980d12004-11-16 06:47:19 +0000312 delete mbr; // We don't need this member in the list of members.
Reid Spencerf9d7a512004-11-14 21:58:33 +0000313 } else {
Reid Spencer4a980d12004-11-16 06:47:19 +0000314 // This is just a regular file. If its the first one, save its offset.
315 // Otherwise just push it on the list and move on to the next file.
Reid Spencerf9d7a512004-11-14 21:58:33 +0000316 if (!foundFirstFile) {
317 firstFileOffset = Save - base;
318 foundFirstFile = true;
319 }
320 members.push_back(mbr);
321 At += mbr->getSize();
Reid Spencer6405c9e2004-11-19 17:08:00 +0000322 if ((intptr_t(At) & 1) == 1)
Reid Spencerf9d7a512004-11-14 21:58:33 +0000323 At++;
324 }
325 }
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000326 return true;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000327}
328
329// Open and completely load the archive file.
330Archive*
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000331Archive::OpenAndLoad(const sys::Path& file, std::string* ErrorMessage)
332{
Reid Spencer0ff2d312006-08-24 23:45:08 +0000333 std::auto_ptr<Archive> result ( new Archive(file));
334 if (result->mapToMemory(ErrorMessage))
335 return 0;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000336 if (!result->loadArchive(ErrorMessage))
Reid Spencer5af46882004-12-13 02:59:03 +0000337 return 0;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000338 return result.release();
Reid Spencerf9d7a512004-11-14 21:58:33 +0000339}
340
Gabor Greifa99be512007-07-05 17:07:56 +0000341// Get all the bitcode modules from the archive
Reid Spencerf9d7a512004-11-14 21:58:33 +0000342bool
343Archive::getAllModules(std::vector<Module*>& Modules, std::string* ErrMessage) {
344
345 for (iterator I=begin(), E=end(); I != E; ++I) {
Gabor Greife75ca3d2007-07-06 13:38:17 +0000346 if (I->isBitcode()) {
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000347 std::string FullMemberName = archPath.toString() +
Reid Spencer1fce0912004-12-11 00:14:15 +0000348 "(" + I->getPath().toString() + ")";
Chris Lattnerc1d56242007-05-06 09:28:33 +0000349 MemoryBuffer *Buffer =
350 MemoryBuffer::getNewMemBuffer(I->getSize(), FullMemberName.c_str());
351 memcpy((char*)Buffer->getBufferStart(), I->getData(), I->getSize());
Chris Lattnere07c15c2007-05-06 06:18:07 +0000352
Chris Lattnerc1d56242007-05-06 09:28:33 +0000353 Module *M = ParseBitcodeFile(Buffer, ErrMessage);
354 delete Buffer;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000355 if (!M)
356 return true;
357
358 Modules.push_back(M);
359 }
360 }
Brian Gaeke2c61d7b2003-11-16 23:08:48 +0000361 return false;
362}
Chris Lattner968cfd02003-04-19 21:45:34 +0000363
Reid Spencerf9d7a512004-11-14 21:58:33 +0000364// Load just the symbol table from the archive file
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000365bool
366Archive::loadSymbolTable(std::string* ErrorMsg) {
Chris Lattner968cfd02003-04-19 21:45:34 +0000367
Reid Spencerf9d7a512004-11-14 21:58:33 +0000368 // Set up parsing
369 members.clear();
370 symTab.clear();
371 const char *At = base;
372 const char *End = base + mapfile->size();
Chris Lattner968cfd02003-04-19 21:45:34 +0000373
Reid Spencerf9d7a512004-11-14 21:58:33 +0000374 // Make sure we're dealing with an archive
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000375 if (!checkSignature(ErrorMsg))
376 return false;
Chris Lattner968cfd02003-04-19 21:45:34 +0000377
Reid Spencerf9d7a512004-11-14 21:58:33 +0000378 At += 8; // Skip signature
379
380 // Parse the first file member header
381 const char* FirstFile = At;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000382 ArchiveMember* mbr = parseMemberHeader(At, End, ErrorMsg);
383 if (!mbr)
384 return false;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000385
Reid Spencer9a29db42004-11-20 07:29:40 +0000386 if (mbr->isSVR4SymbolTable() || mbr->isBSD4SymbolTable()) {
Reid Spencerf9d7a512004-11-14 21:58:33 +0000387 // Skip the foreign symbol table, we don't do anything with it
388 At += mbr->getSize();
Reid Spencer6405c9e2004-11-19 17:08:00 +0000389 if ((intptr_t(At) & 1) == 1)
Reid Spencerb3231132004-11-15 01:40:20 +0000390 At++;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000391 delete mbr;
392
Reid Spencerb3231132004-11-15 01:40:20 +0000393 // Read the next one
Reid Spencerf9d7a512004-11-14 21:58:33 +0000394 FirstFile = At;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000395 mbr = parseMemberHeader(At, End, ErrorMsg);
396 if (!mbr) {
397 delete mbr;
398 return false;
399 }
Reid Spencerb3231132004-11-15 01:40:20 +0000400 }
401
402 if (mbr->isStringTable()) {
403 // Process the string table entry
Reid Spencer9a29db42004-11-20 07:29:40 +0000404 strtab.assign((const char*)mbr->getData(), mbr->getSize());
Reid Spencerb3231132004-11-15 01:40:20 +0000405 At += mbr->getSize();
Reid Spencer6405c9e2004-11-19 17:08:00 +0000406 if ((intptr_t(At) & 1) == 1)
Reid Spencerb3231132004-11-15 01:40:20 +0000407 At++;
408 delete mbr;
409 // Get the next one
410 FirstFile = At;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000411 mbr = parseMemberHeader(At, End, ErrorMsg);
412 if (!mbr) {
413 delete mbr;
414 return false;
415 }
Chris Lattner968cfd02003-04-19 21:45:34 +0000416 }
417
Reid Spencerf9d7a512004-11-14 21:58:33 +0000418 // See if its the symbol table
419 if (mbr->isLLVMSymbolTable()) {
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000420 if (!parseSymbolTable(mbr->getData(), mbr->getSize(), ErrorMsg)) {
421 delete mbr;
422 return false;
423 }
424
Reid Spencer8dde18f2004-11-28 03:13:02 +0000425 At += mbr->getSize();
Reid Spencer6405c9e2004-11-19 17:08:00 +0000426 if ((intptr_t(At) & 1) == 1)
Reid Spencer8dde18f2004-11-28 03:13:02 +0000427 At++;
Reid Spencerffb4d622006-05-12 17:56:20 +0000428 delete mbr;
429 // Can't be any more symtab headers so just advance
Reid Spencer8dde18f2004-11-28 03:13:02 +0000430 FirstFile = At;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000431 } else {
432 // There's no symbol table in the file. We have to rebuild it from scratch
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000433 // because the intent of this method is to get the symbol table loaded so
434 // it can be searched efficiently.
Reid Spencerf9d7a512004-11-14 21:58:33 +0000435 // Add the member to the members list
436 members.push_back(mbr);
437 }
438
439 firstFileOffset = FirstFile - base;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000440 return true;
Chris Lattner968cfd02003-04-19 21:45:34 +0000441}
442
Reid Spencerf9d7a512004-11-14 21:58:33 +0000443// Open the archive and load just the symbol tables
444Archive*
Reid Spencer5af46882004-12-13 02:59:03 +0000445Archive::OpenAndLoadSymbols(const sys::Path& file, std::string* ErrorMessage) {
Reid Spencer0ff2d312006-08-24 23:45:08 +0000446 std::auto_ptr<Archive> result ( new Archive(file) );
447 if (result->mapToMemory(ErrorMessage))
448 return 0;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000449 if (!result->loadSymbolTable(ErrorMessage))
Reid Spencer5af46882004-12-13 02:59:03 +0000450 return 0;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000451 return result.release();
Reid Spencerf9d7a512004-11-14 21:58:33 +0000452}
Chris Lattner968cfd02003-04-19 21:45:34 +0000453
Reid Spencerf9d7a512004-11-14 21:58:33 +0000454// Look up one symbol in the symbol table and return a ModuleProvider for the
455// module that defines that symbol.
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000456ModuleProvider*
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000457Archive::findModuleDefiningSymbol(const std::string& symbol,
458 std::string* ErrMsg) {
Reid Spencerf9d7a512004-11-14 21:58:33 +0000459 SymTabType::iterator SI = symTab.find(symbol);
460 if (SI == symTab.end())
461 return 0;
462
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000463 // The symbol table was previously constructed assuming that the members were
Reid Spencerf9d7a512004-11-14 21:58:33 +0000464 // written without the symbol table header. Because VBR encoding is used, the
465 // values could not be adjusted to account for the offset of the symbol table
466 // because that could affect the size of the symbol table due to VBR encoding.
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000467 // We now have to account for this by adjusting the offset by the size of the
Reid Spencerf9d7a512004-11-14 21:58:33 +0000468 // symbol table and its header.
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000469 unsigned fileOffset =
Reid Spencerf9d7a512004-11-14 21:58:33 +0000470 SI->second + // offset in symbol-table-less file
471 firstFileOffset; // add offset to first "real" file in archive
472
473 // See if the module is already loaded
474 ModuleMap::iterator MI = modules.find(fileOffset);
475 if (MI != modules.end())
476 return MI->second.first;
477
478 // Module hasn't been loaded yet, we need to load it
479 const char* modptr = base + fileOffset;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000480 ArchiveMember* mbr = parseMemberHeader(modptr, base + mapfile->size(),ErrMsg);
481 if (!mbr)
Reid Spencer0b5a5042006-08-25 17:43:11 +0000482 return 0;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000483
Gabor Greifa99be512007-07-05 17:07:56 +0000484 // Now, load the bitcode module to get the ModuleProvider
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000485 std::string FullMemberName = archPath.toString() + "(" +
Reid Spencer1fce0912004-12-11 00:14:15 +0000486 mbr->getPath().toString() + ")";
Chris Lattnerc1d56242007-05-06 09:28:33 +0000487 MemoryBuffer *Buffer =MemoryBuffer::getNewMemBuffer(mbr->getSize(),
488 FullMemberName.c_str());
489 memcpy((char*)Buffer->getBufferStart(), mbr->getData(), mbr->getSize());
490
491 ModuleProvider *mp = getBitcodeModuleProvider(Buffer, ErrMsg);
Reid Spencer0b5a5042006-08-25 17:43:11 +0000492 if (!mp)
493 return 0;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000494
Reid Spencer9a29db42004-11-20 07:29:40 +0000495 modules.insert(std::make_pair(fileOffset, std::make_pair(mp, mbr)));
Reid Spencerf9d7a512004-11-14 21:58:33 +0000496
497 return mp;
498}
499
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000500// Look up multiple symbols in the symbol table and return a set of
Reid Spencerf9d7a512004-11-14 21:58:33 +0000501// ModuleProviders that define those symbols.
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000502bool
Reid Spencer7783e8a2004-11-19 03:18:22 +0000503Archive::findModulesDefiningSymbols(std::set<std::string>& symbols,
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000504 std::set<ModuleProvider*>& result,
Chris Lattnerf2e292c2007-02-07 21:41:02 +0000505 std::string* error) {
Reid Spencercd5561a2006-12-15 19:44:51 +0000506 if (!mapfile || !base) {
507 if (error)
508 *error = "Empty archive invalid for finding modules defining symbols";
509 return false;
510 }
511
Reid Spencer766b7932004-11-15 01:20:11 +0000512 if (symTab.empty()) {
513 // We don't have a symbol table, so we must build it now but lets also
514 // make sure that we populate the modules table as we do this to ensure
515 // that we don't load them twice when findModuleDefiningSymbol is called
516 // below.
517
518 // Get a pointer to the first file
519 const char* At = ((const char*)base) + firstFileOffset;
520 const char* End = ((const char*)base) + mapfile->size();
521
522 while ( At < End) {
523 // Compute the offset to be put in the symbol table
524 unsigned offset = At - base - firstFileOffset;
525
526 // Parse the file's header
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000527 ArchiveMember* mbr = parseMemberHeader(At, End, error);
528 if (!mbr)
529 return false;
Reid Spencer766b7932004-11-15 01:20:11 +0000530
531 // If it contains symbols
Gabor Greife75ca3d2007-07-06 13:38:17 +0000532 if (mbr->isBitcode()) {
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000533 // Get the symbols
Reid Spencer766b7932004-11-15 01:20:11 +0000534 std::vector<std::string> symbols;
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000535 std::string FullMemberName = archPath.toString() + "(" +
Reid Spencer1fce0912004-12-11 00:14:15 +0000536 mbr->getPath().toString() + ")";
Chris Lattnerf2e292c2007-02-07 21:41:02 +0000537 ModuleProvider* MP =
Gabor Greifa99be512007-07-05 17:07:56 +0000538 GetBitcodeSymbols((const unsigned char*)At, mbr->getSize(),
539 FullMemberName, symbols, error);
Reid Spencer766b7932004-11-15 01:20:11 +0000540
541 if (MP) {
542 // Insert the module's symbols into the symbol table
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000543 for (std::vector<std::string>::iterator I = symbols.begin(),
Reid Spencer766b7932004-11-15 01:20:11 +0000544 E=symbols.end(); I != E; ++I ) {
Reid Spencer9a29db42004-11-20 07:29:40 +0000545 symTab.insert(std::make_pair(*I, offset));
Reid Spencer766b7932004-11-15 01:20:11 +0000546 }
547 // Insert the ModuleProvider and the ArchiveMember into the table of
548 // modules.
Reid Spencer9a29db42004-11-20 07:29:40 +0000549 modules.insert(std::make_pair(offset, std::make_pair(MP, mbr)));
Reid Spencer766b7932004-11-15 01:20:11 +0000550 } else {
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000551 if (error)
Gabor Greifa99be512007-07-05 17:07:56 +0000552 *error = "Can't parse bitcode member: " +
Reid Spencer0b5a5042006-08-25 17:43:11 +0000553 mbr->getPath().toString() + ": " + *error;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000554 delete mbr;
555 return false;
Reid Spencer766b7932004-11-15 01:20:11 +0000556 }
557 }
Reid Spencerb3231132004-11-15 01:40:20 +0000558
559 // Go to the next file location
560 At += mbr->getSize();
Reid Spencer6405c9e2004-11-19 17:08:00 +0000561 if ((intptr_t(At) & 1) == 1)
Reid Spencerb3231132004-11-15 01:40:20 +0000562 At++;
Reid Spencer766b7932004-11-15 01:20:11 +0000563 }
564 }
565
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000566 // At this point we have a valid symbol table (one way or another) so we
Reid Spencer766b7932004-11-15 01:20:11 +0000567 // just use it to quickly find the symbols requested.
568
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000569 for (std::set<std::string>::iterator I=symbols.begin(),
Reid Spencer7783e8a2004-11-19 03:18:22 +0000570 E=symbols.end(); I != E;) {
571 // See if this symbol exists
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000572 ModuleProvider* mp = findModuleDefiningSymbol(*I,error);
Reid Spencerf9d7a512004-11-14 21:58:33 +0000573 if (mp) {
Reid Spencer7783e8a2004-11-19 03:18:22 +0000574 // The symbol exists, insert the ModuleProvider into our result,
575 // duplicates wil be ignored
Reid Spencer766b7932004-11-15 01:20:11 +0000576 result.insert(mp);
Reid Spencer7783e8a2004-11-19 03:18:22 +0000577
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000578 // Remove the symbol now that its been resolved, being careful to
Reid Spencer57646ec2004-11-19 03:44:10 +0000579 // post-increment the iterator.
580 symbols.erase(I++);
Reid Spencer7783e8a2004-11-19 03:18:22 +0000581 } else {
582 ++I;
Chris Lattner968cfd02003-04-19 21:45:34 +0000583 }
Reid Spencerf9d7a512004-11-14 21:58:33 +0000584 }
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000585 return true;
Chris Lattner968cfd02003-04-19 21:45:34 +0000586}
Reid Spencereaa06bb2005-02-26 22:00:32 +0000587
Gabor Greifa99be512007-07-05 17:07:56 +0000588bool Archive::isBitcodeArchive() {
Chris Lattner7c439922005-09-23 06:22:58 +0000589 // Make sure the symTab has been loaded. In most cases this should have been
590 // done when the archive was constructed, but still, this is just in case.
591 if (!symTab.size())
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000592 if (!loadSymbolTable(0))
593 return false;
Reid Spencereaa06bb2005-02-26 22:00:32 +0000594
Chris Lattner7c439922005-09-23 06:22:58 +0000595 // Now that we know it's been loaded, return true
596 // if it has a size
597 if (symTab.size()) return true;
Reid Spencereaa06bb2005-02-26 22:00:32 +0000598
Gabor Greifa99be512007-07-05 17:07:56 +0000599 // We still can't be sure it isn't a bitcode archive
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000600 if (!loadArchive(0))
601 return false;
Reid Spencereaa06bb2005-02-26 22:00:32 +0000602
603 std::vector<Module *> Modules;
604 std::string ErrorMessage;
605
Gabor Greifa99be512007-07-05 17:07:56 +0000606 // Scan the archive, trying to load a bitcode member. We only load one to
Chris Lattner7c439922005-09-23 06:22:58 +0000607 // see if this works.
608 for (iterator I = begin(), E = end(); I != E; ++I) {
Gabor Greife75ca3d2007-07-06 13:38:17 +0000609 if (!I->isBitcode())
Chris Lattner7c439922005-09-23 06:22:58 +0000610 continue;
611
612 std::string FullMemberName =
613 archPath.toString() + "(" + I->getPath().toString() + ")";
Chris Lattnerc1d56242007-05-06 09:28:33 +0000614
615 MemoryBuffer *Buffer =
616 MemoryBuffer::getNewMemBuffer(I->getSize(), FullMemberName.c_str());
617 memcpy((char*)Buffer->getBufferStart(), I->getData(), I->getSize());
618 Module *M = ParseBitcodeFile(Buffer);
619 delete Buffer;
Chris Lattner7c439922005-09-23 06:22:58 +0000620 if (!M)
Gabor Greifa99be512007-07-05 17:07:56 +0000621 return false; // Couldn't parse bitcode, not a bitcode archive.
Chris Lattner7c439922005-09-23 06:22:58 +0000622 delete M;
623 return true;
624 }
625
626 return false;
Reid Spencereaa06bb2005-02-26 22:00:32 +0000627}