blob: 82ff9edc8ae427b9688ab2409b24578e46a230bf [file] [log] [blame]
Reid Spencerf9d7a512004-11-14 21:58:33 +00001//===-- ArchiveReader.cpp - Read LLVM archive files -------------*- C++ -*-===//
Misha Brukman2b37d7c2005-04-21 21:13:18 +00002//
John Criswellb576c942003-10-20 19:43:21 +00003// The LLVM Compiler Infrastructure
4//
Misha Brukman2b37d7c2005-04-21 21:13:18 +00005// This file was developed by Reid Spencer and is distributed under the
Reid Spencerf9d7a512004-11-14 21:58:33 +00006// University of Illinois Open Source License. See LICENSE.TXT for details.
Misha Brukman2b37d7c2005-04-21 21:13:18 +00007//
John Criswellb576c942003-10-20 19:43:21 +00008//===----------------------------------------------------------------------===//
Chris Lattner968cfd02003-04-19 21:45:34 +00009//
Reid Spencerf9d7a512004-11-14 21:58:33 +000010// Builds up standard unix archive files (.a) containing LLVM bytecode.
Chris Lattner968cfd02003-04-19 21:45:34 +000011//
12//===----------------------------------------------------------------------===//
13
Reid Spencerf9d7a512004-11-14 21:58:33 +000014#include "ArchiveInternals.h"
Chris Lattner968cfd02003-04-19 21:45:34 +000015#include "llvm/Bytecode/Reader.h"
Duraid Madina87487462005-12-26 14:31:26 +000016#include <memory>
Reid Spencerf9d7a512004-11-14 21:58:33 +000017
Chris Lattner3446ae82004-01-10 19:00:15 +000018using namespace llvm;
Brian Gaeked0fde302003-11-11 22:41:34 +000019
Reid Spencerf9d7a512004-11-14 21:58:33 +000020/// Read a variable-bit-rate encoded unsigned integer
Reid Spencer8d8a7ff2006-07-07 20:56:50 +000021inline unsigned readInteger(const char*&At, const char*End){
Reid Spencerf9d7a512004-11-14 21:58:33 +000022 unsigned Shift = 0;
23 unsigned Result = 0;
Misha Brukman2b37d7c2005-04-21 21:13:18 +000024
Reid Spencerf9d7a512004-11-14 21:58:33 +000025 do {
Misha Brukman2b37d7c2005-04-21 21:13:18 +000026 if (At == End)
Reid Spencer8d8a7ff2006-07-07 20:56:50 +000027 return Result;
Reid Spencerf9d7a512004-11-14 21:58:33 +000028 Result |= (unsigned)((*At++) & 0x7F) << Shift;
29 Shift += 7;
30 } while (At[-1] & 0x80);
31 return Result;
Chris Lattner968cfd02003-04-19 21:45:34 +000032}
33
Reid Spencerf9d7a512004-11-14 21:58:33 +000034// Completely parse the Archive's symbol table and populate symTab member var.
Reid Spencer8d8a7ff2006-07-07 20:56:50 +000035bool
36Archive::parseSymbolTable(const void* data, unsigned size, std::string* error) {
Reid Spencerf9d7a512004-11-14 21:58:33 +000037 const char* At = (const char*) data;
38 const char* End = At + size;
39 while (At < End) {
40 unsigned offset = readInteger(At, End);
Reid Spencer8d8a7ff2006-07-07 20:56:50 +000041 if (At == End) {
42 if (error)
43 *error = "Ran out of data reading vbr_uint for symtab offset!";
44 return false;
45 }
Reid Spencerf9d7a512004-11-14 21:58:33 +000046 unsigned length = readInteger(At, End);
Reid Spencer8d8a7ff2006-07-07 20:56:50 +000047 if (At == End) {
48 if (error)
49 *error = "Ran out of data reading vbr_uint for symtab length!";
50 return false;
51 }
52 if (At + length > End) {
53 if (error)
54 *error = "Malformed symbol table: length not consistent with size";
55 return false;
56 }
Reid Spencerf9d7a512004-11-14 21:58:33 +000057 // we don't care if it can't be inserted (duplicate entry)
Reid Spencer9a29db42004-11-20 07:29:40 +000058 symTab.insert(std::make_pair(std::string(At, length), offset));
Reid Spencerf9d7a512004-11-14 21:58:33 +000059 At += length;
60 }
61 symTabSize = size;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +000062 return true;
Chris Lattner968cfd02003-04-19 21:45:34 +000063}
64
Reid Spencerf9d7a512004-11-14 21:58:33 +000065// This member parses an ArchiveMemberHeader that is presumed to be pointed to
66// by At. The At pointer is updated to the byte just after the header, which
Misha Brukman2b37d7c2005-04-21 21:13:18 +000067// can be variable in size.
Reid Spencerf9d7a512004-11-14 21:58:33 +000068ArchiveMember*
Reid Spencer8d8a7ff2006-07-07 20:56:50 +000069Archive::parseMemberHeader(const char*& At, const char* End, std::string* error)
70{
Reid Spencercd5561a2006-12-15 19:44:51 +000071 if (At + sizeof(ArchiveMemberHeader) >= End) {
72 if (error)
73 *error = "Unexpected end of file";
74 return 0;
75 }
Reid Spencerf9d7a512004-11-14 21:58:33 +000076
77 // Cast archive member header
78 ArchiveMemberHeader* Hdr = (ArchiveMemberHeader*)At;
79 At += sizeof(ArchiveMemberHeader);
80
Misha Brukman2b37d7c2005-04-21 21:13:18 +000081 // Extract the size and determine if the file is
Reid Spencerf9d7a512004-11-14 21:58:33 +000082 // compressed or not (negative length).
83 int flags = 0;
84 int MemberSize = atoi(Hdr->size);
85 if (MemberSize < 0) {
86 flags |= ArchiveMember::CompressedFlag;
87 MemberSize = -MemberSize;
88 }
89
90 // Check the size of the member for sanity
Reid Spencer8d8a7ff2006-07-07 20:56:50 +000091 if (At + MemberSize > End) {
92 if (error)
93 *error = "invalid member length in archive file";
94 return 0;
95 }
Reid Spencerf9d7a512004-11-14 21:58:33 +000096
97 // Check the member signature
Reid Spencer8d8a7ff2006-07-07 20:56:50 +000098 if (!Hdr->checkSignature()) {
99 if (error)
100 *error = "invalid file member signature";
101 return 0;
102 }
Reid Spencerf9d7a512004-11-14 21:58:33 +0000103
104 // Convert and check the member name
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000105 // The empty name ( '/' and 15 blanks) is for a foreign (non-LLVM) symbol
106 // table. The special name "//" and 14 blanks is for a string table, used
Reid Spencerf9d7a512004-11-14 21:58:33 +0000107 // for long file names. This library doesn't generate either of those but
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000108 // it will accept them. If the name starts with #1/ and the remainder is
109 // digits, then those digits specify the length of the name that is
110 // stored immediately following the header. The special name
111 // __LLVM_SYM_TAB__ identifies the symbol table for LLVM bytecode.
112 // Anything else is a regular, short filename that is terminated with
Reid Spencerf9d7a512004-11-14 21:58:33 +0000113 // a '/' and blanks.
114
115 std::string pathname;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000116 switch (Hdr->name[0]) {
117 case '#':
118 if (Hdr->name[1] == '1' && Hdr->name[2] == '/') {
119 if (isdigit(Hdr->name[3])) {
120 unsigned len = atoi(&Hdr->name[3]);
Reid Spencer9a29db42004-11-20 07:29:40 +0000121 pathname.assign(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:
Reid Spencer9a29db42004-11-20 07:29:40 +0000199 char* slash = (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
206 // Determine if this is a bytecode file
Reid Spencer9a29db42004-11-20 07:29:40 +0000207 switch (sys::IdentifyFileType(At, 4)) {
Reid Spencerf9d7a512004-11-14 21:58:33 +0000208 case sys::BytecodeFileType:
209 flags |= ArchiveMember::BytecodeFlag;
210 break;
211 case sys::CompressedBytecodeFileType:
212 flags |= ArchiveMember::CompressedBytecodeFlag;
213 flags &= ~ArchiveMember::CompressedFlag;
214 break;
215 default:
216 flags &= ~(ArchiveMember::BytecodeFlag|
217 ArchiveMember::CompressedBytecodeFlag);
218 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
225 member->next = 0;
226 member->prev = 0;
227 member->parent = this;
Reid Spencerdd04df02005-07-07 23:21:43 +0000228 member->path.set(pathname);
Reid Spencerf9d7a512004-11-14 21:58:33 +0000229 member->info.fileSize = MemberSize;
230 member->info.modTime.fromEpochTime(atoi(Hdr->date));
Reid Spencer56127622004-12-29 01:20:24 +0000231 unsigned int mode;
232 sscanf(Hdr->mode, "%o", &mode);
233 member->info.mode = mode;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000234 member->info.user = atoi(Hdr->uid);
235 member->info.group = atoi(Hdr->gid);
236 member->flags = flags;
237 member->data = At;
238
239 return member;
240}
241
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000242bool
243Archive::checkSignature(std::string* error) {
Reid Spencerf9d7a512004-11-14 21:58:33 +0000244 // Check the magic string at file's header
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000245 if (mapfile->size() < 8 || memcmp(base, ARFILE_MAGIC, 8)) {
246 if (error)
247 *error = "invalid signature for an archive file";
248 return false;
249 }
250 return true;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000251}
252
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000253// This function loads the entire archive and fully populates its ilist with
Reid Spencerf9d7a512004-11-14 21:58:33 +0000254// the members of the archive file. This is typically used in preparation for
255// editing the contents of the archive.
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000256bool
257Archive::loadArchive(std::string* error) {
Reid Spencerf9d7a512004-11-14 21:58:33 +0000258
259 // Set up parsing
260 members.clear();
261 symTab.clear();
262 const char *At = base;
263 const char *End = base + mapfile->size();
264
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000265 if (!checkSignature(error))
266 return false;
267
Reid Spencerf9d7a512004-11-14 21:58:33 +0000268 At += 8; // Skip the magic string.
269
270 bool seenSymbolTable = false;
271 bool foundFirstFile = false;
272 while (At < End) {
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000273 // parse the member header
Reid Spencerf9d7a512004-11-14 21:58:33 +0000274 const char* Save = At;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000275 ArchiveMember* mbr = parseMemberHeader(At, End, error);
276 if (!mbr)
277 return false;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000278
279 // check if this is the foreign symbol table
Reid Spencer9a29db42004-11-20 07:29:40 +0000280 if (mbr->isSVR4SymbolTable() || mbr->isBSD4SymbolTable()) {
Reid Spencer4a980d12004-11-16 06:47:19 +0000281 // We just save this but don't do anything special
282 // with it. It doesn't count as the "first file".
Reid Spencer9a29db42004-11-20 07:29:40 +0000283 if (foreignST) {
284 // What? Multiple foreign symbol tables? Just chuck it
285 // and retain the last one found.
286 delete foreignST;
287 }
Reid Spencer4a980d12004-11-16 06:47:19 +0000288 foreignST = mbr;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000289 At += mbr->getSize();
Reid Spencer6405c9e2004-11-19 17:08:00 +0000290 if ((intptr_t(At) & 1) == 1)
Reid Spencerf9d7a512004-11-14 21:58:33 +0000291 At++;
292 } else if (mbr->isStringTable()) {
Reid Spencer4a980d12004-11-16 06:47:19 +0000293 // Simply suck the entire string table into a string
294 // variable. This will be used to get the names of the
295 // members that use the "/ddd" format for their names
296 // (SVR4 style long names).
Reid Spencer9a29db42004-11-20 07:29:40 +0000297 strtab.assign(At, mbr->getSize());
Reid Spencerf9d7a512004-11-14 21:58:33 +0000298 At += mbr->getSize();
Reid Spencer6405c9e2004-11-19 17:08:00 +0000299 if ((intptr_t(At) & 1) == 1)
Reid Spencerf9d7a512004-11-14 21:58:33 +0000300 At++;
301 delete mbr;
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000302 } else if (mbr->isLLVMSymbolTable()) {
Reid Spencer4a980d12004-11-16 06:47:19 +0000303 // This is the LLVM symbol table for the archive. If we've seen it
304 // already, its an error. Otherwise, parse the symbol table and move on.
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000305 if (seenSymbolTable) {
306 if (error)
307 *error = "invalid archive: multiple symbol tables";
308 return false;
309 }
310 if (!parseSymbolTable(mbr->getData(), mbr->getSize(), error))
311 return false;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000312 seenSymbolTable = true;
313 At += mbr->getSize();
Reid Spencer6405c9e2004-11-19 17:08:00 +0000314 if ((intptr_t(At) & 1) == 1)
Reid Spencerf9d7a512004-11-14 21:58:33 +0000315 At++;
Reid Spencer4a980d12004-11-16 06:47:19 +0000316 delete mbr; // We don't need this member in the list of members.
Reid Spencerf9d7a512004-11-14 21:58:33 +0000317 } else {
Reid Spencer4a980d12004-11-16 06:47:19 +0000318 // This is just a regular file. If its the first one, save its offset.
319 // Otherwise just push it on the list and move on to the next file.
Reid Spencerf9d7a512004-11-14 21:58:33 +0000320 if (!foundFirstFile) {
321 firstFileOffset = Save - base;
322 foundFirstFile = true;
323 }
324 members.push_back(mbr);
325 At += mbr->getSize();
Reid Spencer6405c9e2004-11-19 17:08:00 +0000326 if ((intptr_t(At) & 1) == 1)
Reid Spencerf9d7a512004-11-14 21:58:33 +0000327 At++;
328 }
329 }
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000330 return true;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000331}
332
333// Open and completely load the archive file.
334Archive*
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000335Archive::OpenAndLoad(const sys::Path& file, std::string* ErrorMessage)
336{
Reid Spencer0ff2d312006-08-24 23:45:08 +0000337 std::auto_ptr<Archive> result ( new Archive(file));
338 if (result->mapToMemory(ErrorMessage))
339 return 0;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000340 if (!result->loadArchive(ErrorMessage))
Reid Spencer5af46882004-12-13 02:59:03 +0000341 return 0;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000342 return result.release();
Reid Spencerf9d7a512004-11-14 21:58:33 +0000343}
344
345// Get all the bytecode modules from the archive
346bool
347Archive::getAllModules(std::vector<Module*>& Modules, std::string* ErrMessage) {
348
349 for (iterator I=begin(), E=end(); I != E; ++I) {
350 if (I->isBytecode() || I->isCompressedBytecode()) {
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000351 std::string FullMemberName = archPath.toString() +
Reid Spencer1fce0912004-12-11 00:14:15 +0000352 "(" + I->getPath().toString() + ")";
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000353 Module* M = ParseBytecodeBuffer((const unsigned char*)I->getData(),
Reid Spencer6ab7a4f2004-11-17 18:25:21 +0000354 I->getSize(), FullMemberName, ErrMessage);
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
484 // Now, load the bytecode 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() + ")";
Reid Spencerf9d7a512004-11-14 21:58:33 +0000487 ModuleProvider* mp = getBytecodeBufferModuleProvider(
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000488 (const unsigned char*) mbr->getData(), mbr->getSize(),
Reid Spencer0b5a5042006-08-25 17:43:11 +0000489 FullMemberName, ErrMsg, 0);
490 if (!mp)
491 return 0;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000492
Reid Spencer9a29db42004-11-20 07:29:40 +0000493 modules.insert(std::make_pair(fileOffset, std::make_pair(mp, mbr)));
Reid Spencerf9d7a512004-11-14 21:58:33 +0000494
495 return mp;
496}
497
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000498// Look up multiple symbols in the symbol table and return a set of
Reid Spencerf9d7a512004-11-14 21:58:33 +0000499// ModuleProviders that define those symbols.
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000500bool
Reid Spencer7783e8a2004-11-19 03:18:22 +0000501Archive::findModulesDefiningSymbols(std::set<std::string>& symbols,
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000502 std::set<ModuleProvider*>& result,
503 std::string* error)
Reid Spencerf9d7a512004-11-14 21:58:33 +0000504{
Reid Spencercd5561a2006-12-15 19:44:51 +0000505 if (!mapfile || !base) {
506 if (error)
507 *error = "Empty archive invalid for finding modules defining symbols";
508 return false;
509 }
510
Reid Spencer766b7932004-11-15 01:20:11 +0000511 if (symTab.empty()) {
512 // We don't have a symbol table, so we must build it now but lets also
513 // make sure that we populate the modules table as we do this to ensure
514 // that we don't load them twice when findModuleDefiningSymbol is called
515 // below.
516
517 // Get a pointer to the first file
518 const char* At = ((const char*)base) + firstFileOffset;
519 const char* End = ((const char*)base) + mapfile->size();
520
521 while ( At < End) {
522 // Compute the offset to be put in the symbol table
523 unsigned offset = At - base - firstFileOffset;
524
525 // Parse the file's header
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000526 ArchiveMember* mbr = parseMemberHeader(At, End, error);
527 if (!mbr)
528 return false;
Reid Spencer766b7932004-11-15 01:20:11 +0000529
530 // If it contains symbols
531 if (mbr->isBytecode() || mbr->isCompressedBytecode()) {
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000532 // Get the symbols
Reid Spencer766b7932004-11-15 01:20:11 +0000533 std::vector<std::string> symbols;
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000534 std::string FullMemberName = archPath.toString() + "(" +
Reid Spencer1fce0912004-12-11 00:14:15 +0000535 mbr->getPath().toString() + ")";
Reid Spencer766b7932004-11-15 01:20:11 +0000536 ModuleProvider* MP = GetBytecodeSymbols((const unsigned char*)At,
Reid Spencer0b5a5042006-08-25 17:43:11 +0000537 mbr->getSize(), FullMemberName, symbols, error);
Reid Spencer766b7932004-11-15 01:20:11 +0000538
539 if (MP) {
540 // Insert the module's symbols into the symbol table
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000541 for (std::vector<std::string>::iterator I = symbols.begin(),
Reid Spencer766b7932004-11-15 01:20:11 +0000542 E=symbols.end(); I != E; ++I ) {
Reid Spencer9a29db42004-11-20 07:29:40 +0000543 symTab.insert(std::make_pair(*I, offset));
Reid Spencer766b7932004-11-15 01:20:11 +0000544 }
545 // Insert the ModuleProvider and the ArchiveMember into the table of
546 // modules.
Reid Spencer9a29db42004-11-20 07:29:40 +0000547 modules.insert(std::make_pair(offset, std::make_pair(MP, mbr)));
Reid Spencer766b7932004-11-15 01:20:11 +0000548 } else {
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000549 if (error)
550 *error = "Can't parse bytecode member: " +
Reid Spencer0b5a5042006-08-25 17:43:11 +0000551 mbr->getPath().toString() + ": " + *error;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000552 delete mbr;
553 return false;
Reid Spencer766b7932004-11-15 01:20:11 +0000554 }
555 }
Reid Spencerb3231132004-11-15 01:40:20 +0000556
557 // Go to the next file location
558 At += mbr->getSize();
Reid Spencer6405c9e2004-11-19 17:08:00 +0000559 if ((intptr_t(At) & 1) == 1)
Reid Spencerb3231132004-11-15 01:40:20 +0000560 At++;
Reid Spencer766b7932004-11-15 01:20:11 +0000561 }
562 }
563
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000564 // At this point we have a valid symbol table (one way or another) so we
Reid Spencer766b7932004-11-15 01:20:11 +0000565 // just use it to quickly find the symbols requested.
566
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000567 for (std::set<std::string>::iterator I=symbols.begin(),
Reid Spencer7783e8a2004-11-19 03:18:22 +0000568 E=symbols.end(); I != E;) {
569 // See if this symbol exists
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000570 ModuleProvider* mp = findModuleDefiningSymbol(*I,error);
Reid Spencerf9d7a512004-11-14 21:58:33 +0000571 if (mp) {
Reid Spencer7783e8a2004-11-19 03:18:22 +0000572 // The symbol exists, insert the ModuleProvider into our result,
573 // duplicates wil be ignored
Reid Spencer766b7932004-11-15 01:20:11 +0000574 result.insert(mp);
Reid Spencer7783e8a2004-11-19 03:18:22 +0000575
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000576 // Remove the symbol now that its been resolved, being careful to
Reid Spencer57646ec2004-11-19 03:44:10 +0000577 // post-increment the iterator.
578 symbols.erase(I++);
Reid Spencer7783e8a2004-11-19 03:18:22 +0000579 } else {
580 ++I;
Chris Lattner968cfd02003-04-19 21:45:34 +0000581 }
Reid Spencerf9d7a512004-11-14 21:58:33 +0000582 }
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000583 return true;
Chris Lattner968cfd02003-04-19 21:45:34 +0000584}
Reid Spencereaa06bb2005-02-26 22:00:32 +0000585
Chris Lattner7c439922005-09-23 06:22:58 +0000586bool Archive::isBytecodeArchive() {
587 // Make sure the symTab has been loaded. In most cases this should have been
588 // done when the archive was constructed, but still, this is just in case.
589 if (!symTab.size())
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000590 if (!loadSymbolTable(0))
591 return false;
Reid Spencereaa06bb2005-02-26 22:00:32 +0000592
Chris Lattner7c439922005-09-23 06:22:58 +0000593 // Now that we know it's been loaded, return true
594 // if it has a size
595 if (symTab.size()) return true;
Reid Spencereaa06bb2005-02-26 22:00:32 +0000596
597 //We still can't be sure it isn't a bytecode archive
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000598 if (!loadArchive(0))
599 return false;
Reid Spencereaa06bb2005-02-26 22:00:32 +0000600
601 std::vector<Module *> Modules;
602 std::string ErrorMessage;
603
Chris Lattner7c439922005-09-23 06:22:58 +0000604 // Scan the archive, trying to load a bytecode member. We only load one to
605 // see if this works.
606 for (iterator I = begin(), E = end(); I != E; ++I) {
607 if (!I->isBytecode() && !I->isCompressedBytecode())
608 continue;
609
610 std::string FullMemberName =
611 archPath.toString() + "(" + I->getPath().toString() + ")";
612 Module* M = ParseBytecodeBuffer((const unsigned char*)I->getData(),
613 I->getSize(), FullMemberName);
614 if (!M)
615 return false; // Couldn't parse bytecode, not a bytecode archive.
616 delete M;
617 return true;
618 }
619
620 return false;
Reid Spencereaa06bb2005-02-26 22:00:32 +0000621}