blob: e4769cac888a4573c08a688c6278ebb58bbb136d [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 Spencerf9d7a512004-11-14 21:58:33 +000071 assert(At + sizeof(ArchiveMemberHeader) < End && "Not enough data");
72
73 // Cast archive member header
74 ArchiveMemberHeader* Hdr = (ArchiveMemberHeader*)At;
75 At += sizeof(ArchiveMemberHeader);
76
Misha Brukman2b37d7c2005-04-21 21:13:18 +000077 // Extract the size and determine if the file is
Reid Spencerf9d7a512004-11-14 21:58:33 +000078 // compressed or not (negative length).
79 int flags = 0;
80 int MemberSize = atoi(Hdr->size);
81 if (MemberSize < 0) {
82 flags |= ArchiveMember::CompressedFlag;
83 MemberSize = -MemberSize;
84 }
85
86 // Check the size of the member for sanity
Reid Spencer8d8a7ff2006-07-07 20:56:50 +000087 if (At + MemberSize > End) {
88 if (error)
89 *error = "invalid member length in archive file";
90 return 0;
91 }
Reid Spencerf9d7a512004-11-14 21:58:33 +000092
93 // Check the member signature
Reid Spencer8d8a7ff2006-07-07 20:56:50 +000094 if (!Hdr->checkSignature()) {
95 if (error)
96 *error = "invalid file member signature";
97 return 0;
98 }
Reid Spencerf9d7a512004-11-14 21:58:33 +000099
100 // Convert and check the member name
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000101 // The empty name ( '/' and 15 blanks) is for a foreign (non-LLVM) symbol
102 // table. The special name "//" and 14 blanks is for a string table, used
Reid Spencerf9d7a512004-11-14 21:58:33 +0000103 // for long file names. This library doesn't generate either of those but
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000104 // it will accept them. If the name starts with #1/ and the remainder is
105 // digits, then those digits specify the length of the name that is
106 // stored immediately following the header. The special name
107 // __LLVM_SYM_TAB__ identifies the symbol table for LLVM bytecode.
108 // Anything else is a regular, short filename that is terminated with
Reid Spencerf9d7a512004-11-14 21:58:33 +0000109 // a '/' and blanks.
110
111 std::string pathname;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000112 switch (Hdr->name[0]) {
113 case '#':
114 if (Hdr->name[1] == '1' && Hdr->name[2] == '/') {
115 if (isdigit(Hdr->name[3])) {
116 unsigned len = atoi(&Hdr->name[3]);
Reid Spencer9a29db42004-11-20 07:29:40 +0000117 pathname.assign(At, len);
Reid Spencerdd95e8d2004-11-17 16:13:11 +0000118 At += len;
119 MemberSize -= len;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000120 flags |= ArchiveMember::HasLongFilenameFlag;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000121 } else {
122 if (error)
123 *error = "invalid long filename";
124 return 0;
125 }
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000126 } else if (Hdr->name[1] == '_' &&
Reid Spencer9a29db42004-11-20 07:29:40 +0000127 (0 == memcmp(Hdr->name, ARFILE_LLVM_SYMTAB_NAME, 16))) {
Reid Spencerf9d7a512004-11-14 21:58:33 +0000128 // The member is using a long file name (>15 chars) format.
129 // This format is standard for 4.4BSD and Mac OSX operating
130 // systems. LLVM uses it similarly. In this format, the
131 // remainder of the name field (after #1/) specifies the
132 // length of the file name which occupy the first bytes of
133 // the member's data. The pathname already has the #1/ stripped.
134 pathname.assign(ARFILE_LLVM_SYMTAB_NAME);
135 flags |= ArchiveMember::LLVMSymbolTableFlag;
136 }
137 break;
138 case '/':
139 if (Hdr->name[1]== '/') {
Reid Spencer9a29db42004-11-20 07:29:40 +0000140 if (0 == memcmp(Hdr->name, ARFILE_STRTAB_NAME, 16)) {
Reid Spencerf9d7a512004-11-14 21:58:33 +0000141 pathname.assign(ARFILE_STRTAB_NAME);
142 flags |= ArchiveMember::StringTableFlag;
143 } else {
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000144 if (error)
145 *error = "invalid string table name";
146 return 0;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000147 }
148 } else if (Hdr->name[1] == ' ') {
Reid Spencer9a29db42004-11-20 07:29:40 +0000149 if (0 == memcmp(Hdr->name, ARFILE_SVR4_SYMTAB_NAME, 16)) {
150 pathname.assign(ARFILE_SVR4_SYMTAB_NAME);
151 flags |= ArchiveMember::SVR4SymbolTableFlag;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000152 } else {
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000153 if (error)
154 *error = "invalid SVR4 symbol table name";
155 return 0;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000156 }
157 } else if (isdigit(Hdr->name[1])) {
158 unsigned index = atoi(&Hdr->name[1]);
159 if (index < strtab.length()) {
160 const char* namep = strtab.c_str() + index;
161 const char* endp = strtab.c_str() + strtab.length();
162 const char* p = namep;
163 const char* last_p = p;
164 while (p < endp) {
165 if (*p == '\n' && *last_p == '/') {
Reid Spencer9a29db42004-11-20 07:29:40 +0000166 pathname.assign(namep, last_p - namep);
Reid Spencerf9d7a512004-11-14 21:58:33 +0000167 flags |= ArchiveMember::HasLongFilenameFlag;
168 break;
169 }
170 last_p = p;
171 p++;
172 }
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000173 if (p >= endp) {
174 if (error)
175 *error = "missing name termiantor in string table";
176 return 0;
177 }
Reid Spencerf9d7a512004-11-14 21:58:33 +0000178 } else {
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000179 if (error)
180 *error = "name index beyond string table";
181 return 0;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000182 }
183 }
184 break;
Reid Spencer9a29db42004-11-20 07:29:40 +0000185 case '_':
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000186 if (Hdr->name[1] == '_' &&
Reid Spencer9a29db42004-11-20 07:29:40 +0000187 (0 == memcmp(Hdr->name, ARFILE_BSD4_SYMTAB_NAME, 16))) {
188 pathname.assign(ARFILE_BSD4_SYMTAB_NAME);
189 flags |= ArchiveMember::BSD4SymbolTableFlag;
Reid Spencer84b9ced2004-11-23 22:35:39 +0000190 break;
Reid Spencer9a29db42004-11-20 07:29:40 +0000191 }
Reid Spencer84b9ced2004-11-23 22:35:39 +0000192 /* FALL THROUGH */
Reid Spencerf9d7a512004-11-14 21:58:33 +0000193
194 default:
Reid Spencer9a29db42004-11-20 07:29:40 +0000195 char* slash = (char*) memchr(Hdr->name, '/', 16);
Reid Spencerf9d7a512004-11-14 21:58:33 +0000196 if (slash == 0)
Reid Spencerdd95e8d2004-11-17 16:13:11 +0000197 slash = Hdr->name + 16;
Reid Spencer9a29db42004-11-20 07:29:40 +0000198 pathname.assign(Hdr->name, slash - Hdr->name);
Reid Spencerf9d7a512004-11-14 21:58:33 +0000199 break;
200 }
201
202 // Determine if this is a bytecode file
Reid Spencer9a29db42004-11-20 07:29:40 +0000203 switch (sys::IdentifyFileType(At, 4)) {
Reid Spencerf9d7a512004-11-14 21:58:33 +0000204 case sys::BytecodeFileType:
205 flags |= ArchiveMember::BytecodeFlag;
206 break;
207 case sys::CompressedBytecodeFileType:
208 flags |= ArchiveMember::CompressedBytecodeFlag;
209 flags &= ~ArchiveMember::CompressedFlag;
210 break;
211 default:
212 flags &= ~(ArchiveMember::BytecodeFlag|
213 ArchiveMember::CompressedBytecodeFlag);
214 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
341// Get all the bytecode modules from the archive
342bool
343Archive::getAllModules(std::vector<Module*>& Modules, std::string* ErrMessage) {
344
345 for (iterator I=begin(), E=end(); I != E; ++I) {
346 if (I->isBytecode() || I->isCompressedBytecode()) {
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000347 std::string FullMemberName = archPath.toString() +
Reid Spencer1fce0912004-12-11 00:14:15 +0000348 "(" + I->getPath().toString() + ")";
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000349 Module* M = ParseBytecodeBuffer((const unsigned char*)I->getData(),
Reid Spencer6ab7a4f2004-11-17 18:25:21 +0000350 I->getSize(), FullMemberName, ErrMessage);
Reid Spencerf9d7a512004-11-14 21:58:33 +0000351 if (!M)
352 return true;
353
354 Modules.push_back(M);
355 }
356 }
Brian Gaeke2c61d7b2003-11-16 23:08:48 +0000357 return false;
358}
Chris Lattner968cfd02003-04-19 21:45:34 +0000359
Reid Spencerf9d7a512004-11-14 21:58:33 +0000360// Load just the symbol table from the archive file
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000361bool
362Archive::loadSymbolTable(std::string* ErrorMsg) {
Chris Lattner968cfd02003-04-19 21:45:34 +0000363
Reid Spencerf9d7a512004-11-14 21:58:33 +0000364 // Set up parsing
365 members.clear();
366 symTab.clear();
367 const char *At = base;
368 const char *End = base + mapfile->size();
Chris Lattner968cfd02003-04-19 21:45:34 +0000369
Reid Spencerf9d7a512004-11-14 21:58:33 +0000370 // Make sure we're dealing with an archive
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000371 if (!checkSignature(ErrorMsg))
372 return false;
Chris Lattner968cfd02003-04-19 21:45:34 +0000373
Reid Spencerf9d7a512004-11-14 21:58:33 +0000374 At += 8; // Skip signature
375
376 // Parse the first file member header
377 const char* FirstFile = At;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000378 ArchiveMember* mbr = parseMemberHeader(At, End, ErrorMsg);
379 if (!mbr)
380 return false;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000381
Reid Spencer9a29db42004-11-20 07:29:40 +0000382 if (mbr->isSVR4SymbolTable() || mbr->isBSD4SymbolTable()) {
Reid Spencerf9d7a512004-11-14 21:58:33 +0000383 // Skip the foreign symbol table, we don't do anything with it
384 At += mbr->getSize();
Reid Spencer6405c9e2004-11-19 17:08:00 +0000385 if ((intptr_t(At) & 1) == 1)
Reid Spencerb3231132004-11-15 01:40:20 +0000386 At++;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000387 delete mbr;
388
Reid Spencerb3231132004-11-15 01:40:20 +0000389 // Read the next one
Reid Spencerf9d7a512004-11-14 21:58:33 +0000390 FirstFile = At;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000391 mbr = parseMemberHeader(At, End, ErrorMsg);
392 if (!mbr) {
393 delete mbr;
394 return false;
395 }
Reid Spencerb3231132004-11-15 01:40:20 +0000396 }
397
398 if (mbr->isStringTable()) {
399 // Process the string table entry
Reid Spencer9a29db42004-11-20 07:29:40 +0000400 strtab.assign((const char*)mbr->getData(), mbr->getSize());
Reid Spencerb3231132004-11-15 01:40:20 +0000401 At += mbr->getSize();
Reid Spencer6405c9e2004-11-19 17:08:00 +0000402 if ((intptr_t(At) & 1) == 1)
Reid Spencerb3231132004-11-15 01:40:20 +0000403 At++;
404 delete mbr;
405 // Get the next one
406 FirstFile = At;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000407 mbr = parseMemberHeader(At, End, ErrorMsg);
408 if (!mbr) {
409 delete mbr;
410 return false;
411 }
Chris Lattner968cfd02003-04-19 21:45:34 +0000412 }
413
Reid Spencerf9d7a512004-11-14 21:58:33 +0000414 // See if its the symbol table
415 if (mbr->isLLVMSymbolTable()) {
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000416 if (!parseSymbolTable(mbr->getData(), mbr->getSize(), ErrorMsg)) {
417 delete mbr;
418 return false;
419 }
420
Reid Spencer8dde18f2004-11-28 03:13:02 +0000421 At += mbr->getSize();
Reid Spencer6405c9e2004-11-19 17:08:00 +0000422 if ((intptr_t(At) & 1) == 1)
Reid Spencer8dde18f2004-11-28 03:13:02 +0000423 At++;
Reid Spencerffb4d622006-05-12 17:56:20 +0000424 delete mbr;
425 // Can't be any more symtab headers so just advance
Reid Spencer8dde18f2004-11-28 03:13:02 +0000426 FirstFile = At;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000427 } else {
428 // There's no symbol table in the file. We have to rebuild it from scratch
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000429 // because the intent of this method is to get the symbol table loaded so
430 // it can be searched efficiently.
Reid Spencerf9d7a512004-11-14 21:58:33 +0000431 // Add the member to the members list
432 members.push_back(mbr);
433 }
434
435 firstFileOffset = FirstFile - base;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000436 return true;
Chris Lattner968cfd02003-04-19 21:45:34 +0000437}
438
Reid Spencerf9d7a512004-11-14 21:58:33 +0000439// Open the archive and load just the symbol tables
440Archive*
Reid Spencer5af46882004-12-13 02:59:03 +0000441Archive::OpenAndLoadSymbols(const sys::Path& file, std::string* ErrorMessage) {
Reid Spencer0ff2d312006-08-24 23:45:08 +0000442 std::auto_ptr<Archive> result ( new Archive(file) );
443 if (result->mapToMemory(ErrorMessage))
444 return 0;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000445 if (!result->loadSymbolTable(ErrorMessage))
Reid Spencer5af46882004-12-13 02:59:03 +0000446 return 0;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000447 return result.release();
Reid Spencerf9d7a512004-11-14 21:58:33 +0000448}
Chris Lattner968cfd02003-04-19 21:45:34 +0000449
Reid Spencerf9d7a512004-11-14 21:58:33 +0000450// Look up one symbol in the symbol table and return a ModuleProvider for the
451// module that defines that symbol.
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000452ModuleProvider*
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000453Archive::findModuleDefiningSymbol(const std::string& symbol,
454 std::string* ErrMsg) {
Reid Spencerf9d7a512004-11-14 21:58:33 +0000455 SymTabType::iterator SI = symTab.find(symbol);
456 if (SI == symTab.end())
457 return 0;
458
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000459 // The symbol table was previously constructed assuming that the members were
Reid Spencerf9d7a512004-11-14 21:58:33 +0000460 // written without the symbol table header. Because VBR encoding is used, the
461 // values could not be adjusted to account for the offset of the symbol table
462 // because that could affect the size of the symbol table due to VBR encoding.
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000463 // We now have to account for this by adjusting the offset by the size of the
Reid Spencerf9d7a512004-11-14 21:58:33 +0000464 // symbol table and its header.
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000465 unsigned fileOffset =
Reid Spencerf9d7a512004-11-14 21:58:33 +0000466 SI->second + // offset in symbol-table-less file
467 firstFileOffset; // add offset to first "real" file in archive
468
469 // See if the module is already loaded
470 ModuleMap::iterator MI = modules.find(fileOffset);
471 if (MI != modules.end())
472 return MI->second.first;
473
474 // Module hasn't been loaded yet, we need to load it
475 const char* modptr = base + fileOffset;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000476 ArchiveMember* mbr = parseMemberHeader(modptr, base + mapfile->size(),ErrMsg);
477 if (!mbr)
Reid Spencer0b5a5042006-08-25 17:43:11 +0000478 return 0;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000479
480 // Now, load the bytecode module to get the ModuleProvider
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000481 std::string FullMemberName = archPath.toString() + "(" +
Reid Spencer1fce0912004-12-11 00:14:15 +0000482 mbr->getPath().toString() + ")";
Reid Spencerf9d7a512004-11-14 21:58:33 +0000483 ModuleProvider* mp = getBytecodeBufferModuleProvider(
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000484 (const unsigned char*) mbr->getData(), mbr->getSize(),
Reid Spencer0b5a5042006-08-25 17:43:11 +0000485 FullMemberName, ErrMsg, 0);
486 if (!mp)
487 return 0;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000488
Reid Spencer9a29db42004-11-20 07:29:40 +0000489 modules.insert(std::make_pair(fileOffset, std::make_pair(mp, mbr)));
Reid Spencerf9d7a512004-11-14 21:58:33 +0000490
491 return mp;
492}
493
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000494// Look up multiple symbols in the symbol table and return a set of
Reid Spencerf9d7a512004-11-14 21:58:33 +0000495// ModuleProviders that define those symbols.
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000496bool
Reid Spencer7783e8a2004-11-19 03:18:22 +0000497Archive::findModulesDefiningSymbols(std::set<std::string>& symbols,
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000498 std::set<ModuleProvider*>& result,
499 std::string* error)
Reid Spencerf9d7a512004-11-14 21:58:33 +0000500{
Reid Spencer766b7932004-11-15 01:20:11 +0000501 assert(mapfile && base && "Can't findModulesDefiningSymbols on new archive");
502 if (symTab.empty()) {
503 // We don't have a symbol table, so we must build it now but lets also
504 // make sure that we populate the modules table as we do this to ensure
505 // that we don't load them twice when findModuleDefiningSymbol is called
506 // below.
507
508 // Get a pointer to the first file
509 const char* At = ((const char*)base) + firstFileOffset;
510 const char* End = ((const char*)base) + mapfile->size();
511
512 while ( At < End) {
513 // Compute the offset to be put in the symbol table
514 unsigned offset = At - base - firstFileOffset;
515
516 // Parse the file's header
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000517 ArchiveMember* mbr = parseMemberHeader(At, End, error);
518 if (!mbr)
519 return false;
Reid Spencer766b7932004-11-15 01:20:11 +0000520
521 // If it contains symbols
522 if (mbr->isBytecode() || mbr->isCompressedBytecode()) {
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000523 // Get the symbols
Reid Spencer766b7932004-11-15 01:20:11 +0000524 std::vector<std::string> symbols;
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000525 std::string FullMemberName = archPath.toString() + "(" +
Reid Spencer1fce0912004-12-11 00:14:15 +0000526 mbr->getPath().toString() + ")";
Reid Spencer766b7932004-11-15 01:20:11 +0000527 ModuleProvider* MP = GetBytecodeSymbols((const unsigned char*)At,
Reid Spencer0b5a5042006-08-25 17:43:11 +0000528 mbr->getSize(), FullMemberName, symbols, error);
Reid Spencer766b7932004-11-15 01:20:11 +0000529
530 if (MP) {
531 // Insert the module's symbols into the symbol table
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000532 for (std::vector<std::string>::iterator I = symbols.begin(),
Reid Spencer766b7932004-11-15 01:20:11 +0000533 E=symbols.end(); I != E; ++I ) {
Reid Spencer9a29db42004-11-20 07:29:40 +0000534 symTab.insert(std::make_pair(*I, offset));
Reid Spencer766b7932004-11-15 01:20:11 +0000535 }
536 // Insert the ModuleProvider and the ArchiveMember into the table of
537 // modules.
Reid Spencer9a29db42004-11-20 07:29:40 +0000538 modules.insert(std::make_pair(offset, std::make_pair(MP, mbr)));
Reid Spencer766b7932004-11-15 01:20:11 +0000539 } else {
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000540 if (error)
541 *error = "Can't parse bytecode member: " +
Reid Spencer0b5a5042006-08-25 17:43:11 +0000542 mbr->getPath().toString() + ": " + *error;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000543 delete mbr;
544 return false;
Reid Spencer766b7932004-11-15 01:20:11 +0000545 }
546 }
Reid Spencerb3231132004-11-15 01:40:20 +0000547
548 // Go to the next file location
549 At += mbr->getSize();
Reid Spencer6405c9e2004-11-19 17:08:00 +0000550 if ((intptr_t(At) & 1) == 1)
Reid Spencerb3231132004-11-15 01:40:20 +0000551 At++;
Reid Spencer766b7932004-11-15 01:20:11 +0000552 }
553 }
554
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000555 // At this point we have a valid symbol table (one way or another) so we
Reid Spencer766b7932004-11-15 01:20:11 +0000556 // just use it to quickly find the symbols requested.
557
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000558 for (std::set<std::string>::iterator I=symbols.begin(),
Reid Spencer7783e8a2004-11-19 03:18:22 +0000559 E=symbols.end(); I != E;) {
560 // See if this symbol exists
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000561 ModuleProvider* mp = findModuleDefiningSymbol(*I,error);
Reid Spencerf9d7a512004-11-14 21:58:33 +0000562 if (mp) {
Reid Spencer7783e8a2004-11-19 03:18:22 +0000563 // The symbol exists, insert the ModuleProvider into our result,
564 // duplicates wil be ignored
Reid Spencer766b7932004-11-15 01:20:11 +0000565 result.insert(mp);
Reid Spencer7783e8a2004-11-19 03:18:22 +0000566
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000567 // Remove the symbol now that its been resolved, being careful to
Reid Spencer57646ec2004-11-19 03:44:10 +0000568 // post-increment the iterator.
569 symbols.erase(I++);
Reid Spencer7783e8a2004-11-19 03:18:22 +0000570 } else {
571 ++I;
Chris Lattner968cfd02003-04-19 21:45:34 +0000572 }
Reid Spencerf9d7a512004-11-14 21:58:33 +0000573 }
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000574 return true;
Chris Lattner968cfd02003-04-19 21:45:34 +0000575}
Reid Spencereaa06bb2005-02-26 22:00:32 +0000576
Chris Lattner7c439922005-09-23 06:22:58 +0000577bool Archive::isBytecodeArchive() {
578 // Make sure the symTab has been loaded. In most cases this should have been
579 // done when the archive was constructed, but still, this is just in case.
580 if (!symTab.size())
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000581 if (!loadSymbolTable(0))
582 return false;
Reid Spencereaa06bb2005-02-26 22:00:32 +0000583
Chris Lattner7c439922005-09-23 06:22:58 +0000584 // Now that we know it's been loaded, return true
585 // if it has a size
586 if (symTab.size()) return true;
Reid Spencereaa06bb2005-02-26 22:00:32 +0000587
588 //We still can't be sure it isn't a bytecode archive
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000589 if (!loadArchive(0))
590 return false;
Reid Spencereaa06bb2005-02-26 22:00:32 +0000591
592 std::vector<Module *> Modules;
593 std::string ErrorMessage;
594
Chris Lattner7c439922005-09-23 06:22:58 +0000595 // Scan the archive, trying to load a bytecode member. We only load one to
596 // see if this works.
597 for (iterator I = begin(), E = end(); I != E; ++I) {
598 if (!I->isBytecode() && !I->isCompressedBytecode())
599 continue;
600
601 std::string FullMemberName =
602 archPath.toString() + "(" + I->getPath().toString() + ")";
603 Module* M = ParseBytecodeBuffer((const unsigned char*)I->getData(),
604 I->getSize(), FullMemberName);
605 if (!M)
606 return false; // Couldn't parse bytecode, not a bytecode archive.
607 delete M;
608 return true;
609 }
610
611 return false;
Reid Spencereaa06bb2005-02-26 22:00:32 +0000612}