blob: c38389e2d7c08ef8753546051aaf8e8af72653df [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 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
112 // __LLVM_SYM_TAB__ identifies the symbol table for LLVM bytecode.
113 // 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
207 // Determine if this is a bytecode 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:
Reid Spencer20c34892007-04-04 06:31:04 +0000210 case sys::Bytecode_FileType:
Reid Spencerf9d7a512004-11-14 21:58:33 +0000211 flags |= ArchiveMember::BytecodeFlag;
212 break;
Reid Spencer20c34892007-04-04 06:31:04 +0000213 case sys::CompressedBytecode_FileType:
Reid Spencerf9d7a512004-11-14 21:58:33 +0000214 flags |= ArchiveMember::CompressedBytecodeFlag;
215 flags &= ~ArchiveMember::CompressedFlag;
216 break;
217 default:
218 flags &= ~(ArchiveMember::BytecodeFlag|
219 ArchiveMember::CompressedBytecodeFlag);
220 break;
221 }
222
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000223 // Instantiate the ArchiveMember to be filled
224 ArchiveMember* member = new ArchiveMember(this);
225
Reid Spencerf9d7a512004-11-14 21:58:33 +0000226 // Fill in fields of the ArchiveMember
227 member->next = 0;
228 member->prev = 0;
229 member->parent = this;
Reid Spencerdd04df02005-07-07 23:21:43 +0000230 member->path.set(pathname);
Reid Spencerf9d7a512004-11-14 21:58:33 +0000231 member->info.fileSize = MemberSize;
232 member->info.modTime.fromEpochTime(atoi(Hdr->date));
Reid Spencer56127622004-12-29 01:20:24 +0000233 unsigned int mode;
234 sscanf(Hdr->mode, "%o", &mode);
235 member->info.mode = mode;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000236 member->info.user = atoi(Hdr->uid);
237 member->info.group = atoi(Hdr->gid);
238 member->flags = flags;
239 member->data = At;
240
241 return member;
242}
243
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000244bool
245Archive::checkSignature(std::string* error) {
Reid Spencerf9d7a512004-11-14 21:58:33 +0000246 // Check the magic string at file's header
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000247 if (mapfile->size() < 8 || memcmp(base, ARFILE_MAGIC, 8)) {
248 if (error)
249 *error = "invalid signature for an archive file";
250 return false;
251 }
252 return true;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000253}
254
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000255// This function loads the entire archive and fully populates its ilist with
Reid Spencerf9d7a512004-11-14 21:58:33 +0000256// the members of the archive file. This is typically used in preparation for
257// editing the contents of the archive.
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000258bool
259Archive::loadArchive(std::string* error) {
Reid Spencerf9d7a512004-11-14 21:58:33 +0000260
261 // Set up parsing
262 members.clear();
263 symTab.clear();
264 const char *At = base;
265 const char *End = base + mapfile->size();
266
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000267 if (!checkSignature(error))
268 return false;
269
Reid Spencerf9d7a512004-11-14 21:58:33 +0000270 At += 8; // Skip the magic string.
271
272 bool seenSymbolTable = false;
273 bool foundFirstFile = false;
274 while (At < End) {
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000275 // parse the member header
Reid Spencerf9d7a512004-11-14 21:58:33 +0000276 const char* Save = At;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000277 ArchiveMember* mbr = parseMemberHeader(At, End, error);
278 if (!mbr)
279 return false;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000280
281 // check if this is the foreign symbol table
Reid Spencer9a29db42004-11-20 07:29:40 +0000282 if (mbr->isSVR4SymbolTable() || mbr->isBSD4SymbolTable()) {
Reid Spencer4a980d12004-11-16 06:47:19 +0000283 // We just save this but don't do anything special
284 // with it. It doesn't count as the "first file".
Reid Spencer9a29db42004-11-20 07:29:40 +0000285 if (foreignST) {
286 // What? Multiple foreign symbol tables? Just chuck it
287 // and retain the last one found.
288 delete foreignST;
289 }
Reid Spencer4a980d12004-11-16 06:47:19 +0000290 foreignST = mbr;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000291 At += mbr->getSize();
Reid Spencer6405c9e2004-11-19 17:08:00 +0000292 if ((intptr_t(At) & 1) == 1)
Reid Spencerf9d7a512004-11-14 21:58:33 +0000293 At++;
294 } else if (mbr->isStringTable()) {
Reid Spencer4a980d12004-11-16 06:47:19 +0000295 // Simply suck the entire string table into a string
296 // variable. This will be used to get the names of the
297 // members that use the "/ddd" format for their names
298 // (SVR4 style long names).
Reid Spencer9a29db42004-11-20 07:29:40 +0000299 strtab.assign(At, mbr->getSize());
Reid Spencerf9d7a512004-11-14 21:58:33 +0000300 At += mbr->getSize();
Reid Spencer6405c9e2004-11-19 17:08:00 +0000301 if ((intptr_t(At) & 1) == 1)
Reid Spencerf9d7a512004-11-14 21:58:33 +0000302 At++;
303 delete mbr;
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000304 } else if (mbr->isLLVMSymbolTable()) {
Reid Spencer4a980d12004-11-16 06:47:19 +0000305 // This is the LLVM symbol table for the archive. If we've seen it
306 // already, its an error. Otherwise, parse the symbol table and move on.
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000307 if (seenSymbolTable) {
308 if (error)
309 *error = "invalid archive: multiple symbol tables";
310 return false;
311 }
312 if (!parseSymbolTable(mbr->getData(), mbr->getSize(), error))
313 return false;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000314 seenSymbolTable = true;
315 At += mbr->getSize();
Reid Spencer6405c9e2004-11-19 17:08:00 +0000316 if ((intptr_t(At) & 1) == 1)
Reid Spencerf9d7a512004-11-14 21:58:33 +0000317 At++;
Reid Spencer4a980d12004-11-16 06:47:19 +0000318 delete mbr; // We don't need this member in the list of members.
Reid Spencerf9d7a512004-11-14 21:58:33 +0000319 } else {
Reid Spencer4a980d12004-11-16 06:47:19 +0000320 // This is just a regular file. If its the first one, save its offset.
321 // Otherwise just push it on the list and move on to the next file.
Reid Spencerf9d7a512004-11-14 21:58:33 +0000322 if (!foundFirstFile) {
323 firstFileOffset = Save - base;
324 foundFirstFile = true;
325 }
326 members.push_back(mbr);
327 At += mbr->getSize();
Reid Spencer6405c9e2004-11-19 17:08:00 +0000328 if ((intptr_t(At) & 1) == 1)
Reid Spencerf9d7a512004-11-14 21:58:33 +0000329 At++;
330 }
331 }
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000332 return true;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000333}
334
335// Open and completely load the archive file.
336Archive*
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000337Archive::OpenAndLoad(const sys::Path& file, std::string* ErrorMessage)
338{
Reid Spencer0ff2d312006-08-24 23:45:08 +0000339 std::auto_ptr<Archive> result ( new Archive(file));
340 if (result->mapToMemory(ErrorMessage))
341 return 0;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000342 if (!result->loadArchive(ErrorMessage))
Reid Spencer5af46882004-12-13 02:59:03 +0000343 return 0;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000344 return result.release();
Reid Spencerf9d7a512004-11-14 21:58:33 +0000345}
346
347// Get all the bytecode modules from the archive
348bool
349Archive::getAllModules(std::vector<Module*>& Modules, std::string* ErrMessage) {
350
351 for (iterator I=begin(), E=end(); I != E; ++I) {
352 if (I->isBytecode() || I->isCompressedBytecode()) {
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000353 std::string FullMemberName = archPath.toString() +
Reid Spencer1fce0912004-12-11 00:14:15 +0000354 "(" + I->getPath().toString() + ")";
Chris Lattnerc1d56242007-05-06 09:28:33 +0000355 MemoryBuffer *Buffer =
356 MemoryBuffer::getNewMemBuffer(I->getSize(), FullMemberName.c_str());
357 memcpy((char*)Buffer->getBufferStart(), I->getData(), I->getSize());
Chris Lattnere07c15c2007-05-06 06:18:07 +0000358
Chris Lattnerc1d56242007-05-06 09:28:33 +0000359 Module *M = ParseBitcodeFile(Buffer, ErrMessage);
360 delete Buffer;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000361 if (!M)
362 return true;
363
364 Modules.push_back(M);
365 }
366 }
Brian Gaeke2c61d7b2003-11-16 23:08:48 +0000367 return false;
368}
Chris Lattner968cfd02003-04-19 21:45:34 +0000369
Reid Spencerf9d7a512004-11-14 21:58:33 +0000370// Load just the symbol table from the archive file
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000371bool
372Archive::loadSymbolTable(std::string* ErrorMsg) {
Chris Lattner968cfd02003-04-19 21:45:34 +0000373
Reid Spencerf9d7a512004-11-14 21:58:33 +0000374 // Set up parsing
375 members.clear();
376 symTab.clear();
377 const char *At = base;
378 const char *End = base + mapfile->size();
Chris Lattner968cfd02003-04-19 21:45:34 +0000379
Reid Spencerf9d7a512004-11-14 21:58:33 +0000380 // Make sure we're dealing with an archive
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000381 if (!checkSignature(ErrorMsg))
382 return false;
Chris Lattner968cfd02003-04-19 21:45:34 +0000383
Reid Spencerf9d7a512004-11-14 21:58:33 +0000384 At += 8; // Skip signature
385
386 // Parse the first file member header
387 const char* FirstFile = At;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000388 ArchiveMember* mbr = parseMemberHeader(At, End, ErrorMsg);
389 if (!mbr)
390 return false;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000391
Reid Spencer9a29db42004-11-20 07:29:40 +0000392 if (mbr->isSVR4SymbolTable() || mbr->isBSD4SymbolTable()) {
Reid Spencerf9d7a512004-11-14 21:58:33 +0000393 // Skip the foreign symbol table, we don't do anything with it
394 At += mbr->getSize();
Reid Spencer6405c9e2004-11-19 17:08:00 +0000395 if ((intptr_t(At) & 1) == 1)
Reid Spencerb3231132004-11-15 01:40:20 +0000396 At++;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000397 delete mbr;
398
Reid Spencerb3231132004-11-15 01:40:20 +0000399 // Read the next one
Reid Spencerf9d7a512004-11-14 21:58:33 +0000400 FirstFile = At;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000401 mbr = parseMemberHeader(At, End, ErrorMsg);
402 if (!mbr) {
403 delete mbr;
404 return false;
405 }
Reid Spencerb3231132004-11-15 01:40:20 +0000406 }
407
408 if (mbr->isStringTable()) {
409 // Process the string table entry
Reid Spencer9a29db42004-11-20 07:29:40 +0000410 strtab.assign((const char*)mbr->getData(), mbr->getSize());
Reid Spencerb3231132004-11-15 01:40:20 +0000411 At += mbr->getSize();
Reid Spencer6405c9e2004-11-19 17:08:00 +0000412 if ((intptr_t(At) & 1) == 1)
Reid Spencerb3231132004-11-15 01:40:20 +0000413 At++;
414 delete mbr;
415 // Get the next one
416 FirstFile = At;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000417 mbr = parseMemberHeader(At, End, ErrorMsg);
418 if (!mbr) {
419 delete mbr;
420 return false;
421 }
Chris Lattner968cfd02003-04-19 21:45:34 +0000422 }
423
Reid Spencerf9d7a512004-11-14 21:58:33 +0000424 // See if its the symbol table
425 if (mbr->isLLVMSymbolTable()) {
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000426 if (!parseSymbolTable(mbr->getData(), mbr->getSize(), ErrorMsg)) {
427 delete mbr;
428 return false;
429 }
430
Reid Spencer8dde18f2004-11-28 03:13:02 +0000431 At += mbr->getSize();
Reid Spencer6405c9e2004-11-19 17:08:00 +0000432 if ((intptr_t(At) & 1) == 1)
Reid Spencer8dde18f2004-11-28 03:13:02 +0000433 At++;
Reid Spencerffb4d622006-05-12 17:56:20 +0000434 delete mbr;
435 // Can't be any more symtab headers so just advance
Reid Spencer8dde18f2004-11-28 03:13:02 +0000436 FirstFile = At;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000437 } else {
438 // There's no symbol table in the file. We have to rebuild it from scratch
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000439 // because the intent of this method is to get the symbol table loaded so
440 // it can be searched efficiently.
Reid Spencerf9d7a512004-11-14 21:58:33 +0000441 // Add the member to the members list
442 members.push_back(mbr);
443 }
444
445 firstFileOffset = FirstFile - base;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000446 return true;
Chris Lattner968cfd02003-04-19 21:45:34 +0000447}
448
Reid Spencerf9d7a512004-11-14 21:58:33 +0000449// Open the archive and load just the symbol tables
450Archive*
Reid Spencer5af46882004-12-13 02:59:03 +0000451Archive::OpenAndLoadSymbols(const sys::Path& file, std::string* ErrorMessage) {
Reid Spencer0ff2d312006-08-24 23:45:08 +0000452 std::auto_ptr<Archive> result ( new Archive(file) );
453 if (result->mapToMemory(ErrorMessage))
454 return 0;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000455 if (!result->loadSymbolTable(ErrorMessage))
Reid Spencer5af46882004-12-13 02:59:03 +0000456 return 0;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000457 return result.release();
Reid Spencerf9d7a512004-11-14 21:58:33 +0000458}
Chris Lattner968cfd02003-04-19 21:45:34 +0000459
Reid Spencerf9d7a512004-11-14 21:58:33 +0000460// Look up one symbol in the symbol table and return a ModuleProvider for the
461// module that defines that symbol.
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000462ModuleProvider*
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000463Archive::findModuleDefiningSymbol(const std::string& symbol,
464 std::string* ErrMsg) {
Reid Spencerf9d7a512004-11-14 21:58:33 +0000465 SymTabType::iterator SI = symTab.find(symbol);
466 if (SI == symTab.end())
467 return 0;
468
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000469 // The symbol table was previously constructed assuming that the members were
Reid Spencerf9d7a512004-11-14 21:58:33 +0000470 // written without the symbol table header. Because VBR encoding is used, the
471 // values could not be adjusted to account for the offset of the symbol table
472 // because that could affect the size of the symbol table due to VBR encoding.
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000473 // We now have to account for this by adjusting the offset by the size of the
Reid Spencerf9d7a512004-11-14 21:58:33 +0000474 // symbol table and its header.
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000475 unsigned fileOffset =
Reid Spencerf9d7a512004-11-14 21:58:33 +0000476 SI->second + // offset in symbol-table-less file
477 firstFileOffset; // add offset to first "real" file in archive
478
479 // See if the module is already loaded
480 ModuleMap::iterator MI = modules.find(fileOffset);
481 if (MI != modules.end())
482 return MI->second.first;
483
484 // Module hasn't been loaded yet, we need to load it
485 const char* modptr = base + fileOffset;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000486 ArchiveMember* mbr = parseMemberHeader(modptr, base + mapfile->size(),ErrMsg);
487 if (!mbr)
Reid Spencer0b5a5042006-08-25 17:43:11 +0000488 return 0;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000489
490 // Now, load the bytecode module to get the ModuleProvider
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000491 std::string FullMemberName = archPath.toString() + "(" +
Reid Spencer1fce0912004-12-11 00:14:15 +0000492 mbr->getPath().toString() + ")";
Chris Lattnerc1d56242007-05-06 09:28:33 +0000493 MemoryBuffer *Buffer =MemoryBuffer::getNewMemBuffer(mbr->getSize(),
494 FullMemberName.c_str());
495 memcpy((char*)Buffer->getBufferStart(), mbr->getData(), mbr->getSize());
496
497 ModuleProvider *mp = getBitcodeModuleProvider(Buffer, ErrMsg);
Reid Spencer0b5a5042006-08-25 17:43:11 +0000498 if (!mp)
499 return 0;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000500
Reid Spencer9a29db42004-11-20 07:29:40 +0000501 modules.insert(std::make_pair(fileOffset, std::make_pair(mp, mbr)));
Reid Spencerf9d7a512004-11-14 21:58:33 +0000502
503 return mp;
504}
505
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000506// Look up multiple symbols in the symbol table and return a set of
Reid Spencerf9d7a512004-11-14 21:58:33 +0000507// ModuleProviders that define those symbols.
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000508bool
Reid Spencer7783e8a2004-11-19 03:18:22 +0000509Archive::findModulesDefiningSymbols(std::set<std::string>& symbols,
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000510 std::set<ModuleProvider*>& result,
Chris Lattnerf2e292c2007-02-07 21:41:02 +0000511 std::string* error) {
Reid Spencercd5561a2006-12-15 19:44:51 +0000512 if (!mapfile || !base) {
513 if (error)
514 *error = "Empty archive invalid for finding modules defining symbols";
515 return false;
516 }
517
Reid Spencer766b7932004-11-15 01:20:11 +0000518 if (symTab.empty()) {
519 // We don't have a symbol table, so we must build it now but lets also
520 // make sure that we populate the modules table as we do this to ensure
521 // that we don't load them twice when findModuleDefiningSymbol is called
522 // below.
523
524 // Get a pointer to the first file
525 const char* At = ((const char*)base) + firstFileOffset;
526 const char* End = ((const char*)base) + mapfile->size();
527
528 while ( At < End) {
529 // Compute the offset to be put in the symbol table
530 unsigned offset = At - base - firstFileOffset;
531
532 // Parse the file's header
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000533 ArchiveMember* mbr = parseMemberHeader(At, End, error);
534 if (!mbr)
535 return false;
Reid Spencer766b7932004-11-15 01:20:11 +0000536
537 // If it contains symbols
538 if (mbr->isBytecode() || mbr->isCompressedBytecode()) {
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000539 // Get the symbols
Reid Spencer766b7932004-11-15 01:20:11 +0000540 std::vector<std::string> symbols;
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000541 std::string FullMemberName = archPath.toString() + "(" +
Reid Spencer1fce0912004-12-11 00:14:15 +0000542 mbr->getPath().toString() + ")";
Chris Lattnerf2e292c2007-02-07 21:41:02 +0000543 ModuleProvider* MP =
544 GetBytecodeSymbols((const unsigned char*)At, mbr->getSize(),
Chris Lattnerc1d56242007-05-06 09:28:33 +0000545 FullMemberName, symbols, error);
Reid Spencer766b7932004-11-15 01:20:11 +0000546
547 if (MP) {
548 // Insert the module's symbols into the symbol table
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000549 for (std::vector<std::string>::iterator I = symbols.begin(),
Reid Spencer766b7932004-11-15 01:20:11 +0000550 E=symbols.end(); I != E; ++I ) {
Reid Spencer9a29db42004-11-20 07:29:40 +0000551 symTab.insert(std::make_pair(*I, offset));
Reid Spencer766b7932004-11-15 01:20:11 +0000552 }
553 // Insert the ModuleProvider and the ArchiveMember into the table of
554 // modules.
Reid Spencer9a29db42004-11-20 07:29:40 +0000555 modules.insert(std::make_pair(offset, std::make_pair(MP, mbr)));
Reid Spencer766b7932004-11-15 01:20:11 +0000556 } else {
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000557 if (error)
558 *error = "Can't parse bytecode member: " +
Reid Spencer0b5a5042006-08-25 17:43:11 +0000559 mbr->getPath().toString() + ": " + *error;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000560 delete mbr;
561 return false;
Reid Spencer766b7932004-11-15 01:20:11 +0000562 }
563 }
Reid Spencerb3231132004-11-15 01:40:20 +0000564
565 // Go to the next file location
566 At += mbr->getSize();
Reid Spencer6405c9e2004-11-19 17:08:00 +0000567 if ((intptr_t(At) & 1) == 1)
Reid Spencerb3231132004-11-15 01:40:20 +0000568 At++;
Reid Spencer766b7932004-11-15 01:20:11 +0000569 }
570 }
571
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000572 // At this point we have a valid symbol table (one way or another) so we
Reid Spencer766b7932004-11-15 01:20:11 +0000573 // just use it to quickly find the symbols requested.
574
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000575 for (std::set<std::string>::iterator I=symbols.begin(),
Reid Spencer7783e8a2004-11-19 03:18:22 +0000576 E=symbols.end(); I != E;) {
577 // See if this symbol exists
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000578 ModuleProvider* mp = findModuleDefiningSymbol(*I,error);
Reid Spencerf9d7a512004-11-14 21:58:33 +0000579 if (mp) {
Reid Spencer7783e8a2004-11-19 03:18:22 +0000580 // The symbol exists, insert the ModuleProvider into our result,
581 // duplicates wil be ignored
Reid Spencer766b7932004-11-15 01:20:11 +0000582 result.insert(mp);
Reid Spencer7783e8a2004-11-19 03:18:22 +0000583
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000584 // Remove the symbol now that its been resolved, being careful to
Reid Spencer57646ec2004-11-19 03:44:10 +0000585 // post-increment the iterator.
586 symbols.erase(I++);
Reid Spencer7783e8a2004-11-19 03:18:22 +0000587 } else {
588 ++I;
Chris Lattner968cfd02003-04-19 21:45:34 +0000589 }
Reid Spencerf9d7a512004-11-14 21:58:33 +0000590 }
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000591 return true;
Chris Lattner968cfd02003-04-19 21:45:34 +0000592}
Reid Spencereaa06bb2005-02-26 22:00:32 +0000593
Chris Lattner7c439922005-09-23 06:22:58 +0000594bool Archive::isBytecodeArchive() {
595 // Make sure the symTab has been loaded. In most cases this should have been
596 // done when the archive was constructed, but still, this is just in case.
597 if (!symTab.size())
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000598 if (!loadSymbolTable(0))
599 return false;
Reid Spencereaa06bb2005-02-26 22:00:32 +0000600
Chris Lattner7c439922005-09-23 06:22:58 +0000601 // Now that we know it's been loaded, return true
602 // if it has a size
603 if (symTab.size()) return true;
Reid Spencereaa06bb2005-02-26 22:00:32 +0000604
605 //We still can't be sure it isn't a bytecode archive
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000606 if (!loadArchive(0))
607 return false;
Reid Spencereaa06bb2005-02-26 22:00:32 +0000608
609 std::vector<Module *> Modules;
610 std::string ErrorMessage;
611
Chris Lattner7c439922005-09-23 06:22:58 +0000612 // Scan the archive, trying to load a bytecode member. We only load one to
613 // see if this works.
614 for (iterator I = begin(), E = end(); I != E; ++I) {
615 if (!I->isBytecode() && !I->isCompressedBytecode())
616 continue;
617
618 std::string FullMemberName =
619 archPath.toString() + "(" + I->getPath().toString() + ")";
Chris Lattnerc1d56242007-05-06 09:28:33 +0000620
621 MemoryBuffer *Buffer =
622 MemoryBuffer::getNewMemBuffer(I->getSize(), FullMemberName.c_str());
623 memcpy((char*)Buffer->getBufferStart(), I->getData(), I->getSize());
624 Module *M = ParseBitcodeFile(Buffer);
625 delete Buffer;
Chris Lattner7c439922005-09-23 06:22:58 +0000626 if (!M)
627 return false; // Couldn't parse bytecode, not a bytecode archive.
628 delete M;
629 return true;
630 }
631
632 return false;
Reid Spencereaa06bb2005-02-26 22:00:32 +0000633}