blob: 6946ddb7681087b6e8cb5782c768db48c5bef014 [file] [log] [blame]
Reid Spencerf9d7a512004-11-14 21:58:33 +00001//===-- ArchiveReader.cpp - Read LLVM archive files -------------*- C++ -*-===//
Misha Brukman2b37d7c2005-04-21 21:13:18 +00002//
John Criswellb576c942003-10-20 19:43:21 +00003// The LLVM Compiler Infrastructure
4//
Chris Lattner4ee451d2007-12-29 20:36:04 +00005// This file is distributed under the University of Illinois Open Source
6// License. See LICENSE.TXT for details.
Misha Brukman2b37d7c2005-04-21 21:13:18 +00007//
John Criswellb576c942003-10-20 19:43:21 +00008//===----------------------------------------------------------------------===//
Chris Lattner968cfd02003-04-19 21:45:34 +00009//
Gabor Greifa99be512007-07-05 17:07:56 +000010// Builds up standard unix archive files (.a) containing LLVM bitcode.
Chris Lattner968cfd02003-04-19 21:45:34 +000011//
12//===----------------------------------------------------------------------===//
13
Chandler Carruthd04a8d42012-12-03 16:50:05 +000014#include "llvm/Bitcode/Archive.h"
Reid Spencerf9d7a512004-11-14 21:58:33 +000015#include "ArchiveInternals.h"
Matt Arsenault7375f4f2013-02-26 21:20:35 +000016#include "llvm/ADT/OwningPtr.h"
Rafael Espindola603d6b52012-01-23 03:41:53 +000017#include "llvm/ADT/SmallPtrSet.h"
Chris Lattnere07c15c2007-05-06 06:18:07 +000018#include "llvm/Bitcode/ReaderWriter.h"
Chandler Carruth0b8c9a82013-01-02 11:36:10 +000019#include "llvm/IR/Module.h"
Rafael Espindola9e26acb2013-06-11 17:28:56 +000020#include "llvm/Support/FileSystem.h"
Chandler Carruthd04a8d42012-12-03 16:50:05 +000021#include "llvm/Support/MemoryBuffer.h"
Benjamin Kramer901b8582012-03-23 11:35:30 +000022#include <cstdio>
Dan Gohmand68a0762009-01-05 17:59:02 +000023#include <cstdlib>
Chris Lattner3446ae82004-01-10 19:00:15 +000024using namespace llvm;
Brian Gaeked0fde302003-11-11 22:41:34 +000025
Reid Spencerf9d7a512004-11-14 21:58:33 +000026/// Read a variable-bit-rate encoded unsigned integer
Dan Gohman844731a2008-05-13 00:00:25 +000027static inline unsigned readInteger(const char*&At, const char*End) {
Reid Spencerf9d7a512004-11-14 21:58:33 +000028 unsigned Shift = 0;
29 unsigned Result = 0;
Misha Brukman2b37d7c2005-04-21 21:13:18 +000030
Reid Spencerf9d7a512004-11-14 21:58:33 +000031 do {
Misha Brukman2b37d7c2005-04-21 21:13:18 +000032 if (At == End)
Reid Spencer8d8a7ff2006-07-07 20:56:50 +000033 return Result;
Reid Spencerf9d7a512004-11-14 21:58:33 +000034 Result |= (unsigned)((*At++) & 0x7F) << Shift;
35 Shift += 7;
36 } while (At[-1] & 0x80);
37 return Result;
Chris Lattner968cfd02003-04-19 21:45:34 +000038}
39
Reid Spencerf9d7a512004-11-14 21:58:33 +000040// Completely parse the Archive's symbol table and populate symTab member var.
Reid Spencer8d8a7ff2006-07-07 20:56:50 +000041bool
42Archive::parseSymbolTable(const void* data, unsigned size, std::string* error) {
Reid Spencerf9d7a512004-11-14 21:58:33 +000043 const char* At = (const char*) data;
44 const char* End = At + size;
45 while (At < End) {
46 unsigned offset = 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 offset!";
50 return false;
51 }
Reid Spencerf9d7a512004-11-14 21:58:33 +000052 unsigned length = readInteger(At, End);
Reid Spencer8d8a7ff2006-07-07 20:56:50 +000053 if (At == End) {
54 if (error)
55 *error = "Ran out of data reading vbr_uint for symtab length!";
56 return false;
57 }
58 if (At + length > End) {
59 if (error)
60 *error = "Malformed symbol table: length not consistent with size";
61 return false;
62 }
Reid Spencerf9d7a512004-11-14 21:58:33 +000063 // we don't care if it can't be inserted (duplicate entry)
Reid Spencer9a29db42004-11-20 07:29:40 +000064 symTab.insert(std::make_pair(std::string(At, length), offset));
Reid Spencerf9d7a512004-11-14 21:58:33 +000065 At += length;
66 }
67 symTabSize = size;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +000068 return true;
Chris Lattner968cfd02003-04-19 21:45:34 +000069}
70
Reid Spencerf9d7a512004-11-14 21:58:33 +000071// This member parses an ArchiveMemberHeader that is presumed to be pointed to
72// by At. The At pointer is updated to the byte just after the header, which
Misha Brukman2b37d7c2005-04-21 21:13:18 +000073// can be variable in size.
Reid Spencerf9d7a512004-11-14 21:58:33 +000074ArchiveMember*
Reid Spencer8d8a7ff2006-07-07 20:56:50 +000075Archive::parseMemberHeader(const char*& At, const char* End, std::string* error)
76{
Reid Spencercd5561a2006-12-15 19:44:51 +000077 if (At + sizeof(ArchiveMemberHeader) >= End) {
78 if (error)
79 *error = "Unexpected end of file";
80 return 0;
81 }
Reid Spencerf9d7a512004-11-14 21:58:33 +000082
83 // Cast archive member header
Roman Divacky651e8002012-09-05 22:09:23 +000084 const ArchiveMemberHeader* Hdr = (const ArchiveMemberHeader*)At;
Reid Spencerf9d7a512004-11-14 21:58:33 +000085 At += sizeof(ArchiveMemberHeader);
86
Reid Spencerf9d7a512004-11-14 21:58:33 +000087 int flags = 0;
88 int MemberSize = atoi(Hdr->size);
Rafael Espindola94bc2462012-08-10 01:57:52 +000089 assert(MemberSize >= 0);
Reid Spencerf9d7a512004-11-14 21:58:33 +000090
91 // Check the size of the member for sanity
Reid Spencer8d8a7ff2006-07-07 20:56:50 +000092 if (At + MemberSize > End) {
93 if (error)
94 *error = "invalid member length in archive file";
95 return 0;
96 }
Reid Spencerf9d7a512004-11-14 21:58:33 +000097
98 // Check the member signature
Reid Spencer8d8a7ff2006-07-07 20:56:50 +000099 if (!Hdr->checkSignature()) {
100 if (error)
101 *error = "invalid file member signature";
102 return 0;
103 }
Reid Spencerf9d7a512004-11-14 21:58:33 +0000104
105 // Convert and check the member name
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000106 // The empty name ( '/' and 15 blanks) is for a foreign (non-LLVM) symbol
107 // table. The special name "//" and 14 blanks is for a string table, used
Reid Spencerf9d7a512004-11-14 21:58:33 +0000108 // for long file names. This library doesn't generate either of those but
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000109 // it will accept them. If the name starts with #1/ and the remainder is
110 // digits, then those digits specify the length of the name that is
111 // stored immediately following the header. The special name
Gabor Greifa99be512007-07-05 17:07:56 +0000112 // __LLVM_SYM_TAB__ identifies the symbol table for LLVM bitcode.
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000113 // Anything else is a regular, short filename that is terminated with
Reid Spencerf9d7a512004-11-14 21:58:33 +0000114 // a '/' and blanks.
115
116 std::string pathname;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000117 switch (Hdr->name[0]) {
118 case '#':
119 if (Hdr->name[1] == '1' && Hdr->name[2] == '/') {
120 if (isdigit(Hdr->name[3])) {
121 unsigned len = atoi(&Hdr->name[3]);
Chris Lattner46333062010-02-04 06:19:43 +0000122 const char *nulp = (const char *)memchr(At, '\0', len);
Chris Lattner73382422010-04-25 04:44:26 +0000123 pathname.assign(At, nulp != 0 ? (uintptr_t)(nulp - At) : len);
Reid Spencerdd95e8d2004-11-17 16:13:11 +0000124 At += len;
125 MemberSize -= len;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000126 flags |= ArchiveMember::HasLongFilenameFlag;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000127 } else {
128 if (error)
129 *error = "invalid long filename";
130 return 0;
131 }
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000132 } else if (Hdr->name[1] == '_' &&
Reid Spencer9a29db42004-11-20 07:29:40 +0000133 (0 == memcmp(Hdr->name, ARFILE_LLVM_SYMTAB_NAME, 16))) {
Reid Spencerf9d7a512004-11-14 21:58:33 +0000134 // The member is using a long file name (>15 chars) format.
135 // This format is standard for 4.4BSD and Mac OSX operating
136 // systems. LLVM uses it similarly. In this format, the
137 // remainder of the name field (after #1/) specifies the
138 // length of the file name which occupy the first bytes of
139 // the member's data. The pathname already has the #1/ stripped.
140 pathname.assign(ARFILE_LLVM_SYMTAB_NAME);
141 flags |= ArchiveMember::LLVMSymbolTableFlag;
142 }
143 break;
144 case '/':
145 if (Hdr->name[1]== '/') {
Reid Spencer9a29db42004-11-20 07:29:40 +0000146 if (0 == memcmp(Hdr->name, ARFILE_STRTAB_NAME, 16)) {
Reid Spencerf9d7a512004-11-14 21:58:33 +0000147 pathname.assign(ARFILE_STRTAB_NAME);
148 flags |= ArchiveMember::StringTableFlag;
149 } else {
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000150 if (error)
151 *error = "invalid string table name";
152 return 0;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000153 }
154 } else if (Hdr->name[1] == ' ') {
Reid Spencer9a29db42004-11-20 07:29:40 +0000155 if (0 == memcmp(Hdr->name, ARFILE_SVR4_SYMTAB_NAME, 16)) {
156 pathname.assign(ARFILE_SVR4_SYMTAB_NAME);
157 flags |= ArchiveMember::SVR4SymbolTableFlag;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000158 } else {
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000159 if (error)
160 *error = "invalid SVR4 symbol table name";
161 return 0;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000162 }
163 } else if (isdigit(Hdr->name[1])) {
164 unsigned index = atoi(&Hdr->name[1]);
165 if (index < strtab.length()) {
166 const char* namep = strtab.c_str() + index;
167 const char* endp = strtab.c_str() + strtab.length();
168 const char* p = namep;
169 const char* last_p = p;
170 while (p < endp) {
171 if (*p == '\n' && *last_p == '/') {
Reid Spencer9a29db42004-11-20 07:29:40 +0000172 pathname.assign(namep, last_p - namep);
Reid Spencerf9d7a512004-11-14 21:58:33 +0000173 flags |= ArchiveMember::HasLongFilenameFlag;
174 break;
175 }
176 last_p = p;
177 p++;
178 }
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000179 if (p >= endp) {
180 if (error)
Matt Arsenaultd06c9cb2013-02-26 20:13:09 +0000181 *error = "missing name terminator in string table";
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000182 return 0;
183 }
Reid Spencerf9d7a512004-11-14 21:58:33 +0000184 } else {
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000185 if (error)
186 *error = "name index beyond string table";
187 return 0;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000188 }
189 }
190 break;
Reid Spencer9a29db42004-11-20 07:29:40 +0000191 case '_':
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000192 if (Hdr->name[1] == '_' &&
Reid Spencer9a29db42004-11-20 07:29:40 +0000193 (0 == memcmp(Hdr->name, ARFILE_BSD4_SYMTAB_NAME, 16))) {
194 pathname.assign(ARFILE_BSD4_SYMTAB_NAME);
195 flags |= ArchiveMember::BSD4SymbolTableFlag;
Reid Spencer84b9ced2004-11-23 22:35:39 +0000196 break;
Reid Spencer9a29db42004-11-20 07:29:40 +0000197 }
Reid Spencer84b9ced2004-11-23 22:35:39 +0000198 /* FALL THROUGH */
Reid Spencerf9d7a512004-11-14 21:58:33 +0000199
200 default:
Roman Divacky651e8002012-09-05 22:09:23 +0000201 const char* slash = (const char*) memchr(Hdr->name, '/', 16);
Reid Spencerf9d7a512004-11-14 21:58:33 +0000202 if (slash == 0)
Reid Spencerdd95e8d2004-11-17 16:13:11 +0000203 slash = Hdr->name + 16;
Reid Spencer9a29db42004-11-20 07:29:40 +0000204 pathname.assign(Hdr->name, slash - Hdr->name);
Reid Spencerf9d7a512004-11-14 21:58:33 +0000205 break;
206 }
207
Gabor Greifdb5565a2007-07-06 20:28:40 +0000208 // Determine if this is a bitcode file
Rafael Espindola9e26acb2013-06-11 17:28:56 +0000209 if (sys::fs::identify_magic(StringRef(At, 4)) ==
210 sys::fs::file_magic::bitcode)
211 flags |= ArchiveMember::BitcodeFlag;
212 else
213 flags &= ~ArchiveMember::BitcodeFlag;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000214
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000215 // Instantiate the ArchiveMember to be filled
216 ArchiveMember* member = new ArchiveMember(this);
217
Reid Spencerf9d7a512004-11-14 21:58:33 +0000218 // Fill in fields of the ArchiveMember
Reid Spencerf9d7a512004-11-14 21:58:33 +0000219 member->parent = this;
Reid Spencerdd04df02005-07-07 23:21:43 +0000220 member->path.set(pathname);
Reid Spencerf9d7a512004-11-14 21:58:33 +0000221 member->info.fileSize = MemberSize;
222 member->info.modTime.fromEpochTime(atoi(Hdr->date));
Reid Spencer56127622004-12-29 01:20:24 +0000223 unsigned int mode;
224 sscanf(Hdr->mode, "%o", &mode);
225 member->info.mode = mode;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000226 member->info.user = atoi(Hdr->uid);
227 member->info.group = atoi(Hdr->gid);
228 member->flags = flags;
229 member->data = At;
230
231 return member;
232}
233
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000234bool
235Archive::checkSignature(std::string* error) {
Reid Spencerf9d7a512004-11-14 21:58:33 +0000236 // Check the magic string at file's header
Chris Lattner7f6b4472008-04-01 04:26:46 +0000237 if (mapfile->getBufferSize() < 8 || memcmp(base, ARFILE_MAGIC, 8)) {
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000238 if (error)
239 *error = "invalid signature for an archive file";
240 return false;
241 }
242 return true;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000243}
244
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000245// This function loads the entire archive and fully populates its ilist with
Reid Spencerf9d7a512004-11-14 21:58:33 +0000246// the members of the archive file. This is typically used in preparation for
247// editing the contents of the archive.
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000248bool
249Archive::loadArchive(std::string* error) {
Reid Spencerf9d7a512004-11-14 21:58:33 +0000250
251 // Set up parsing
252 members.clear();
253 symTab.clear();
254 const char *At = base;
Chris Lattner7f6b4472008-04-01 04:26:46 +0000255 const char *End = mapfile->getBufferEnd();
Reid Spencerf9d7a512004-11-14 21:58:33 +0000256
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000257 if (!checkSignature(error))
258 return false;
259
Reid Spencerf9d7a512004-11-14 21:58:33 +0000260 At += 8; // Skip the magic string.
261
262 bool seenSymbolTable = false;
263 bool foundFirstFile = false;
264 while (At < End) {
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000265 // parse the member header
Reid Spencerf9d7a512004-11-14 21:58:33 +0000266 const char* Save = At;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000267 ArchiveMember* mbr = parseMemberHeader(At, End, error);
268 if (!mbr)
269 return false;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000270
271 // check if this is the foreign symbol table
Reid Spencer9a29db42004-11-20 07:29:40 +0000272 if (mbr->isSVR4SymbolTable() || mbr->isBSD4SymbolTable()) {
Reid Spencer4a980d12004-11-16 06:47:19 +0000273 // We just save this but don't do anything special
274 // with it. It doesn't count as the "first file".
Reid Spencer9a29db42004-11-20 07:29:40 +0000275 if (foreignST) {
276 // What? Multiple foreign symbol tables? Just chuck it
277 // and retain the last one found.
278 delete foreignST;
279 }
Reid Spencer4a980d12004-11-16 06:47:19 +0000280 foreignST = mbr;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000281 At += mbr->getSize();
Reid Spencer6405c9e2004-11-19 17:08:00 +0000282 if ((intptr_t(At) & 1) == 1)
Reid Spencerf9d7a512004-11-14 21:58:33 +0000283 At++;
284 } else if (mbr->isStringTable()) {
Reid Spencer4a980d12004-11-16 06:47:19 +0000285 // Simply suck the entire string table into a string
286 // variable. This will be used to get the names of the
287 // members that use the "/ddd" format for their names
288 // (SVR4 style long names).
Reid Spencer9a29db42004-11-20 07:29:40 +0000289 strtab.assign(At, mbr->getSize());
Reid Spencerf9d7a512004-11-14 21:58:33 +0000290 At += mbr->getSize();
Reid Spencer6405c9e2004-11-19 17:08:00 +0000291 if ((intptr_t(At) & 1) == 1)
Reid Spencerf9d7a512004-11-14 21:58:33 +0000292 At++;
293 delete mbr;
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000294 } else if (mbr->isLLVMSymbolTable()) {
Reid Spencer4a980d12004-11-16 06:47:19 +0000295 // This is the LLVM symbol table for the archive. If we've seen it
296 // already, its an error. Otherwise, parse the symbol table and move on.
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000297 if (seenSymbolTable) {
298 if (error)
299 *error = "invalid archive: multiple symbol tables";
300 return false;
301 }
302 if (!parseSymbolTable(mbr->getData(), mbr->getSize(), error))
303 return false;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000304 seenSymbolTable = true;
305 At += mbr->getSize();
Reid Spencer6405c9e2004-11-19 17:08:00 +0000306 if ((intptr_t(At) & 1) == 1)
Reid Spencerf9d7a512004-11-14 21:58:33 +0000307 At++;
Reid Spencer4a980d12004-11-16 06:47:19 +0000308 delete mbr; // We don't need this member in the list of members.
Reid Spencerf9d7a512004-11-14 21:58:33 +0000309 } else {
Reid Spencer4a980d12004-11-16 06:47:19 +0000310 // This is just a regular file. If its the first one, save its offset.
311 // Otherwise just push it on the list and move on to the next file.
Reid Spencerf9d7a512004-11-14 21:58:33 +0000312 if (!foundFirstFile) {
313 firstFileOffset = Save - base;
314 foundFirstFile = true;
315 }
316 members.push_back(mbr);
317 At += mbr->getSize();
Reid Spencer6405c9e2004-11-19 17:08:00 +0000318 if ((intptr_t(At) & 1) == 1)
Reid Spencerf9d7a512004-11-14 21:58:33 +0000319 At++;
320 }
321 }
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000322 return true;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000323}
324
325// Open and completely load the archive file.
326Archive*
Matt Arsenault7375f4f2013-02-26 21:20:35 +0000327Archive::OpenAndLoad(const sys::Path& File, LLVMContext& C,
Owen Anderson8b477ed2009-07-01 16:58:40 +0000328 std::string* ErrorMessage) {
Matt Arsenault7375f4f2013-02-26 21:20:35 +0000329 OwningPtr<Archive> result ( new Archive(File, C));
Reid Spencer0ff2d312006-08-24 23:45:08 +0000330 if (result->mapToMemory(ErrorMessage))
Matt Arsenault7375f4f2013-02-26 21:20:35 +0000331 return NULL;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000332 if (!result->loadArchive(ErrorMessage))
Matt Arsenault7375f4f2013-02-26 21:20:35 +0000333 return NULL;
334 return result.take();
Reid Spencerf9d7a512004-11-14 21:58:33 +0000335}
336
Gabor Greifa99be512007-07-05 17:07:56 +0000337// Get all the bitcode modules from the archive
Reid Spencerf9d7a512004-11-14 21:58:33 +0000338bool
Owen Anderson8b477ed2009-07-01 16:58:40 +0000339Archive::getAllModules(std::vector<Module*>& Modules,
340 std::string* ErrMessage) {
Reid Spencerf9d7a512004-11-14 21:58:33 +0000341
342 for (iterator I=begin(), E=end(); I != E; ++I) {
Gabor Greife75ca3d2007-07-06 13:38:17 +0000343 if (I->isBitcode()) {
Chris Lattner74382b72009-08-23 22:45:37 +0000344 std::string FullMemberName = archPath.str() +
345 "(" + I->getPath().str() + ")";
Chris Lattnerc1d56242007-05-06 09:28:33 +0000346 MemoryBuffer *Buffer =
Benjamin Kramer9d44e702010-04-19 16:15:31 +0000347 MemoryBuffer::getMemBufferCopy(StringRef(I->getData(), I->getSize()),
348 FullMemberName.c_str());
Chris Lattnere07c15c2007-05-06 06:18:07 +0000349
Owen Anderson8b477ed2009-07-01 16:58:40 +0000350 Module *M = ParseBitcodeFile(Buffer, Context, ErrMessage);
Chris Lattnerc1d56242007-05-06 09:28:33 +0000351 delete Buffer;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000352 if (!M)
353 return true;
354
355 Modules.push_back(M);
356 }
357 }
Brian Gaeke2c61d7b2003-11-16 23:08:48 +0000358 return false;
359}
Chris Lattner968cfd02003-04-19 21:45:34 +0000360
Reid Spencerf9d7a512004-11-14 21:58:33 +0000361// Load just the symbol table from the archive file
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000362bool
363Archive::loadSymbolTable(std::string* ErrorMsg) {
Chris Lattner968cfd02003-04-19 21:45:34 +0000364
Reid Spencerf9d7a512004-11-14 21:58:33 +0000365 // Set up parsing
366 members.clear();
367 symTab.clear();
368 const char *At = base;
Chris Lattner7f6b4472008-04-01 04:26:46 +0000369 const char *End = mapfile->getBufferEnd();
Chris Lattner968cfd02003-04-19 21:45:34 +0000370
Reid Spencerf9d7a512004-11-14 21:58:33 +0000371 // Make sure we're dealing with an archive
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000372 if (!checkSignature(ErrorMsg))
373 return false;
Chris Lattner968cfd02003-04-19 21:45:34 +0000374
Reid Spencerf9d7a512004-11-14 21:58:33 +0000375 At += 8; // Skip signature
376
377 // Parse the first file member header
378 const char* FirstFile = At;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000379 ArchiveMember* mbr = parseMemberHeader(At, End, ErrorMsg);
380 if (!mbr)
381 return false;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000382
Reid Spencer9a29db42004-11-20 07:29:40 +0000383 if (mbr->isSVR4SymbolTable() || mbr->isBSD4SymbolTable()) {
Reid Spencerf9d7a512004-11-14 21:58:33 +0000384 // Skip the foreign symbol table, we don't do anything with it
385 At += mbr->getSize();
Reid Spencer6405c9e2004-11-19 17:08:00 +0000386 if ((intptr_t(At) & 1) == 1)
Reid Spencerb3231132004-11-15 01:40:20 +0000387 At++;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000388 delete mbr;
389
Reid Spencerb3231132004-11-15 01:40:20 +0000390 // Read the next one
Reid Spencerf9d7a512004-11-14 21:58:33 +0000391 FirstFile = At;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000392 mbr = parseMemberHeader(At, End, ErrorMsg);
393 if (!mbr) {
394 delete mbr;
395 return false;
396 }
Reid Spencerb3231132004-11-15 01:40:20 +0000397 }
398
399 if (mbr->isStringTable()) {
400 // Process the string table entry
Reid Spencer9a29db42004-11-20 07:29:40 +0000401 strtab.assign((const char*)mbr->getData(), mbr->getSize());
Reid Spencerb3231132004-11-15 01:40:20 +0000402 At += mbr->getSize();
Reid Spencer6405c9e2004-11-19 17:08:00 +0000403 if ((intptr_t(At) & 1) == 1)
Reid Spencerb3231132004-11-15 01:40:20 +0000404 At++;
405 delete mbr;
406 // Get the next one
407 FirstFile = At;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000408 mbr = parseMemberHeader(At, End, ErrorMsg);
409 if (!mbr) {
410 delete mbr;
411 return false;
412 }
Chris Lattner968cfd02003-04-19 21:45:34 +0000413 }
414
Reid Spencerf9d7a512004-11-14 21:58:33 +0000415 // See if its the symbol table
416 if (mbr->isLLVMSymbolTable()) {
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000417 if (!parseSymbolTable(mbr->getData(), mbr->getSize(), ErrorMsg)) {
418 delete mbr;
419 return false;
420 }
421
Reid Spencer8dde18f2004-11-28 03:13:02 +0000422 At += mbr->getSize();
Reid Spencer6405c9e2004-11-19 17:08:00 +0000423 if ((intptr_t(At) & 1) == 1)
Reid Spencer8dde18f2004-11-28 03:13:02 +0000424 At++;
Reid Spencerffb4d622006-05-12 17:56:20 +0000425 delete mbr;
426 // Can't be any more symtab headers so just advance
Reid Spencer8dde18f2004-11-28 03:13:02 +0000427 FirstFile = At;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000428 } else {
429 // There's no symbol table in the file. We have to rebuild it from scratch
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000430 // because the intent of this method is to get the symbol table loaded so
431 // it can be searched efficiently.
Reid Spencerf9d7a512004-11-14 21:58:33 +0000432 // Add the member to the members list
433 members.push_back(mbr);
434 }
435
436 firstFileOffset = FirstFile - base;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000437 return true;
Chris Lattner968cfd02003-04-19 21:45:34 +0000438}
439
Reid Spencerf9d7a512004-11-14 21:58:33 +0000440// Open the archive and load just the symbol tables
Matt Arsenault7375f4f2013-02-26 21:20:35 +0000441Archive* Archive::OpenAndLoadSymbols(const sys::Path& File,
Owen Anderson4434ed42009-07-01 23:13:44 +0000442 LLVMContext& C,
Owen Anderson8b477ed2009-07-01 16:58:40 +0000443 std::string* ErrorMessage) {
Matt Arsenault7375f4f2013-02-26 21:20:35 +0000444 OwningPtr<Archive> result ( new Archive(File, C) );
Reid Spencer0ff2d312006-08-24 23:45:08 +0000445 if (result->mapToMemory(ErrorMessage))
Matt Arsenault7375f4f2013-02-26 21:20:35 +0000446 return NULL;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000447 if (!result->loadSymbolTable(ErrorMessage))
Matt Arsenault7375f4f2013-02-26 21:20:35 +0000448 return NULL;
449 return result.take();
Reid Spencerf9d7a512004-11-14 21:58:33 +0000450}
Chris Lattner968cfd02003-04-19 21:45:34 +0000451
Jeffrey Yasskinf0356fe2010-01-27 20:34:15 +0000452// Look up one symbol in the symbol table and return the module that defines
453// that symbol.
454Module*
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000455Archive::findModuleDefiningSymbol(const std::string& symbol,
456 std::string* ErrMsg) {
Reid Spencerf9d7a512004-11-14 21:58:33 +0000457 SymTabType::iterator SI = symTab.find(symbol);
458 if (SI == symTab.end())
459 return 0;
460
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000461 // The symbol table was previously constructed assuming that the members were
Reid Spencerf9d7a512004-11-14 21:58:33 +0000462 // written without the symbol table header. Because VBR encoding is used, the
463 // values could not be adjusted to account for the offset of the symbol table
464 // because that could affect the size of the symbol table due to VBR encoding.
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000465 // We now have to account for this by adjusting the offset by the size of the
Reid Spencerf9d7a512004-11-14 21:58:33 +0000466 // symbol table and its header.
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000467 unsigned fileOffset =
Reid Spencerf9d7a512004-11-14 21:58:33 +0000468 SI->second + // offset in symbol-table-less file
469 firstFileOffset; // add offset to first "real" file in archive
470
471 // See if the module is already loaded
472 ModuleMap::iterator MI = modules.find(fileOffset);
473 if (MI != modules.end())
474 return MI->second.first;
475
476 // Module hasn't been loaded yet, we need to load it
477 const char* modptr = base + fileOffset;
Chris Lattner7f6b4472008-04-01 04:26:46 +0000478 ArchiveMember* mbr = parseMemberHeader(modptr, mapfile->getBufferEnd(),
479 ErrMsg);
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000480 if (!mbr)
Reid Spencer0b5a5042006-08-25 17:43:11 +0000481 return 0;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000482
Jeffrey Yasskinf0356fe2010-01-27 20:34:15 +0000483 // Now, load the bitcode module to get the Module.
Chris Lattner74382b72009-08-23 22:45:37 +0000484 std::string FullMemberName = archPath.str() + "(" +
485 mbr->getPath().str() + ")";
Benjamin Kramer9d44e702010-04-19 16:15:31 +0000486 MemoryBuffer *Buffer =
487 MemoryBuffer::getMemBufferCopy(StringRef(mbr->getData(), mbr->getSize()),
488 FullMemberName.c_str());
Chris Lattnerc1d56242007-05-06 09:28:33 +0000489
Jeffrey Yasskinf0356fe2010-01-27 20:34:15 +0000490 Module *m = getLazyBitcodeModule(Buffer, Context, ErrMsg);
491 if (!m)
Reid Spencer0b5a5042006-08-25 17:43:11 +0000492 return 0;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000493
Jeffrey Yasskinf0356fe2010-01-27 20:34:15 +0000494 modules.insert(std::make_pair(fileOffset, std::make_pair(m, mbr)));
Reid Spencerf9d7a512004-11-14 21:58:33 +0000495
Jeffrey Yasskinf0356fe2010-01-27 20:34:15 +0000496 return m;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000497}
498
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000499// Look up multiple symbols in the symbol table and return a set of
Jeffrey Yasskinf0356fe2010-01-27 20:34:15 +0000500// Modules that define those symbols.
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000501bool
Reid Spencer7783e8a2004-11-19 03:18:22 +0000502Archive::findModulesDefiningSymbols(std::set<std::string>& symbols,
Rafael Espindola603d6b52012-01-23 03:41:53 +0000503 SmallVectorImpl<Module*>& result,
Chris Lattnerf2e292c2007-02-07 21:41:02 +0000504 std::string* error) {
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
Chris Lattner7f6b4472008-04-01 04:26:46 +0000518 const char* At = base + firstFileOffset;
519 const char* End = mapfile->getBufferEnd();
Reid Spencer766b7932004-11-15 01:20:11 +0000520
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
Gabor Greife75ca3d2007-07-06 13:38:17 +0000531 if (mbr->isBitcode()) {
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000532 // Get the symbols
Reid Spencer766b7932004-11-15 01:20:11 +0000533 std::vector<std::string> symbols;
Chris Lattner74382b72009-08-23 22:45:37 +0000534 std::string FullMemberName = archPath.str() + "(" +
535 mbr->getPath().str() + ")";
Jeffrey Yasskinf0356fe2010-01-27 20:34:15 +0000536 Module* M =
Benjamin Kramer9d44e702010-04-19 16:15:31 +0000537 GetBitcodeSymbols(At, mbr->getSize(), FullMemberName, Context,
538 symbols, error);
Reid Spencer766b7932004-11-15 01:20:11 +0000539
Jeffrey Yasskinf0356fe2010-01-27 20:34:15 +0000540 if (M) {
Reid Spencer766b7932004-11-15 01:20:11 +0000541 // Insert the module's symbols into the symbol table
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000542 for (std::vector<std::string>::iterator I = symbols.begin(),
Reid Spencer766b7932004-11-15 01:20:11 +0000543 E=symbols.end(); I != E; ++I ) {
Reid Spencer9a29db42004-11-20 07:29:40 +0000544 symTab.insert(std::make_pair(*I, offset));
Reid Spencer766b7932004-11-15 01:20:11 +0000545 }
Jeffrey Yasskinf0356fe2010-01-27 20:34:15 +0000546 // Insert the Module and the ArchiveMember into the table of
Reid Spencer766b7932004-11-15 01:20:11 +0000547 // modules.
Jeffrey Yasskinf0356fe2010-01-27 20:34:15 +0000548 modules.insert(std::make_pair(offset, std::make_pair(M, mbr)));
Reid Spencer766b7932004-11-15 01:20:11 +0000549 } else {
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000550 if (error)
Gabor Greifa99be512007-07-05 17:07:56 +0000551 *error = "Can't parse bitcode member: " +
Chris Lattner74382b72009-08-23 22:45:37 +0000552 mbr->getPath().str() + ": " + *error;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000553 delete mbr;
554 return false;
Reid Spencer766b7932004-11-15 01:20:11 +0000555 }
556 }
Reid Spencerb3231132004-11-15 01:40:20 +0000557
558 // Go to the next file location
559 At += mbr->getSize();
Reid Spencer6405c9e2004-11-19 17:08:00 +0000560 if ((intptr_t(At) & 1) == 1)
Reid Spencerb3231132004-11-15 01:40:20 +0000561 At++;
Reid Spencer766b7932004-11-15 01:20:11 +0000562 }
563 }
564
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000565 // At this point we have a valid symbol table (one way or another) so we
Reid Spencer766b7932004-11-15 01:20:11 +0000566 // just use it to quickly find the symbols requested.
567
Rafael Espindola603d6b52012-01-23 03:41:53 +0000568 SmallPtrSet<Module*, 16> Added;
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000569 for (std::set<std::string>::iterator I=symbols.begin(),
Rafael Espindola4254df32012-01-23 05:07:16 +0000570 Next = I,
571 E=symbols.end(); I != E; I = Next) {
572 // Increment Next before we invalidate it.
573 ++Next;
574
Reid Spencer7783e8a2004-11-19 03:18:22 +0000575 // See if this symbol exists
Jeffrey Yasskinf0356fe2010-01-27 20:34:15 +0000576 Module* m = findModuleDefiningSymbol(*I,error);
Rafael Espindola603d6b52012-01-23 03:41:53 +0000577 if (!m)
578 continue;
579 bool NewMember = Added.insert(m);
580 if (!NewMember)
581 continue;
Reid Spencer7783e8a2004-11-19 03:18:22 +0000582
Rafael Espindola603d6b52012-01-23 03:41:53 +0000583 // The symbol exists, insert the Module into our result.
584 result.push_back(m);
585
586 // Remove the symbol now that its been resolved.
587 symbols.erase(I);
Reid Spencerf9d7a512004-11-14 21:58:33 +0000588 }
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000589 return true;
Chris Lattner968cfd02003-04-19 21:45:34 +0000590}
Reid Spencereaa06bb2005-02-26 22:00:32 +0000591
Gabor Greifa99be512007-07-05 17:07:56 +0000592bool Archive::isBitcodeArchive() {
Chris Lattner7c439922005-09-23 06:22:58 +0000593 // Make sure the symTab has been loaded. In most cases this should have been
594 // done when the archive was constructed, but still, this is just in case.
Dan Gohman30359592008-01-29 13:02:09 +0000595 if (symTab.empty())
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000596 if (!loadSymbolTable(0))
597 return false;
Reid Spencereaa06bb2005-02-26 22:00:32 +0000598
Chris Lattner7c439922005-09-23 06:22:58 +0000599 // Now that we know it's been loaded, return true
600 // if it has a size
601 if (symTab.size()) return true;
Reid Spencereaa06bb2005-02-26 22:00:32 +0000602
Gabor Greifa99be512007-07-05 17:07:56 +0000603 // We still can't be sure it isn't a bitcode archive
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000604 if (!loadArchive(0))
605 return false;
Reid Spencereaa06bb2005-02-26 22:00:32 +0000606
607 std::vector<Module *> Modules;
608 std::string ErrorMessage;
609
Gabor Greifa99be512007-07-05 17:07:56 +0000610 // Scan the archive, trying to load a bitcode member. We only load one to
Chris Lattner7c439922005-09-23 06:22:58 +0000611 // see if this works.
612 for (iterator I = begin(), E = end(); I != E; ++I) {
Gabor Greife75ca3d2007-07-06 13:38:17 +0000613 if (!I->isBitcode())
Chris Lattner7c439922005-09-23 06:22:58 +0000614 continue;
615
616 std::string FullMemberName =
Chris Lattner74382b72009-08-23 22:45:37 +0000617 archPath.str() + "(" + I->getPath().str() + ")";
Chris Lattnerc1d56242007-05-06 09:28:33 +0000618
619 MemoryBuffer *Buffer =
Benjamin Kramer9d44e702010-04-19 16:15:31 +0000620 MemoryBuffer::getMemBufferCopy(StringRef(I->getData(), I->getSize()),
621 FullMemberName.c_str());
Owen Anderson8b477ed2009-07-01 16:58:40 +0000622 Module *M = ParseBitcodeFile(Buffer, Context);
Chris Lattnerc1d56242007-05-06 09:28:33 +0000623 delete Buffer;
Chris Lattner7c439922005-09-23 06:22:58 +0000624 if (!M)
Gabor Greifa99be512007-07-05 17:07:56 +0000625 return false; // Couldn't parse bitcode, not a bitcode archive.
Chris Lattner7c439922005-09-23 06:22:58 +0000626 delete M;
627 return true;
628 }
629
630 return false;
Reid Spencereaa06bb2005-02-26 22:00:32 +0000631}