blob: 8ad01dca0be6a21172011eb6ba97bf0c6ddcef6d [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
Rafael Espindola8496fae2013-06-17 15:47:20 +000014#include "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// This member parses an ArchiveMemberHeader that is presumed to be pointed to
41// by At. The At pointer is updated to the byte just after the header, which
Misha Brukman2b37d7c2005-04-21 21:13:18 +000042// can be variable in size.
Reid Spencerf9d7a512004-11-14 21:58:33 +000043ArchiveMember*
Reid Spencer8d8a7ff2006-07-07 20:56:50 +000044Archive::parseMemberHeader(const char*& At, const char* End, std::string* error)
45{
Reid Spencercd5561a2006-12-15 19:44:51 +000046 if (At + sizeof(ArchiveMemberHeader) >= End) {
47 if (error)
48 *error = "Unexpected end of file";
49 return 0;
50 }
Reid Spencerf9d7a512004-11-14 21:58:33 +000051
52 // Cast archive member header
Roman Divacky651e8002012-09-05 22:09:23 +000053 const ArchiveMemberHeader* Hdr = (const ArchiveMemberHeader*)At;
Reid Spencerf9d7a512004-11-14 21:58:33 +000054 At += sizeof(ArchiveMemberHeader);
55
Reid Spencerf9d7a512004-11-14 21:58:33 +000056 int flags = 0;
57 int MemberSize = atoi(Hdr->size);
Rafael Espindola94bc2462012-08-10 01:57:52 +000058 assert(MemberSize >= 0);
Reid Spencerf9d7a512004-11-14 21:58:33 +000059
60 // Check the size of the member for sanity
Reid Spencer8d8a7ff2006-07-07 20:56:50 +000061 if (At + MemberSize > End) {
62 if (error)
63 *error = "invalid member length in archive file";
64 return 0;
65 }
Reid Spencerf9d7a512004-11-14 21:58:33 +000066
67 // Check the member signature
Reid Spencer8d8a7ff2006-07-07 20:56:50 +000068 if (!Hdr->checkSignature()) {
69 if (error)
70 *error = "invalid file member signature";
71 return 0;
72 }
Reid Spencerf9d7a512004-11-14 21:58:33 +000073
74 // Convert and check the member name
Misha Brukman2b37d7c2005-04-21 21:13:18 +000075 // The empty name ( '/' and 15 blanks) is for a foreign (non-LLVM) symbol
76 // table. The special name "//" and 14 blanks is for a string table, used
Reid Spencerf9d7a512004-11-14 21:58:33 +000077 // for long file names. This library doesn't generate either of those but
Misha Brukman2b37d7c2005-04-21 21:13:18 +000078 // it will accept them. If the name starts with #1/ and the remainder is
79 // digits, then those digits specify the length of the name that is
Rafael Espindola250bfb12013-06-14 23:25:53 +000080 // stored immediately following the header. Anything else is a regular, short
81 // filename that is terminated with a '/' and blanks.
Reid Spencerf9d7a512004-11-14 21:58:33 +000082
83 std::string pathname;
Reid Spencerf9d7a512004-11-14 21:58:33 +000084 switch (Hdr->name[0]) {
85 case '#':
86 if (Hdr->name[1] == '1' && Hdr->name[2] == '/') {
87 if (isdigit(Hdr->name[3])) {
88 unsigned len = atoi(&Hdr->name[3]);
Chris Lattner46333062010-02-04 06:19:43 +000089 const char *nulp = (const char *)memchr(At, '\0', len);
Chris Lattner73382422010-04-25 04:44:26 +000090 pathname.assign(At, nulp != 0 ? (uintptr_t)(nulp - At) : len);
Reid Spencerdd95e8d2004-11-17 16:13:11 +000091 At += len;
92 MemberSize -= len;
Reid Spencerf9d7a512004-11-14 21:58:33 +000093 flags |= ArchiveMember::HasLongFilenameFlag;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +000094 } else {
95 if (error)
96 *error = "invalid long filename";
97 return 0;
98 }
Reid Spencerf9d7a512004-11-14 21:58:33 +000099 }
100 break;
101 case '/':
102 if (Hdr->name[1]== '/') {
Reid Spencer9a29db42004-11-20 07:29:40 +0000103 if (0 == memcmp(Hdr->name, ARFILE_STRTAB_NAME, 16)) {
Reid Spencerf9d7a512004-11-14 21:58:33 +0000104 pathname.assign(ARFILE_STRTAB_NAME);
105 flags |= ArchiveMember::StringTableFlag;
106 } else {
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000107 if (error)
108 *error = "invalid string table name";
109 return 0;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000110 }
111 } else if (Hdr->name[1] == ' ') {
Reid Spencer9a29db42004-11-20 07:29:40 +0000112 if (0 == memcmp(Hdr->name, ARFILE_SVR4_SYMTAB_NAME, 16)) {
113 pathname.assign(ARFILE_SVR4_SYMTAB_NAME);
114 flags |= ArchiveMember::SVR4SymbolTableFlag;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000115 } else {
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000116 if (error)
117 *error = "invalid SVR4 symbol table name";
118 return 0;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000119 }
120 } else if (isdigit(Hdr->name[1])) {
121 unsigned index = atoi(&Hdr->name[1]);
122 if (index < strtab.length()) {
123 const char* namep = strtab.c_str() + index;
124 const char* endp = strtab.c_str() + strtab.length();
125 const char* p = namep;
126 const char* last_p = p;
127 while (p < endp) {
128 if (*p == '\n' && *last_p == '/') {
Reid Spencer9a29db42004-11-20 07:29:40 +0000129 pathname.assign(namep, last_p - namep);
Reid Spencerf9d7a512004-11-14 21:58:33 +0000130 flags |= ArchiveMember::HasLongFilenameFlag;
131 break;
132 }
133 last_p = p;
134 p++;
135 }
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000136 if (p >= endp) {
137 if (error)
Matt Arsenaultd06c9cb2013-02-26 20:13:09 +0000138 *error = "missing name terminator in string table";
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000139 return 0;
140 }
Reid Spencerf9d7a512004-11-14 21:58:33 +0000141 } else {
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000142 if (error)
143 *error = "name index beyond string table";
144 return 0;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000145 }
146 }
147 break;
Reid Spencer9a29db42004-11-20 07:29:40 +0000148 case '_':
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000149 if (Hdr->name[1] == '_' &&
Reid Spencer9a29db42004-11-20 07:29:40 +0000150 (0 == memcmp(Hdr->name, ARFILE_BSD4_SYMTAB_NAME, 16))) {
151 pathname.assign(ARFILE_BSD4_SYMTAB_NAME);
152 flags |= ArchiveMember::BSD4SymbolTableFlag;
Reid Spencer84b9ced2004-11-23 22:35:39 +0000153 break;
Reid Spencer9a29db42004-11-20 07:29:40 +0000154 }
Reid Spencer84b9ced2004-11-23 22:35:39 +0000155 /* FALL THROUGH */
Reid Spencerf9d7a512004-11-14 21:58:33 +0000156
157 default:
Roman Divacky651e8002012-09-05 22:09:23 +0000158 const char* slash = (const char*) memchr(Hdr->name, '/', 16);
Reid Spencerf9d7a512004-11-14 21:58:33 +0000159 if (slash == 0)
Reid Spencerdd95e8d2004-11-17 16:13:11 +0000160 slash = Hdr->name + 16;
Reid Spencer9a29db42004-11-20 07:29:40 +0000161 pathname.assign(Hdr->name, slash - Hdr->name);
Reid Spencerf9d7a512004-11-14 21:58:33 +0000162 break;
163 }
164
Gabor Greifdb5565a2007-07-06 20:28:40 +0000165 // Determine if this is a bitcode file
Rafael Espindola9e26acb2013-06-11 17:28:56 +0000166 if (sys::fs::identify_magic(StringRef(At, 4)) ==
167 sys::fs::file_magic::bitcode)
168 flags |= ArchiveMember::BitcodeFlag;
169 else
170 flags &= ~ArchiveMember::BitcodeFlag;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000171
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000172 // Instantiate the ArchiveMember to be filled
173 ArchiveMember* member = new ArchiveMember(this);
174
Reid Spencerf9d7a512004-11-14 21:58:33 +0000175 // Fill in fields of the ArchiveMember
Reid Spencerf9d7a512004-11-14 21:58:33 +0000176 member->parent = this;
Reid Spencerdd04df02005-07-07 23:21:43 +0000177 member->path.set(pathname);
Reid Spencerf9d7a512004-11-14 21:58:33 +0000178 member->info.fileSize = MemberSize;
179 member->info.modTime.fromEpochTime(atoi(Hdr->date));
Reid Spencer56127622004-12-29 01:20:24 +0000180 unsigned int mode;
181 sscanf(Hdr->mode, "%o", &mode);
182 member->info.mode = mode;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000183 member->info.user = atoi(Hdr->uid);
184 member->info.group = atoi(Hdr->gid);
185 member->flags = flags;
186 member->data = At;
187
188 return member;
189}
190
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000191bool
192Archive::checkSignature(std::string* error) {
Reid Spencerf9d7a512004-11-14 21:58:33 +0000193 // Check the magic string at file's header
Chris Lattner7f6b4472008-04-01 04:26:46 +0000194 if (mapfile->getBufferSize() < 8 || memcmp(base, ARFILE_MAGIC, 8)) {
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000195 if (error)
196 *error = "invalid signature for an archive file";
197 return false;
198 }
199 return true;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000200}
201
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000202// This function loads the entire archive and fully populates its ilist with
Reid Spencerf9d7a512004-11-14 21:58:33 +0000203// the members of the archive file. This is typically used in preparation for
204// editing the contents of the archive.
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000205bool
206Archive::loadArchive(std::string* error) {
Reid Spencerf9d7a512004-11-14 21:58:33 +0000207
208 // Set up parsing
209 members.clear();
210 symTab.clear();
211 const char *At = base;
Chris Lattner7f6b4472008-04-01 04:26:46 +0000212 const char *End = mapfile->getBufferEnd();
Reid Spencerf9d7a512004-11-14 21:58:33 +0000213
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000214 if (!checkSignature(error))
215 return false;
216
Reid Spencerf9d7a512004-11-14 21:58:33 +0000217 At += 8; // Skip the magic string.
218
Reid Spencerf9d7a512004-11-14 21:58:33 +0000219 bool foundFirstFile = false;
220 while (At < End) {
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000221 // parse the member header
Reid Spencerf9d7a512004-11-14 21:58:33 +0000222 const char* Save = At;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000223 ArchiveMember* mbr = parseMemberHeader(At, End, error);
224 if (!mbr)
225 return false;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000226
227 // check if this is the foreign symbol table
Reid Spencer9a29db42004-11-20 07:29:40 +0000228 if (mbr->isSVR4SymbolTable() || mbr->isBSD4SymbolTable()) {
Reid Spencer4a980d12004-11-16 06:47:19 +0000229 // We just save this but don't do anything special
230 // with it. It doesn't count as the "first file".
Reid Spencer9a29db42004-11-20 07:29:40 +0000231 if (foreignST) {
232 // What? Multiple foreign symbol tables? Just chuck it
233 // and retain the last one found.
234 delete foreignST;
235 }
Reid Spencer4a980d12004-11-16 06:47:19 +0000236 foreignST = mbr;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000237 At += mbr->getSize();
Reid Spencer6405c9e2004-11-19 17:08:00 +0000238 if ((intptr_t(At) & 1) == 1)
Reid Spencerf9d7a512004-11-14 21:58:33 +0000239 At++;
240 } else if (mbr->isStringTable()) {
Reid Spencer4a980d12004-11-16 06:47:19 +0000241 // Simply suck the entire string table into a string
242 // variable. This will be used to get the names of the
243 // members that use the "/ddd" format for their names
244 // (SVR4 style long names).
Reid Spencer9a29db42004-11-20 07:29:40 +0000245 strtab.assign(At, mbr->getSize());
Reid Spencerf9d7a512004-11-14 21:58:33 +0000246 At += mbr->getSize();
Reid Spencer6405c9e2004-11-19 17:08:00 +0000247 if ((intptr_t(At) & 1) == 1)
Reid Spencerf9d7a512004-11-14 21:58:33 +0000248 At++;
249 delete mbr;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000250 } else {
Reid Spencer4a980d12004-11-16 06:47:19 +0000251 // This is just a regular file. If its the first one, save its offset.
252 // Otherwise just push it on the list and move on to the next file.
Reid Spencerf9d7a512004-11-14 21:58:33 +0000253 if (!foundFirstFile) {
254 firstFileOffset = Save - base;
255 foundFirstFile = true;
256 }
257 members.push_back(mbr);
258 At += mbr->getSize();
Reid Spencer6405c9e2004-11-19 17:08:00 +0000259 if ((intptr_t(At) & 1) == 1)
Reid Spencerf9d7a512004-11-14 21:58:33 +0000260 At++;
261 }
262 }
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000263 return true;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000264}
265
266// Open and completely load the archive file.
267Archive*
Matt Arsenault7375f4f2013-02-26 21:20:35 +0000268Archive::OpenAndLoad(const sys::Path& File, LLVMContext& C,
Owen Anderson8b477ed2009-07-01 16:58:40 +0000269 std::string* ErrorMessage) {
Matt Arsenault7375f4f2013-02-26 21:20:35 +0000270 OwningPtr<Archive> result ( new Archive(File, C));
Reid Spencer0ff2d312006-08-24 23:45:08 +0000271 if (result->mapToMemory(ErrorMessage))
Matt Arsenault7375f4f2013-02-26 21:20:35 +0000272 return NULL;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000273 if (!result->loadArchive(ErrorMessage))
Matt Arsenault7375f4f2013-02-26 21:20:35 +0000274 return NULL;
275 return result.take();
Reid Spencerf9d7a512004-11-14 21:58:33 +0000276}
277
Gabor Greifa99be512007-07-05 17:07:56 +0000278// Get all the bitcode modules from the archive
Reid Spencerf9d7a512004-11-14 21:58:33 +0000279bool
Owen Anderson8b477ed2009-07-01 16:58:40 +0000280Archive::getAllModules(std::vector<Module*>& Modules,
281 std::string* ErrMessage) {
Reid Spencerf9d7a512004-11-14 21:58:33 +0000282
283 for (iterator I=begin(), E=end(); I != E; ++I) {
Gabor Greife75ca3d2007-07-06 13:38:17 +0000284 if (I->isBitcode()) {
Chris Lattner74382b72009-08-23 22:45:37 +0000285 std::string FullMemberName = archPath.str() +
286 "(" + I->getPath().str() + ")";
Chris Lattnerc1d56242007-05-06 09:28:33 +0000287 MemoryBuffer *Buffer =
Benjamin Kramer9d44e702010-04-19 16:15:31 +0000288 MemoryBuffer::getMemBufferCopy(StringRef(I->getData(), I->getSize()),
289 FullMemberName.c_str());
Chris Lattnere07c15c2007-05-06 06:18:07 +0000290
Owen Anderson8b477ed2009-07-01 16:58:40 +0000291 Module *M = ParseBitcodeFile(Buffer, Context, ErrMessage);
Chris Lattnerc1d56242007-05-06 09:28:33 +0000292 delete Buffer;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000293 if (!M)
294 return true;
295
296 Modules.push_back(M);
297 }
298 }
Brian Gaeke2c61d7b2003-11-16 23:08:48 +0000299 return false;
300}
Chris Lattner968cfd02003-04-19 21:45:34 +0000301
Reid Spencerf9d7a512004-11-14 21:58:33 +0000302// Load just the symbol table from the archive file
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000303bool
304Archive::loadSymbolTable(std::string* ErrorMsg) {
Chris Lattner968cfd02003-04-19 21:45:34 +0000305
Reid Spencerf9d7a512004-11-14 21:58:33 +0000306 // Set up parsing
307 members.clear();
308 symTab.clear();
309 const char *At = base;
Chris Lattner7f6b4472008-04-01 04:26:46 +0000310 const char *End = mapfile->getBufferEnd();
Chris Lattner968cfd02003-04-19 21:45:34 +0000311
Reid Spencerf9d7a512004-11-14 21:58:33 +0000312 // Make sure we're dealing with an archive
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000313 if (!checkSignature(ErrorMsg))
314 return false;
Chris Lattner968cfd02003-04-19 21:45:34 +0000315
Reid Spencerf9d7a512004-11-14 21:58:33 +0000316 At += 8; // Skip signature
317
318 // Parse the first file member header
319 const char* FirstFile = At;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000320 ArchiveMember* mbr = parseMemberHeader(At, End, ErrorMsg);
321 if (!mbr)
322 return false;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000323
Reid Spencer9a29db42004-11-20 07:29:40 +0000324 if (mbr->isSVR4SymbolTable() || mbr->isBSD4SymbolTable()) {
Reid Spencerf9d7a512004-11-14 21:58:33 +0000325 // Skip the foreign symbol table, we don't do anything with it
326 At += mbr->getSize();
Reid Spencer6405c9e2004-11-19 17:08:00 +0000327 if ((intptr_t(At) & 1) == 1)
Reid Spencerb3231132004-11-15 01:40:20 +0000328 At++;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000329 delete mbr;
330
Reid Spencerb3231132004-11-15 01:40:20 +0000331 // Read the next one
Reid Spencerf9d7a512004-11-14 21:58:33 +0000332 FirstFile = At;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000333 mbr = parseMemberHeader(At, End, ErrorMsg);
334 if (!mbr) {
335 delete mbr;
336 return false;
337 }
Reid Spencerb3231132004-11-15 01:40:20 +0000338 }
339
340 if (mbr->isStringTable()) {
341 // Process the string table entry
Reid Spencer9a29db42004-11-20 07:29:40 +0000342 strtab.assign((const char*)mbr->getData(), mbr->getSize());
Reid Spencerb3231132004-11-15 01:40:20 +0000343 At += mbr->getSize();
Reid Spencer6405c9e2004-11-19 17:08:00 +0000344 if ((intptr_t(At) & 1) == 1)
Reid Spencerb3231132004-11-15 01:40:20 +0000345 At++;
346 delete mbr;
347 // Get the next one
348 FirstFile = At;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000349 mbr = parseMemberHeader(At, End, ErrorMsg);
350 if (!mbr) {
351 delete mbr;
352 return false;
353 }
Chris Lattner968cfd02003-04-19 21:45:34 +0000354 }
355
Rafael Espindola250bfb12013-06-14 23:25:53 +0000356 // There's no symbol table in the file. We have to rebuild it from scratch
357 // because the intent of this method is to get the symbol table loaded so
358 // it can be searched efficiently.
359 // Add the member to the members list
360 members.push_back(mbr);
Reid Spencerf9d7a512004-11-14 21:58:33 +0000361
362 firstFileOffset = FirstFile - base;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000363 return true;
Chris Lattner968cfd02003-04-19 21:45:34 +0000364}
365
Reid Spencerf9d7a512004-11-14 21:58:33 +0000366// Open the archive and load just the symbol tables
Matt Arsenault7375f4f2013-02-26 21:20:35 +0000367Archive* Archive::OpenAndLoadSymbols(const sys::Path& File,
Owen Anderson4434ed42009-07-01 23:13:44 +0000368 LLVMContext& C,
Owen Anderson8b477ed2009-07-01 16:58:40 +0000369 std::string* ErrorMessage) {
Matt Arsenault7375f4f2013-02-26 21:20:35 +0000370 OwningPtr<Archive> result ( new Archive(File, C) );
Reid Spencer0ff2d312006-08-24 23:45:08 +0000371 if (result->mapToMemory(ErrorMessage))
Matt Arsenault7375f4f2013-02-26 21:20:35 +0000372 return NULL;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000373 if (!result->loadSymbolTable(ErrorMessage))
Matt Arsenault7375f4f2013-02-26 21:20:35 +0000374 return NULL;
375 return result.take();
Reid Spencerf9d7a512004-11-14 21:58:33 +0000376}
Chris Lattner968cfd02003-04-19 21:45:34 +0000377
Jeffrey Yasskinf0356fe2010-01-27 20:34:15 +0000378// Look up one symbol in the symbol table and return the module that defines
379// that symbol.
380Module*
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000381Archive::findModuleDefiningSymbol(const std::string& symbol,
382 std::string* ErrMsg) {
Reid Spencerf9d7a512004-11-14 21:58:33 +0000383 SymTabType::iterator SI = symTab.find(symbol);
384 if (SI == symTab.end())
385 return 0;
386
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000387 // The symbol table was previously constructed assuming that the members were
Reid Spencerf9d7a512004-11-14 21:58:33 +0000388 // written without the symbol table header. Because VBR encoding is used, the
389 // values could not be adjusted to account for the offset of the symbol table
390 // because that could affect the size of the symbol table due to VBR encoding.
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000391 // We now have to account for this by adjusting the offset by the size of the
Reid Spencerf9d7a512004-11-14 21:58:33 +0000392 // symbol table and its header.
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000393 unsigned fileOffset =
Reid Spencerf9d7a512004-11-14 21:58:33 +0000394 SI->second + // offset in symbol-table-less file
395 firstFileOffset; // add offset to first "real" file in archive
396
397 // See if the module is already loaded
398 ModuleMap::iterator MI = modules.find(fileOffset);
399 if (MI != modules.end())
400 return MI->second.first;
401
402 // Module hasn't been loaded yet, we need to load it
403 const char* modptr = base + fileOffset;
Chris Lattner7f6b4472008-04-01 04:26:46 +0000404 ArchiveMember* mbr = parseMemberHeader(modptr, mapfile->getBufferEnd(),
405 ErrMsg);
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000406 if (!mbr)
Reid Spencer0b5a5042006-08-25 17:43:11 +0000407 return 0;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000408
Jeffrey Yasskinf0356fe2010-01-27 20:34:15 +0000409 // Now, load the bitcode module to get the Module.
Chris Lattner74382b72009-08-23 22:45:37 +0000410 std::string FullMemberName = archPath.str() + "(" +
411 mbr->getPath().str() + ")";
Benjamin Kramer9d44e702010-04-19 16:15:31 +0000412 MemoryBuffer *Buffer =
413 MemoryBuffer::getMemBufferCopy(StringRef(mbr->getData(), mbr->getSize()),
414 FullMemberName.c_str());
Chris Lattnerc1d56242007-05-06 09:28:33 +0000415
Jeffrey Yasskinf0356fe2010-01-27 20:34:15 +0000416 Module *m = getLazyBitcodeModule(Buffer, Context, ErrMsg);
417 if (!m)
Reid Spencer0b5a5042006-08-25 17:43:11 +0000418 return 0;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000419
Jeffrey Yasskinf0356fe2010-01-27 20:34:15 +0000420 modules.insert(std::make_pair(fileOffset, std::make_pair(m, mbr)));
Reid Spencerf9d7a512004-11-14 21:58:33 +0000421
Jeffrey Yasskinf0356fe2010-01-27 20:34:15 +0000422 return m;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000423}
424
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000425// Look up multiple symbols in the symbol table and return a set of
Jeffrey Yasskinf0356fe2010-01-27 20:34:15 +0000426// Modules that define those symbols.
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000427bool
Reid Spencer7783e8a2004-11-19 03:18:22 +0000428Archive::findModulesDefiningSymbols(std::set<std::string>& symbols,
Rafael Espindola603d6b52012-01-23 03:41:53 +0000429 SmallVectorImpl<Module*>& result,
Chris Lattnerf2e292c2007-02-07 21:41:02 +0000430 std::string* error) {
Reid Spencercd5561a2006-12-15 19:44:51 +0000431 if (!mapfile || !base) {
432 if (error)
433 *error = "Empty archive invalid for finding modules defining symbols";
434 return false;
435 }
436
Reid Spencer766b7932004-11-15 01:20:11 +0000437 if (symTab.empty()) {
438 // We don't have a symbol table, so we must build it now but lets also
439 // make sure that we populate the modules table as we do this to ensure
440 // that we don't load them twice when findModuleDefiningSymbol is called
441 // below.
442
443 // Get a pointer to the first file
Chris Lattner7f6b4472008-04-01 04:26:46 +0000444 const char* At = base + firstFileOffset;
445 const char* End = mapfile->getBufferEnd();
Reid Spencer766b7932004-11-15 01:20:11 +0000446
447 while ( At < End) {
448 // Compute the offset to be put in the symbol table
449 unsigned offset = At - base - firstFileOffset;
450
451 // Parse the file's header
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000452 ArchiveMember* mbr = parseMemberHeader(At, End, error);
453 if (!mbr)
454 return false;
Reid Spencer766b7932004-11-15 01:20:11 +0000455
456 // If it contains symbols
Gabor Greife75ca3d2007-07-06 13:38:17 +0000457 if (mbr->isBitcode()) {
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000458 // Get the symbols
Reid Spencer766b7932004-11-15 01:20:11 +0000459 std::vector<std::string> symbols;
Chris Lattner74382b72009-08-23 22:45:37 +0000460 std::string FullMemberName = archPath.str() + "(" +
461 mbr->getPath().str() + ")";
Jeffrey Yasskinf0356fe2010-01-27 20:34:15 +0000462 Module* M =
Benjamin Kramer9d44e702010-04-19 16:15:31 +0000463 GetBitcodeSymbols(At, mbr->getSize(), FullMemberName, Context,
464 symbols, error);
Reid Spencer766b7932004-11-15 01:20:11 +0000465
Jeffrey Yasskinf0356fe2010-01-27 20:34:15 +0000466 if (M) {
Reid Spencer766b7932004-11-15 01:20:11 +0000467 // Insert the module's symbols into the symbol table
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000468 for (std::vector<std::string>::iterator I = symbols.begin(),
Reid Spencer766b7932004-11-15 01:20:11 +0000469 E=symbols.end(); I != E; ++I ) {
Reid Spencer9a29db42004-11-20 07:29:40 +0000470 symTab.insert(std::make_pair(*I, offset));
Reid Spencer766b7932004-11-15 01:20:11 +0000471 }
Jeffrey Yasskinf0356fe2010-01-27 20:34:15 +0000472 // Insert the Module and the ArchiveMember into the table of
Reid Spencer766b7932004-11-15 01:20:11 +0000473 // modules.
Jeffrey Yasskinf0356fe2010-01-27 20:34:15 +0000474 modules.insert(std::make_pair(offset, std::make_pair(M, mbr)));
Reid Spencer766b7932004-11-15 01:20:11 +0000475 } else {
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000476 if (error)
Gabor Greifa99be512007-07-05 17:07:56 +0000477 *error = "Can't parse bitcode member: " +
Chris Lattner74382b72009-08-23 22:45:37 +0000478 mbr->getPath().str() + ": " + *error;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000479 delete mbr;
480 return false;
Reid Spencer766b7932004-11-15 01:20:11 +0000481 }
482 }
Reid Spencerb3231132004-11-15 01:40:20 +0000483
484 // Go to the next file location
485 At += mbr->getSize();
Reid Spencer6405c9e2004-11-19 17:08:00 +0000486 if ((intptr_t(At) & 1) == 1)
Reid Spencerb3231132004-11-15 01:40:20 +0000487 At++;
Reid Spencer766b7932004-11-15 01:20:11 +0000488 }
489 }
490
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000491 // At this point we have a valid symbol table (one way or another) so we
Reid Spencer766b7932004-11-15 01:20:11 +0000492 // just use it to quickly find the symbols requested.
493
Rafael Espindola603d6b52012-01-23 03:41:53 +0000494 SmallPtrSet<Module*, 16> Added;
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000495 for (std::set<std::string>::iterator I=symbols.begin(),
Rafael Espindola4254df32012-01-23 05:07:16 +0000496 Next = I,
497 E=symbols.end(); I != E; I = Next) {
498 // Increment Next before we invalidate it.
499 ++Next;
500
Reid Spencer7783e8a2004-11-19 03:18:22 +0000501 // See if this symbol exists
Jeffrey Yasskinf0356fe2010-01-27 20:34:15 +0000502 Module* m = findModuleDefiningSymbol(*I,error);
Rafael Espindola603d6b52012-01-23 03:41:53 +0000503 if (!m)
504 continue;
505 bool NewMember = Added.insert(m);
506 if (!NewMember)
507 continue;
Reid Spencer7783e8a2004-11-19 03:18:22 +0000508
Rafael Espindola603d6b52012-01-23 03:41:53 +0000509 // The symbol exists, insert the Module into our result.
510 result.push_back(m);
511
512 // Remove the symbol now that its been resolved.
513 symbols.erase(I);
Reid Spencerf9d7a512004-11-14 21:58:33 +0000514 }
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000515 return true;
Chris Lattner968cfd02003-04-19 21:45:34 +0000516}
Reid Spencereaa06bb2005-02-26 22:00:32 +0000517
Gabor Greifa99be512007-07-05 17:07:56 +0000518bool Archive::isBitcodeArchive() {
Chris Lattner7c439922005-09-23 06:22:58 +0000519 // Make sure the symTab has been loaded. In most cases this should have been
520 // done when the archive was constructed, but still, this is just in case.
Dan Gohman30359592008-01-29 13:02:09 +0000521 if (symTab.empty())
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000522 if (!loadSymbolTable(0))
523 return false;
Reid Spencereaa06bb2005-02-26 22:00:32 +0000524
Chris Lattner7c439922005-09-23 06:22:58 +0000525 // Now that we know it's been loaded, return true
526 // if it has a size
527 if (symTab.size()) return true;
Reid Spencereaa06bb2005-02-26 22:00:32 +0000528
Gabor Greifa99be512007-07-05 17:07:56 +0000529 // We still can't be sure it isn't a bitcode archive
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000530 if (!loadArchive(0))
531 return false;
Reid Spencereaa06bb2005-02-26 22:00:32 +0000532
533 std::vector<Module *> Modules;
534 std::string ErrorMessage;
535
Gabor Greifa99be512007-07-05 17:07:56 +0000536 // Scan the archive, trying to load a bitcode member. We only load one to
Chris Lattner7c439922005-09-23 06:22:58 +0000537 // see if this works.
538 for (iterator I = begin(), E = end(); I != E; ++I) {
Gabor Greife75ca3d2007-07-06 13:38:17 +0000539 if (!I->isBitcode())
Chris Lattner7c439922005-09-23 06:22:58 +0000540 continue;
541
542 std::string FullMemberName =
Chris Lattner74382b72009-08-23 22:45:37 +0000543 archPath.str() + "(" + I->getPath().str() + ")";
Chris Lattnerc1d56242007-05-06 09:28:33 +0000544
545 MemoryBuffer *Buffer =
Benjamin Kramer9d44e702010-04-19 16:15:31 +0000546 MemoryBuffer::getMemBufferCopy(StringRef(I->getData(), I->getSize()),
547 FullMemberName.c_str());
Owen Anderson8b477ed2009-07-01 16:58:40 +0000548 Module *M = ParseBitcodeFile(Buffer, Context);
Chris Lattnerc1d56242007-05-06 09:28:33 +0000549 delete Buffer;
Chris Lattner7c439922005-09-23 06:22:58 +0000550 if (!M)
Gabor Greifa99be512007-07-05 17:07:56 +0000551 return false; // Couldn't parse bitcode, not a bitcode archive.
Chris Lattner7c439922005-09-23 06:22:58 +0000552 delete M;
553 return true;
554 }
555
556 return false;
Reid Spencereaa06bb2005-02-26 22:00:32 +0000557}