blob: 7b363a11dc12f763a3e163da596b7f116ac66e5a [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;
Rafael Espindola4d07abb2013-06-19 15:45:37 +0000177 member->path = pathname;
Rafael Espindolab4900b12013-06-19 21:13:59 +0000178 member->Size = MemberSize;
179 member->ModTime.fromEpochTime(atoi(Hdr->date));
Reid Spencer56127622004-12-29 01:20:24 +0000180 unsigned int mode;
181 sscanf(Hdr->mode, "%o", &mode);
Rafael Espindolab4900b12013-06-19 21:13:59 +0000182 member->Mode = mode;
183 member->User = atoi(Hdr->uid);
184 member->Group = atoi(Hdr->gid);
Reid Spencerf9d7a512004-11-14 21:58:33 +0000185 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*
Rafael Espindola13f4fd72013-06-19 17:49:07 +0000268Archive::OpenAndLoad(StringRef File, LLVMContext& C,
Owen Anderson8b477ed2009-07-01 16:58:40 +0000269 std::string* ErrorMessage) {
Rafael Espindola13f4fd72013-06-19 17:49:07 +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()) {
Rafael Espindola13f4fd72013-06-19 17:49:07 +0000285 std::string FullMemberName = archPath + "(" + I->getPath().str() + ")";
Chris Lattnerc1d56242007-05-06 09:28:33 +0000286 MemoryBuffer *Buffer =
Benjamin Kramer9d44e702010-04-19 16:15:31 +0000287 MemoryBuffer::getMemBufferCopy(StringRef(I->getData(), I->getSize()),
288 FullMemberName.c_str());
Chris Lattnere07c15c2007-05-06 06:18:07 +0000289
Owen Anderson8b477ed2009-07-01 16:58:40 +0000290 Module *M = ParseBitcodeFile(Buffer, Context, ErrMessage);
Chris Lattnerc1d56242007-05-06 09:28:33 +0000291 delete Buffer;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000292 if (!M)
293 return true;
294
295 Modules.push_back(M);
296 }
297 }
Brian Gaeke2c61d7b2003-11-16 23:08:48 +0000298 return false;
299}
Chris Lattner968cfd02003-04-19 21:45:34 +0000300
Reid Spencerf9d7a512004-11-14 21:58:33 +0000301// Load just the symbol table from the archive file
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000302bool
303Archive::loadSymbolTable(std::string* ErrorMsg) {
Chris Lattner968cfd02003-04-19 21:45:34 +0000304
Reid Spencerf9d7a512004-11-14 21:58:33 +0000305 // Set up parsing
306 members.clear();
307 symTab.clear();
308 const char *At = base;
Chris Lattner7f6b4472008-04-01 04:26:46 +0000309 const char *End = mapfile->getBufferEnd();
Chris Lattner968cfd02003-04-19 21:45:34 +0000310
Reid Spencerf9d7a512004-11-14 21:58:33 +0000311 // Make sure we're dealing with an archive
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000312 if (!checkSignature(ErrorMsg))
313 return false;
Chris Lattner968cfd02003-04-19 21:45:34 +0000314
Reid Spencerf9d7a512004-11-14 21:58:33 +0000315 At += 8; // Skip signature
316
317 // Parse the first file member header
318 const char* FirstFile = At;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000319 ArchiveMember* mbr = parseMemberHeader(At, End, ErrorMsg);
320 if (!mbr)
321 return false;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000322
Reid Spencer9a29db42004-11-20 07:29:40 +0000323 if (mbr->isSVR4SymbolTable() || mbr->isBSD4SymbolTable()) {
Reid Spencerf9d7a512004-11-14 21:58:33 +0000324 // Skip the foreign symbol table, we don't do anything with it
325 At += mbr->getSize();
Reid Spencer6405c9e2004-11-19 17:08:00 +0000326 if ((intptr_t(At) & 1) == 1)
Reid Spencerb3231132004-11-15 01:40:20 +0000327 At++;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000328 delete mbr;
329
Reid Spencerb3231132004-11-15 01:40:20 +0000330 // Read the next one
Reid Spencerf9d7a512004-11-14 21:58:33 +0000331 FirstFile = At;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000332 mbr = parseMemberHeader(At, End, ErrorMsg);
333 if (!mbr) {
334 delete mbr;
335 return false;
336 }
Reid Spencerb3231132004-11-15 01:40:20 +0000337 }
338
339 if (mbr->isStringTable()) {
340 // Process the string table entry
Reid Spencer9a29db42004-11-20 07:29:40 +0000341 strtab.assign((const char*)mbr->getData(), mbr->getSize());
Reid Spencerb3231132004-11-15 01:40:20 +0000342 At += mbr->getSize();
Reid Spencer6405c9e2004-11-19 17:08:00 +0000343 if ((intptr_t(At) & 1) == 1)
Reid Spencerb3231132004-11-15 01:40:20 +0000344 At++;
345 delete mbr;
346 // Get the next one
347 FirstFile = At;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000348 mbr = parseMemberHeader(At, End, ErrorMsg);
349 if (!mbr) {
350 delete mbr;
351 return false;
352 }
Chris Lattner968cfd02003-04-19 21:45:34 +0000353 }
354
Rafael Espindola250bfb12013-06-14 23:25:53 +0000355 // There's no symbol table in the file. We have to rebuild it from scratch
356 // because the intent of this method is to get the symbol table loaded so
357 // it can be searched efficiently.
358 // Add the member to the members list
359 members.push_back(mbr);
Reid Spencerf9d7a512004-11-14 21:58:33 +0000360
361 firstFileOffset = FirstFile - base;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000362 return true;
Chris Lattner968cfd02003-04-19 21:45:34 +0000363}
364
Jeffrey Yasskinf0356fe2010-01-27 20:34:15 +0000365// Look up one symbol in the symbol table and return the module that defines
366// that symbol.
367Module*
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000368Archive::findModuleDefiningSymbol(const std::string& symbol,
369 std::string* ErrMsg) {
Reid Spencerf9d7a512004-11-14 21:58:33 +0000370 SymTabType::iterator SI = symTab.find(symbol);
371 if (SI == symTab.end())
372 return 0;
373
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000374 // The symbol table was previously constructed assuming that the members were
Reid Spencerf9d7a512004-11-14 21:58:33 +0000375 // written without the symbol table header. Because VBR encoding is used, the
376 // values could not be adjusted to account for the offset of the symbol table
377 // because that could affect the size of the symbol table due to VBR encoding.
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000378 // We now have to account for this by adjusting the offset by the size of the
Reid Spencerf9d7a512004-11-14 21:58:33 +0000379 // symbol table and its header.
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000380 unsigned fileOffset =
Reid Spencerf9d7a512004-11-14 21:58:33 +0000381 SI->second + // offset in symbol-table-less file
382 firstFileOffset; // add offset to first "real" file in archive
383
384 // See if the module is already loaded
385 ModuleMap::iterator MI = modules.find(fileOffset);
386 if (MI != modules.end())
387 return MI->second.first;
388
389 // Module hasn't been loaded yet, we need to load it
390 const char* modptr = base + fileOffset;
Chris Lattner7f6b4472008-04-01 04:26:46 +0000391 ArchiveMember* mbr = parseMemberHeader(modptr, mapfile->getBufferEnd(),
392 ErrMsg);
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000393 if (!mbr)
Reid Spencer0b5a5042006-08-25 17:43:11 +0000394 return 0;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000395
Jeffrey Yasskinf0356fe2010-01-27 20:34:15 +0000396 // Now, load the bitcode module to get the Module.
Rafael Espindola13f4fd72013-06-19 17:49:07 +0000397 std::string FullMemberName = archPath + "(" + mbr->getPath().str() + ")";
Benjamin Kramer9d44e702010-04-19 16:15:31 +0000398 MemoryBuffer *Buffer =
399 MemoryBuffer::getMemBufferCopy(StringRef(mbr->getData(), mbr->getSize()),
400 FullMemberName.c_str());
Chris Lattnerc1d56242007-05-06 09:28:33 +0000401
Jeffrey Yasskinf0356fe2010-01-27 20:34:15 +0000402 Module *m = getLazyBitcodeModule(Buffer, Context, ErrMsg);
403 if (!m)
Reid Spencer0b5a5042006-08-25 17:43:11 +0000404 return 0;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000405
Jeffrey Yasskinf0356fe2010-01-27 20:34:15 +0000406 modules.insert(std::make_pair(fileOffset, std::make_pair(m, mbr)));
Reid Spencerf9d7a512004-11-14 21:58:33 +0000407
Jeffrey Yasskinf0356fe2010-01-27 20:34:15 +0000408 return m;
Reid Spencerf9d7a512004-11-14 21:58:33 +0000409}
410
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000411// Look up multiple symbols in the symbol table and return a set of
Jeffrey Yasskinf0356fe2010-01-27 20:34:15 +0000412// Modules that define those symbols.
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000413bool
Reid Spencer7783e8a2004-11-19 03:18:22 +0000414Archive::findModulesDefiningSymbols(std::set<std::string>& symbols,
Rafael Espindola603d6b52012-01-23 03:41:53 +0000415 SmallVectorImpl<Module*>& result,
Chris Lattnerf2e292c2007-02-07 21:41:02 +0000416 std::string* error) {
Reid Spencercd5561a2006-12-15 19:44:51 +0000417 if (!mapfile || !base) {
418 if (error)
419 *error = "Empty archive invalid for finding modules defining symbols";
420 return false;
421 }
422
Reid Spencer766b7932004-11-15 01:20:11 +0000423 if (symTab.empty()) {
424 // We don't have a symbol table, so we must build it now but lets also
425 // make sure that we populate the modules table as we do this to ensure
426 // that we don't load them twice when findModuleDefiningSymbol is called
427 // below.
428
429 // Get a pointer to the first file
Chris Lattner7f6b4472008-04-01 04:26:46 +0000430 const char* At = base + firstFileOffset;
431 const char* End = mapfile->getBufferEnd();
Reid Spencer766b7932004-11-15 01:20:11 +0000432
433 while ( At < End) {
434 // Compute the offset to be put in the symbol table
435 unsigned offset = At - base - firstFileOffset;
436
437 // Parse the file's header
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000438 ArchiveMember* mbr = parseMemberHeader(At, End, error);
439 if (!mbr)
440 return false;
Reid Spencer766b7932004-11-15 01:20:11 +0000441
442 // If it contains symbols
Gabor Greife75ca3d2007-07-06 13:38:17 +0000443 if (mbr->isBitcode()) {
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000444 // Get the symbols
Reid Spencer766b7932004-11-15 01:20:11 +0000445 std::vector<std::string> symbols;
Rafael Espindola13f4fd72013-06-19 17:49:07 +0000446 std::string FullMemberName =
447 archPath + "(" + mbr->getPath().str() + ")";
Jeffrey Yasskinf0356fe2010-01-27 20:34:15 +0000448 Module* M =
Benjamin Kramer9d44e702010-04-19 16:15:31 +0000449 GetBitcodeSymbols(At, mbr->getSize(), FullMemberName, Context,
450 symbols, error);
Reid Spencer766b7932004-11-15 01:20:11 +0000451
Jeffrey Yasskinf0356fe2010-01-27 20:34:15 +0000452 if (M) {
Reid Spencer766b7932004-11-15 01:20:11 +0000453 // Insert the module's symbols into the symbol table
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000454 for (std::vector<std::string>::iterator I = symbols.begin(),
Reid Spencer766b7932004-11-15 01:20:11 +0000455 E=symbols.end(); I != E; ++I ) {
Reid Spencer9a29db42004-11-20 07:29:40 +0000456 symTab.insert(std::make_pair(*I, offset));
Reid Spencer766b7932004-11-15 01:20:11 +0000457 }
Jeffrey Yasskinf0356fe2010-01-27 20:34:15 +0000458 // Insert the Module and the ArchiveMember into the table of
Reid Spencer766b7932004-11-15 01:20:11 +0000459 // modules.
Jeffrey Yasskinf0356fe2010-01-27 20:34:15 +0000460 modules.insert(std::make_pair(offset, std::make_pair(M, mbr)));
Reid Spencer766b7932004-11-15 01:20:11 +0000461 } else {
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000462 if (error)
Gabor Greifa99be512007-07-05 17:07:56 +0000463 *error = "Can't parse bitcode member: " +
Chris Lattner74382b72009-08-23 22:45:37 +0000464 mbr->getPath().str() + ": " + *error;
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000465 delete mbr;
466 return false;
Reid Spencer766b7932004-11-15 01:20:11 +0000467 }
468 }
Reid Spencerb3231132004-11-15 01:40:20 +0000469
470 // Go to the next file location
471 At += mbr->getSize();
Reid Spencer6405c9e2004-11-19 17:08:00 +0000472 if ((intptr_t(At) & 1) == 1)
Reid Spencerb3231132004-11-15 01:40:20 +0000473 At++;
Reid Spencer766b7932004-11-15 01:20:11 +0000474 }
475 }
476
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000477 // At this point we have a valid symbol table (one way or another) so we
Reid Spencer766b7932004-11-15 01:20:11 +0000478 // just use it to quickly find the symbols requested.
479
Rafael Espindola603d6b52012-01-23 03:41:53 +0000480 SmallPtrSet<Module*, 16> Added;
Misha Brukman2b37d7c2005-04-21 21:13:18 +0000481 for (std::set<std::string>::iterator I=symbols.begin(),
Rafael Espindola4254df32012-01-23 05:07:16 +0000482 Next = I,
483 E=symbols.end(); I != E; I = Next) {
484 // Increment Next before we invalidate it.
485 ++Next;
486
Reid Spencer7783e8a2004-11-19 03:18:22 +0000487 // See if this symbol exists
Jeffrey Yasskinf0356fe2010-01-27 20:34:15 +0000488 Module* m = findModuleDefiningSymbol(*I,error);
Rafael Espindola603d6b52012-01-23 03:41:53 +0000489 if (!m)
490 continue;
491 bool NewMember = Added.insert(m);
492 if (!NewMember)
493 continue;
Reid Spencer7783e8a2004-11-19 03:18:22 +0000494
Rafael Espindola603d6b52012-01-23 03:41:53 +0000495 // The symbol exists, insert the Module into our result.
496 result.push_back(m);
497
498 // Remove the symbol now that its been resolved.
499 symbols.erase(I);
Reid Spencerf9d7a512004-11-14 21:58:33 +0000500 }
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000501 return true;
Chris Lattner968cfd02003-04-19 21:45:34 +0000502}
Reid Spencereaa06bb2005-02-26 22:00:32 +0000503
Gabor Greifa99be512007-07-05 17:07:56 +0000504bool Archive::isBitcodeArchive() {
Chris Lattner7c439922005-09-23 06:22:58 +0000505 // Make sure the symTab has been loaded. In most cases this should have been
506 // done when the archive was constructed, but still, this is just in case.
Dan Gohman30359592008-01-29 13:02:09 +0000507 if (symTab.empty())
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000508 if (!loadSymbolTable(0))
509 return false;
Reid Spencereaa06bb2005-02-26 22:00:32 +0000510
Chris Lattner7c439922005-09-23 06:22:58 +0000511 // Now that we know it's been loaded, return true
512 // if it has a size
513 if (symTab.size()) return true;
Reid Spencereaa06bb2005-02-26 22:00:32 +0000514
Gabor Greifa99be512007-07-05 17:07:56 +0000515 // We still can't be sure it isn't a bitcode archive
Reid Spencer8d8a7ff2006-07-07 20:56:50 +0000516 if (!loadArchive(0))
517 return false;
Reid Spencereaa06bb2005-02-26 22:00:32 +0000518
519 std::vector<Module *> Modules;
520 std::string ErrorMessage;
521
Gabor Greifa99be512007-07-05 17:07:56 +0000522 // Scan the archive, trying to load a bitcode member. We only load one to
Chris Lattner7c439922005-09-23 06:22:58 +0000523 // see if this works.
524 for (iterator I = begin(), E = end(); I != E; ++I) {
Gabor Greife75ca3d2007-07-06 13:38:17 +0000525 if (!I->isBitcode())
Chris Lattner7c439922005-09-23 06:22:58 +0000526 continue;
Rafael Espindola13f4fd72013-06-19 17:49:07 +0000527
528 std::string FullMemberName = archPath + "(" + I->getPath().str() + ")";
Chris Lattnerc1d56242007-05-06 09:28:33 +0000529
530 MemoryBuffer *Buffer =
Benjamin Kramer9d44e702010-04-19 16:15:31 +0000531 MemoryBuffer::getMemBufferCopy(StringRef(I->getData(), I->getSize()),
532 FullMemberName.c_str());
Owen Anderson8b477ed2009-07-01 16:58:40 +0000533 Module *M = ParseBitcodeFile(Buffer, Context);
Chris Lattnerc1d56242007-05-06 09:28:33 +0000534 delete Buffer;
Chris Lattner7c439922005-09-23 06:22:58 +0000535 if (!M)
Gabor Greifa99be512007-07-05 17:07:56 +0000536 return false; // Couldn't parse bitcode, not a bitcode archive.
Chris Lattner7c439922005-09-23 06:22:58 +0000537 delete M;
538 return true;
539 }
540
541 return false;
Reid Spencereaa06bb2005-02-26 22:00:32 +0000542}