blob: 5c507708b4a021837f543e5bc17fa4ed613a6527 [file] [log] [blame]
Devang Patela89d47f2006-08-03 15:44:57 +00001//===-lto.cpp - LLVM Link Time Optimizer ----------------------------------===//
2//
3// The LLVM Compiler Infrastructure
4//
5// This file was developed by Devang Patel and is distributed under
6// the University of Illinois Open Source License. See LICENSE.TXT for details.
7//
8//===----------------------------------------------------------------------===//
9//
10// This file implementes link time optimization library. This library is
11// intended to be used by linker to optimize code at link time.
12//
13//===----------------------------------------------------------------------===//
14
15#include "llvm/Module.h"
16#include "llvm/PassManager.h"
17#include "llvm/Linker.h"
18#include "llvm/Constants.h"
19#include "llvm/DerivedTypes.h"
20#include "llvm/SymbolTable.h"
21#include "llvm/Bytecode/Reader.h"
22#include "llvm/Bytecode/Writer.h"
23#include "llvm/Support/CommandLine.h"
24#include "llvm/Support/FileUtilities.h"
25#include "llvm/Support/SystemUtils.h"
Devang Patel30235da2006-08-14 22:36:16 +000026#include "llvm/Support/Mangler.h"
Devang Patela89d47f2006-08-03 15:44:57 +000027#include "llvm/System/Program.h"
28#include "llvm/System/Signals.h"
29#include "llvm/Analysis/Passes.h"
30#include "llvm/Analysis/Verifier.h"
31#include "llvm/Target/SubtargetFeature.h"
32#include "llvm/Target/TargetData.h"
33#include "llvm/Target/TargetMachine.h"
34#include "llvm/Target/TargetMachineRegistry.h"
Devang Patel6152b7e2006-10-24 18:41:02 +000035#include "llvm/Target/TargetAsmInfo.h"
Devang Patela89d47f2006-08-03 15:44:57 +000036#include "llvm/Transforms/IPO.h"
37#include "llvm/Transforms/Scalar.h"
38#include "llvm/Analysis/LoadValueNumbering.h"
39#include "llvm/LinkTimeOptimizer.h"
40#include <fstream>
41#include <iostream>
42
43using namespace llvm;
44
45extern "C"
46llvm::LinkTimeOptimizer *createLLVMOptimizer()
47{
Devang Patelc7cfbc52006-09-21 17:22:55 +000048 llvm::LTO *l = new llvm::LTO();
Devang Patela89d47f2006-08-03 15:44:57 +000049 return l;
50}
51
52
53
54/// If symbol is not used then make it internal and let optimizer takes
55/// care of it.
56void LLVMSymbol::mayBeNotUsed() {
57 gv->setLinkage(GlobalValue::InternalLinkage);
58}
59
Devang Patela89d47f2006-08-03 15:44:57 +000060// Map LLVM LinkageType to LTO LinakgeType
61static LTOLinkageTypes
62getLTOLinkageType(GlobalValue *v)
63{
64 LTOLinkageTypes lt;
65 if (v->hasExternalLinkage())
66 lt = LTOExternalLinkage;
67 else if (v->hasLinkOnceLinkage())
68 lt = LTOLinkOnceLinkage;
69 else if (v->hasWeakLinkage())
70 lt = LTOWeakLinkage;
71 else
72 // Otherwise it is internal linkage for link time optimizer
73 lt = LTOInternalLinkage;
74 return lt;
75}
76
77// Find exeternal symbols referenced by VALUE. This is a recursive function.
78static void
Devang Patel30235da2006-08-14 22:36:16 +000079findExternalRefs(Value *value, std::set<std::string> &references,
Devang Patel2198f9c2006-08-14 23:37:18 +000080 Mangler &mangler) {
Devang Patel304d5f22006-08-04 19:10:26 +000081
82 if (GlobalValue *gv = dyn_cast<GlobalValue>(value)) {
Devang Patela89d47f2006-08-03 15:44:57 +000083 LTOLinkageTypes lt = getLTOLinkageType(gv);
84 if (lt != LTOInternalLinkage && strncmp (gv->getName().c_str(), "llvm.", 5))
Devang Patel30235da2006-08-14 22:36:16 +000085 references.insert(mangler.getValueName(gv));
Devang Patela89d47f2006-08-03 15:44:57 +000086 }
Devang Patel544ea342006-09-14 05:49:10 +000087
88 // GlobalValue, even with InternalLinkage type, may have operands with
89 // ExternalLinkage type. Do not ignore these operands.
Devang Patel97d92d52006-09-14 01:35:13 +000090 if (Constant *c = dyn_cast<Constant>(value))
Devang Patel304d5f22006-08-04 19:10:26 +000091 // Handle ConstantExpr, ConstantStruct, ConstantArry etc..
92 for (unsigned i = 0, e = c->getNumOperands(); i != e; ++i)
Devang Patel30235da2006-08-14 22:36:16 +000093 findExternalRefs(c->getOperand(i), references, mangler);
Devang Patela89d47f2006-08-03 15:44:57 +000094}
95
Devang Patel2a4dd682006-10-23 23:57:53 +000096/// If Module with InputFilename is available then remove it from allModules
97/// and call delete on it.
Devang Patelf2ca21f2006-10-23 23:12:26 +000098void
99LTO::removeModule (const std::string &InputFilename)
100{
101 NameToModuleMap::iterator pos = allModules.find(InputFilename.c_str());
Devang Patel2a4dd682006-10-23 23:57:53 +0000102 if (pos == allModules.end())
103 return;
104
105 Module *m = pos->second;
106 allModules.erase(pos);
107 delete m;
Devang Patelf2ca21f2006-10-23 23:12:26 +0000108}
109
Devang Patel0701a2f2006-09-06 18:50:26 +0000110/// InputFilename is a LLVM bytecode file. If Module with InputFilename is
111/// available then return it. Otherwise parseInputFilename.
112Module *
Devang Patelc7cfbc52006-09-21 17:22:55 +0000113LTO::getModule(const std::string &InputFilename)
Devang Patel0701a2f2006-09-06 18:50:26 +0000114{
115 Module *m = NULL;
116
117 NameToModuleMap::iterator pos = allModules.find(InputFilename.c_str());
118 if (pos != allModules.end())
119 m = allModules[InputFilename.c_str()];
120 else {
121 m = ParseBytecodeFile(InputFilename);
122 allModules[InputFilename.c_str()] = m;
123 }
124 return m;
125}
126
Devang Patela291a682006-09-06 20:16:28 +0000127/// InputFilename is a LLVM bytecode file. Reade this bytecode file and
128/// set corresponding target triplet string.
129void
Devang Patelc7cfbc52006-09-21 17:22:55 +0000130LTO::getTargetTriple(const std::string &InputFilename,
Devang Patela291a682006-09-06 20:16:28 +0000131 std::string &targetTriple)
132{
133 Module *m = getModule(InputFilename);
134 if (m)
135 targetTriple = m->getTargetTriple();
136}
137
Devang Patela89d47f2006-08-03 15:44:57 +0000138/// InputFilename is a LLVM bytecode file. Read it using bytecode reader.
139/// Collect global functions and symbol names in symbols vector.
140/// Collect external references in references vector.
141/// Return LTO_READ_SUCCESS if there is no error.
142enum LTOStatus
Devang Patelc7cfbc52006-09-21 17:22:55 +0000143LTO::readLLVMObjectFile(const std::string &InputFilename,
Devang Patel2198f9c2006-08-14 23:37:18 +0000144 NameToSymbolMap &symbols,
145 std::set<std::string> &references)
Devang Patela89d47f2006-08-03 15:44:57 +0000146{
Devang Patel0701a2f2006-09-06 18:50:26 +0000147 Module *m = getModule(InputFilename);
Devang Patela89d47f2006-08-03 15:44:57 +0000148 if (!m)
149 return LTO_READ_FAILURE;
Devang Patel30235da2006-08-14 22:36:16 +0000150
Devang Patel6152b7e2006-10-24 18:41:02 +0000151 // Collect Target info
152 if (!Target)
153 getTarget(m);
154
155 if (!Target)
156 return LTO_READ_FAILURE;
157
Devang Patel30235da2006-08-14 22:36:16 +0000158 // Use mangler to add GlobalPrefix to names to match linker names.
159 // FIXME : Instead of hard coding "-" use GlobalPrefix.
Devang Patel6152b7e2006-10-24 18:41:02 +0000160 Mangler mangler(*m, Target->getTargetAsmInfo()->getGlobalPrefix());
Devang Patela89d47f2006-08-03 15:44:57 +0000161 modules.push_back(m);
162
163 for (Module::iterator f = m->begin(), e = m->end(); f != e; ++f) {
164
165 LTOLinkageTypes lt = getLTOLinkageType(f);
166
167 if (!f->isExternal() && lt != LTOInternalLinkage
Devang Patel2198f9c2006-08-14 23:37:18 +0000168 && strncmp (f->getName().c_str(), "llvm.", 5)) {
Devang Patel30235da2006-08-14 22:36:16 +0000169 LLVMSymbol *newSymbol = new LLVMSymbol(lt, f, f->getName(),
Devang Patel2198f9c2006-08-14 23:37:18 +0000170 mangler.getValueName(f));
Devang Patel30235da2006-08-14 22:36:16 +0000171 symbols[newSymbol->getMangledName()] = newSymbol;
172 allSymbols[newSymbol->getMangledName()] = newSymbol;
Devang Patela89d47f2006-08-03 15:44:57 +0000173 }
Devang Patel30235da2006-08-14 22:36:16 +0000174
Devang Patela89d47f2006-08-03 15:44:57 +0000175 // Collect external symbols referenced by this function.
176 for (Function::iterator b = f->begin(), fe = f->end(); b != fe; ++b)
177 for (BasicBlock::iterator i = b->begin(), be = b->end();
Devang Patel2198f9c2006-08-14 23:37:18 +0000178 i != be; ++i)
179 for (unsigned count = 0, total = i->getNumOperands();
180 count != total; ++count)
181 findExternalRefs(i->getOperand(count), references, mangler);
Devang Patela89d47f2006-08-03 15:44:57 +0000182 }
183
184 for (Module::global_iterator v = m->global_begin(), e = m->global_end();
185 v != e; ++v) {
186 LTOLinkageTypes lt = getLTOLinkageType(v);
187 if (!v->isExternal() && lt != LTOInternalLinkage
Devang Patel2198f9c2006-08-14 23:37:18 +0000188 && strncmp (v->getName().c_str(), "llvm.", 5)) {
Devang Patel30235da2006-08-14 22:36:16 +0000189 LLVMSymbol *newSymbol = new LLVMSymbol(lt, v, v->getName(),
Devang Patel2198f9c2006-08-14 23:37:18 +0000190 mangler.getValueName(v));
Devang Patel30235da2006-08-14 22:36:16 +0000191 symbols[newSymbol->getMangledName()] = newSymbol;
Devang Pateled872862006-09-06 00:45:52 +0000192 allSymbols[newSymbol->getMangledName()] = newSymbol;
Devang Patel304d5f22006-08-04 19:10:26 +0000193
194 for (unsigned count = 0, total = v->getNumOperands();
Devang Patel2198f9c2006-08-14 23:37:18 +0000195 count != total; ++count)
196 findExternalRefs(v->getOperand(count), references, mangler);
Devang Patel304d5f22006-08-04 19:10:26 +0000197
Devang Patela89d47f2006-08-03 15:44:57 +0000198 }
199 }
200
201 return LTO_READ_SUCCESS;
202}
203
Devang Patel6152b7e2006-10-24 18:41:02 +0000204/// Get TargetMachine.
205/// Use module M to find appropriate Target.
206void
207LTO::getTarget (Module *M) {
208
Devang Patela89d47f2006-08-03 15:44:57 +0000209 std::string Err;
210 const TargetMachineRegistry::Entry* March =
211 TargetMachineRegistry::getClosestStaticTargetForModule(*M, Err);
212
213 if (March == 0)
Devang Patel6152b7e2006-10-24 18:41:02 +0000214 return;
Devang Patela89d47f2006-08-03 15:44:57 +0000215
216 // Create target
217 std::string Features;
Devang Patel6152b7e2006-10-24 18:41:02 +0000218 Target = March->CtorFn(*M, Features);
219}
220
221/// Optimize module M using various IPO passes. Use exportList to
222/// internalize selected symbols. Target platform is selected
223/// based on information available to module M. No new target
224/// features are selected.
225enum LTOStatus
226LTO::optimize(Module *M, std::ostream &Out,
227 std::vector<const char *> &exportList)
228{
229 // Instantiate the pass manager to organize the passes.
230 PassManager Passes;
Devang Patela89d47f2006-08-03 15:44:57 +0000231
Devang Patel6152b7e2006-10-24 18:41:02 +0000232 // Collect Target info
233 if (!Target)
234 getTarget(M);
235
236 if (!Target)
237 return LTO_NO_TARGET;
Devang Patela89d47f2006-08-03 15:44:57 +0000238
239 // Start off with a verification pass.
240 Passes.add(createVerifierPass());
241
242 // Add an appropriate TargetData instance for this module...
Devang Patel6152b7e2006-10-24 18:41:02 +0000243 Passes.add(new TargetData(*Target->getTargetData()));
Devang Patela89d47f2006-08-03 15:44:57 +0000244
245 // Often if the programmer does not specify proper prototypes for the
246 // functions they are calling, they end up calling a vararg version of the
247 // function that does not get a body filled in (the real function has typed
248 // arguments). This pass merges the two functions.
249 Passes.add(createFunctionResolvingPass());
250
251 // Internalize symbols if export list is nonemty
252 if (!exportList.empty())
253 Passes.add(createInternalizePass(exportList));
254
255 // Now that we internalized some globals, see if we can hack on them!
256 Passes.add(createGlobalOptimizerPass());
257
258 // Linking modules together can lead to duplicated global constants, only
259 // keep one copy of each constant...
260 Passes.add(createConstantMergePass());
261
262 // If the -s command line option was specified, strip the symbols out of the
263 // resulting program to make it smaller. -s is a GLD option that we are
264 // supporting.
265 Passes.add(createStripSymbolsPass());
266
267 // Propagate constants at call sites into the functions they call.
268 Passes.add(createIPConstantPropagationPass());
269
270 // Remove unused arguments from functions...
271 Passes.add(createDeadArgEliminationPass());
272
273 Passes.add(createFunctionInliningPass()); // Inline small functions
274
275 Passes.add(createPruneEHPass()); // Remove dead EH info
276
277 Passes.add(createGlobalDCEPass()); // Remove dead functions
278
279 // If we didn't decide to inline a function, check to see if we can
280 // transform it to pass arguments by value instead of by reference.
281 Passes.add(createArgumentPromotionPass());
282
283 // The IPO passes may leave cruft around. Clean up after them.
284 Passes.add(createInstructionCombiningPass());
285
286 Passes.add(createScalarReplAggregatesPass()); // Break up allocas
287
288 // Run a few AA driven optimizations here and now, to cleanup the code.
289 Passes.add(createGlobalsModRefPass()); // IP alias analysis
290
291 Passes.add(createLICMPass()); // Hoist loop invariants
292 Passes.add(createLoadValueNumberingPass()); // GVN for load instrs
293 Passes.add(createGCSEPass()); // Remove common subexprs
294 Passes.add(createDeadStoreEliminationPass()); // Nuke dead stores
295
296 // Cleanup and simplify the code after the scalar optimizations.
297 Passes.add(createInstructionCombiningPass());
298
299 // Delete basic blocks, which optimization passes may have killed...
300 Passes.add(createCFGSimplificationPass());
301
302 // Now that we have optimized the program, discard unreachable functions...
303 Passes.add(createGlobalDCEPass());
304
305 // Make sure everything is still good.
306 Passes.add(createVerifierPass());
307
Devang Patel998051a2006-09-07 21:41:11 +0000308 FunctionPassManager *CodeGenPasses =
309 new FunctionPassManager(new ExistingModuleProvider(M));
310
Devang Patel6152b7e2006-10-24 18:41:02 +0000311 CodeGenPasses->add(new TargetData(*Target->getTargetData()));
312 Target->addPassesToEmitFile(*CodeGenPasses, Out, TargetMachine::AssemblyFile,
Devang Patel998051a2006-09-07 21:41:11 +0000313 true);
Devang Patela89d47f2006-08-03 15:44:57 +0000314
315 // Run our queue of passes all at once now, efficiently.
316 Passes.run(*M);
317
Devang Patel998051a2006-09-07 21:41:11 +0000318 // Run the code generator, if present.
319 CodeGenPasses->doInitialization();
320 for (Module::iterator I = M->begin(), E = M->end(); I != E; ++I) {
321 if (!I->isExternal())
322 CodeGenPasses->run(*I);
323 }
324 CodeGenPasses->doFinalization();
325
Devang Patela89d47f2006-08-03 15:44:57 +0000326 return LTO_OPT_SUCCESS;
327}
328
329///Link all modules together and optimize them using IPO. Generate
330/// native object file using OutputFilename
331/// Return appropriate LTOStatus.
332enum LTOStatus
Devang Patelc7cfbc52006-09-21 17:22:55 +0000333LTO::optimizeModules(const std::string &OutputFilename,
Devang Patel26810232006-09-06 00:28:22 +0000334 std::vector<const char *> &exportList,
335 std::string &targetTriple)
Devang Patela89d47f2006-08-03 15:44:57 +0000336{
337 if (modules.empty())
338 return LTO_NO_WORK;
339
340 std::ios::openmode io_mode =
341 std::ios::out | std::ios::trunc | std::ios::binary;
342 std::string *errMsg = NULL;
343 Module *bigOne = modules[0];
344 Linker theLinker("LinkTimeOptimizer", bigOne, false);
345 for (unsigned i = 1, e = modules.size(); i != e; ++i)
346 if (theLinker.LinkModules(bigOne, modules[i], errMsg))
347 return LTO_MODULE_MERGE_FAILURE;
348
349#if 0
350 // Enable this when -save-temps is used
351 std::ofstream Out("big.bc", io_mode);
352 WriteBytecodeToFile(bigOne, Out, true);
353#endif
354
355 // Strip leading underscore because it was added to match names
Devang Patel94a0ac92006-08-03 17:25:36 +0000356 // seen by linker.
Devang Patela89d47f2006-08-03 15:44:57 +0000357 for (unsigned i = 0, e = exportList.size(); i != e; ++i) {
358 const char *name = exportList[i];
Devang Pateled872862006-09-06 00:45:52 +0000359 NameToSymbolMap::iterator itr = allSymbols.find(name);
360 if (itr != allSymbols.end())
361 exportList[i] = allSymbols[name]->getName();
Devang Patela89d47f2006-08-03 15:44:57 +0000362 }
363
Devang Patel3f0e5e22006-10-09 19:04:51 +0000364
Reid Spencere5c9cb52006-08-23 00:39:35 +0000365 std::string ErrMsg;
Devang Patel3f0e5e22006-10-09 19:04:51 +0000366 sys::Path TempDir = sys::Path::GetTemporaryDirectory(&ErrMsg);
Devang Patel9f5d48b2006-10-09 20:20:13 +0000367 if (TempDir.isEmpty()) {
368 std::cerr << "lto: " << ErrMsg << "\n";
369 return LTO_WRITE_FAILURE;
370 }
Devang Patel3f0e5e22006-10-09 19:04:51 +0000371 sys::Path tmpAsmFilePath(TempDir);
372 if (!tmpAsmFilePath.appendComponent("lto")) {
373 std::cerr << "lto: " << ErrMsg << "\n";
374 TempDir.eraseFromDisk(true);
375 return LTO_WRITE_FAILURE;
376 }
Reid Spencere5c9cb52006-08-23 00:39:35 +0000377 if (tmpAsmFilePath.createTemporaryFileOnDisk(&ErrMsg)) {
378 std::cerr << "lto: " << ErrMsg << "\n";
Devang Patel3f0e5e22006-10-09 19:04:51 +0000379 TempDir.eraseFromDisk(true);
Devang Patelca640122006-08-23 16:59:25 +0000380 return LTO_WRITE_FAILURE;
Reid Spencere5c9cb52006-08-23 00:39:35 +0000381 }
Devang Patela89d47f2006-08-03 15:44:57 +0000382 sys::RemoveFileOnSignal(tmpAsmFilePath);
383
384 std::ofstream asmFile(tmpAsmFilePath.c_str(), io_mode);
385 if (!asmFile.is_open() || asmFile.bad()) {
Devang Patel3f0e5e22006-10-09 19:04:51 +0000386 if (tmpAsmFilePath.exists()) {
Devang Patela89d47f2006-08-03 15:44:57 +0000387 tmpAsmFilePath.eraseFromDisk();
Devang Patel3f0e5e22006-10-09 19:04:51 +0000388 TempDir.eraseFromDisk(true);
389 }
Devang Patela89d47f2006-08-03 15:44:57 +0000390 return LTO_WRITE_FAILURE;
391 }
392
Devang Patel6152b7e2006-10-24 18:41:02 +0000393 enum LTOStatus status = optimize(bigOne, asmFile, exportList);
Devang Patela89d47f2006-08-03 15:44:57 +0000394 asmFile.close();
395 if (status != LTO_OPT_SUCCESS) {
396 tmpAsmFilePath.eraseFromDisk();
Devang Patel3f0e5e22006-10-09 19:04:51 +0000397 TempDir.eraseFromDisk(true);
Devang Patela89d47f2006-08-03 15:44:57 +0000398 return status;
399 }
400
Devang Patel26810232006-09-06 00:28:22 +0000401 targetTriple = bigOne->getTargetTriple();
402
Devang Patela89d47f2006-08-03 15:44:57 +0000403 // Run GCC to assemble and link the program into native code.
404 //
405 // Note:
406 // We can't just assemble and link the file with the system assembler
407 // and linker because we don't know where to put the _start symbol.
408 // GCC mysteriously knows how to do it.
Devang Pateldc4c3822006-10-09 21:16:05 +0000409 const sys::Path gcc = sys::Program::FindProgramByName("gcc");
Devang Patela89d47f2006-08-03 15:44:57 +0000410 if (gcc.isEmpty()) {
411 tmpAsmFilePath.eraseFromDisk();
Devang Patel3f0e5e22006-10-09 19:04:51 +0000412 TempDir.eraseFromDisk(true);
Devang Patela89d47f2006-08-03 15:44:57 +0000413 return LTO_ASM_FAILURE;
414 }
415
416 std::vector<const char*> args;
417 args.push_back(gcc.c_str());
418 args.push_back("-c");
419 args.push_back("-x");
420 args.push_back("assembler");
421 args.push_back("-o");
422 args.push_back(OutputFilename.c_str());
423 args.push_back(tmpAsmFilePath.c_str());
424 args.push_back(0);
425
Devang Patel9f5d48b2006-10-09 20:20:13 +0000426 if (sys::Program::ExecuteAndWait(gcc, &args[0], 0, 0, 1, &ErrMsg)) {
427 std::cerr << "lto: " << ErrMsg << "\n";
428 return LTO_ASM_FAILURE;
429 }
Devang Patela89d47f2006-08-03 15:44:57 +0000430
431 tmpAsmFilePath.eraseFromDisk();
Devang Patel3f0e5e22006-10-09 19:04:51 +0000432 TempDir.eraseFromDisk(true);
Devang Patela89d47f2006-08-03 15:44:57 +0000433
434 return LTO_OPT_SUCCESS;
435}
Devang Patel6152b7e2006-10-24 18:41:02 +0000436
437/// Destruct LTO. Delete all modules, symbols and target.
438LTO::~LTO() {
439
440 for (std::vector<Module *>::iterator itr = modules.begin(), e = modules.end();
441 itr != e; ++itr)
442 delete *itr;
443
444 modules.clear();
445
446 for (NameToSymbolMap::iterator itr = allSymbols.begin(), e = allSymbols.end();
447 itr != e; ++itr)
448 delete itr->second;
449
450 allSymbols.clear();
451
452 delete Target;
453}