blob: b00f76539b3ce99160464d7d8182aadc719667b4 [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"
35#include "llvm/Transforms/IPO.h"
36#include "llvm/Transforms/Scalar.h"
37#include "llvm/Analysis/LoadValueNumbering.h"
38#include "llvm/LinkTimeOptimizer.h"
39#include <fstream>
40#include <iostream>
41
42using namespace llvm;
43
44extern "C"
45llvm::LinkTimeOptimizer *createLLVMOptimizer()
46{
Devang Patelc7cfbc52006-09-21 17:22:55 +000047 llvm::LTO *l = new llvm::LTO();
Devang Patela89d47f2006-08-03 15:44:57 +000048 return l;
49}
50
51
52
53/// If symbol is not used then make it internal and let optimizer takes
54/// care of it.
55void LLVMSymbol::mayBeNotUsed() {
56 gv->setLinkage(GlobalValue::InternalLinkage);
57}
58
59// Helper routine
60// FIXME : Take advantage of GlobalPrefix from AsmPrinter
61static const char *addUnderscore(const char *name) {
62 size_t namelen = strlen(name);
63 char *symName = (char*)malloc(namelen+2);
64 symName[0] = '_';
65 strcpy(&symName[1], name);
66 return symName;
67}
68
69// Map LLVM LinkageType to LTO LinakgeType
70static LTOLinkageTypes
71getLTOLinkageType(GlobalValue *v)
72{
73 LTOLinkageTypes lt;
74 if (v->hasExternalLinkage())
75 lt = LTOExternalLinkage;
76 else if (v->hasLinkOnceLinkage())
77 lt = LTOLinkOnceLinkage;
78 else if (v->hasWeakLinkage())
79 lt = LTOWeakLinkage;
80 else
81 // Otherwise it is internal linkage for link time optimizer
82 lt = LTOInternalLinkage;
83 return lt;
84}
85
86// Find exeternal symbols referenced by VALUE. This is a recursive function.
87static void
Devang Patel30235da2006-08-14 22:36:16 +000088findExternalRefs(Value *value, std::set<std::string> &references,
Devang Patel2198f9c2006-08-14 23:37:18 +000089 Mangler &mangler) {
Devang Patel304d5f22006-08-04 19:10:26 +000090
91 if (GlobalValue *gv = dyn_cast<GlobalValue>(value)) {
Devang Patela89d47f2006-08-03 15:44:57 +000092 LTOLinkageTypes lt = getLTOLinkageType(gv);
93 if (lt != LTOInternalLinkage && strncmp (gv->getName().c_str(), "llvm.", 5))
Devang Patel30235da2006-08-14 22:36:16 +000094 references.insert(mangler.getValueName(gv));
Devang Patela89d47f2006-08-03 15:44:57 +000095 }
Devang Patel544ea342006-09-14 05:49:10 +000096
97 // GlobalValue, even with InternalLinkage type, may have operands with
98 // ExternalLinkage type. Do not ignore these operands.
Devang Patel97d92d52006-09-14 01:35:13 +000099 if (Constant *c = dyn_cast<Constant>(value))
Devang Patel304d5f22006-08-04 19:10:26 +0000100 // Handle ConstantExpr, ConstantStruct, ConstantArry etc..
101 for (unsigned i = 0, e = c->getNumOperands(); i != e; ++i)
Devang Patel30235da2006-08-14 22:36:16 +0000102 findExternalRefs(c->getOperand(i), references, mangler);
Devang Patela89d47f2006-08-03 15:44:57 +0000103}
104
Devang Patel2a4dd682006-10-23 23:57:53 +0000105/// If Module with InputFilename is available then remove it from allModules
106/// and call delete on it.
Devang Patelf2ca21f2006-10-23 23:12:26 +0000107void
108LTO::removeModule (const std::string &InputFilename)
109{
110 NameToModuleMap::iterator pos = allModules.find(InputFilename.c_str());
Devang Patel2a4dd682006-10-23 23:57:53 +0000111 if (pos == allModules.end())
112 return;
113
114 Module *m = pos->second;
115 allModules.erase(pos);
116 delete m;
Devang Patelf2ca21f2006-10-23 23:12:26 +0000117}
118
Devang Patel0701a2f2006-09-06 18:50:26 +0000119/// InputFilename is a LLVM bytecode file. If Module with InputFilename is
120/// available then return it. Otherwise parseInputFilename.
121Module *
Devang Patelc7cfbc52006-09-21 17:22:55 +0000122LTO::getModule(const std::string &InputFilename)
Devang Patel0701a2f2006-09-06 18:50:26 +0000123{
124 Module *m = NULL;
125
126 NameToModuleMap::iterator pos = allModules.find(InputFilename.c_str());
127 if (pos != allModules.end())
128 m = allModules[InputFilename.c_str()];
129 else {
130 m = ParseBytecodeFile(InputFilename);
131 allModules[InputFilename.c_str()] = m;
132 }
133 return m;
134}
135
Devang Patela291a682006-09-06 20:16:28 +0000136/// InputFilename is a LLVM bytecode file. Reade this bytecode file and
137/// set corresponding target triplet string.
138void
Devang Patelc7cfbc52006-09-21 17:22:55 +0000139LTO::getTargetTriple(const std::string &InputFilename,
Devang Patela291a682006-09-06 20:16:28 +0000140 std::string &targetTriple)
141{
142 Module *m = getModule(InputFilename);
143 if (m)
144 targetTriple = m->getTargetTriple();
145}
146
Devang Patela89d47f2006-08-03 15:44:57 +0000147/// InputFilename is a LLVM bytecode file. Read it using bytecode reader.
148/// Collect global functions and symbol names in symbols vector.
149/// Collect external references in references vector.
150/// Return LTO_READ_SUCCESS if there is no error.
151enum LTOStatus
Devang Patelc7cfbc52006-09-21 17:22:55 +0000152LTO::readLLVMObjectFile(const std::string &InputFilename,
Devang Patel2198f9c2006-08-14 23:37:18 +0000153 NameToSymbolMap &symbols,
154 std::set<std::string> &references)
Devang Patela89d47f2006-08-03 15:44:57 +0000155{
Devang Patel0701a2f2006-09-06 18:50:26 +0000156 Module *m = getModule(InputFilename);
Devang Patela89d47f2006-08-03 15:44:57 +0000157 if (!m)
158 return LTO_READ_FAILURE;
Devang Patel30235da2006-08-14 22:36:16 +0000159
160 // Use mangler to add GlobalPrefix to names to match linker names.
161 // FIXME : Instead of hard coding "-" use GlobalPrefix.
162 Mangler mangler(*m, "_");
Devang Patela89d47f2006-08-03 15:44:57 +0000163
164 modules.push_back(m);
165
166 for (Module::iterator f = m->begin(), e = m->end(); f != e; ++f) {
167
168 LTOLinkageTypes lt = getLTOLinkageType(f);
169
170 if (!f->isExternal() && lt != LTOInternalLinkage
Devang Patel2198f9c2006-08-14 23:37:18 +0000171 && strncmp (f->getName().c_str(), "llvm.", 5)) {
Devang Patel30235da2006-08-14 22:36:16 +0000172 LLVMSymbol *newSymbol = new LLVMSymbol(lt, f, f->getName(),
Devang Patel2198f9c2006-08-14 23:37:18 +0000173 mangler.getValueName(f));
Devang Patel30235da2006-08-14 22:36:16 +0000174 symbols[newSymbol->getMangledName()] = newSymbol;
175 allSymbols[newSymbol->getMangledName()] = newSymbol;
Devang Patela89d47f2006-08-03 15:44:57 +0000176 }
Devang Patel30235da2006-08-14 22:36:16 +0000177
Devang Patela89d47f2006-08-03 15:44:57 +0000178 // Collect external symbols referenced by this function.
179 for (Function::iterator b = f->begin(), fe = f->end(); b != fe; ++b)
180 for (BasicBlock::iterator i = b->begin(), be = b->end();
Devang Patel2198f9c2006-08-14 23:37:18 +0000181 i != be; ++i)
182 for (unsigned count = 0, total = i->getNumOperands();
183 count != total; ++count)
184 findExternalRefs(i->getOperand(count), references, mangler);
Devang Patela89d47f2006-08-03 15:44:57 +0000185 }
186
187 for (Module::global_iterator v = m->global_begin(), e = m->global_end();
188 v != e; ++v) {
189 LTOLinkageTypes lt = getLTOLinkageType(v);
190 if (!v->isExternal() && lt != LTOInternalLinkage
Devang Patel2198f9c2006-08-14 23:37:18 +0000191 && strncmp (v->getName().c_str(), "llvm.", 5)) {
Devang Patel30235da2006-08-14 22:36:16 +0000192 LLVMSymbol *newSymbol = new LLVMSymbol(lt, v, v->getName(),
Devang Patel2198f9c2006-08-14 23:37:18 +0000193 mangler.getValueName(v));
Devang Patel30235da2006-08-14 22:36:16 +0000194 symbols[newSymbol->getMangledName()] = newSymbol;
Devang Pateled872862006-09-06 00:45:52 +0000195 allSymbols[newSymbol->getMangledName()] = newSymbol;
Devang Patel304d5f22006-08-04 19:10:26 +0000196
197 for (unsigned count = 0, total = v->getNumOperands();
Devang Patel2198f9c2006-08-14 23:37:18 +0000198 count != total; ++count)
199 findExternalRefs(v->getOperand(count), references, mangler);
Devang Patel304d5f22006-08-04 19:10:26 +0000200
Devang Patela89d47f2006-08-03 15:44:57 +0000201 }
202 }
203
204 return LTO_READ_SUCCESS;
205}
206
207/// Optimize module M using various IPO passes. Use exportList to
208/// internalize selected symbols. Target platform is selected
209/// based on information available to module M. No new target
210/// features are selected.
211static enum LTOStatus lto_optimize(Module *M, std::ostream &Out,
Devang Patel2198f9c2006-08-14 23:37:18 +0000212 std::vector<const char *> &exportList)
Devang Patela89d47f2006-08-03 15:44:57 +0000213{
214 // Instantiate the pass manager to organize the passes.
215 PassManager Passes;
216
217 // Collect Target info
218 std::string Err;
219 const TargetMachineRegistry::Entry* March =
220 TargetMachineRegistry::getClosestStaticTargetForModule(*M, Err);
221
222 if (March == 0)
223 return LTO_NO_TARGET;
224
225 // Create target
226 std::string Features;
227 std::auto_ptr<TargetMachine> target(March->CtorFn(*M, Features));
228 if (!target.get())
229 return LTO_NO_TARGET;
230
231 TargetMachine &Target = *target.get();
232
233 // Start off with a verification pass.
234 Passes.add(createVerifierPass());
235
236 // Add an appropriate TargetData instance for this module...
237 Passes.add(new TargetData(*Target.getTargetData()));
238
239 // Often if the programmer does not specify proper prototypes for the
240 // functions they are calling, they end up calling a vararg version of the
241 // function that does not get a body filled in (the real function has typed
242 // arguments). This pass merges the two functions.
243 Passes.add(createFunctionResolvingPass());
244
245 // Internalize symbols if export list is nonemty
246 if (!exportList.empty())
247 Passes.add(createInternalizePass(exportList));
248
249 // Now that we internalized some globals, see if we can hack on them!
250 Passes.add(createGlobalOptimizerPass());
251
252 // Linking modules together can lead to duplicated global constants, only
253 // keep one copy of each constant...
254 Passes.add(createConstantMergePass());
255
256 // If the -s command line option was specified, strip the symbols out of the
257 // resulting program to make it smaller. -s is a GLD option that we are
258 // supporting.
259 Passes.add(createStripSymbolsPass());
260
261 // Propagate constants at call sites into the functions they call.
262 Passes.add(createIPConstantPropagationPass());
263
264 // Remove unused arguments from functions...
265 Passes.add(createDeadArgEliminationPass());
266
267 Passes.add(createFunctionInliningPass()); // Inline small functions
268
269 Passes.add(createPruneEHPass()); // Remove dead EH info
270
271 Passes.add(createGlobalDCEPass()); // Remove dead functions
272
273 // If we didn't decide to inline a function, check to see if we can
274 // transform it to pass arguments by value instead of by reference.
275 Passes.add(createArgumentPromotionPass());
276
277 // The IPO passes may leave cruft around. Clean up after them.
278 Passes.add(createInstructionCombiningPass());
279
280 Passes.add(createScalarReplAggregatesPass()); // Break up allocas
281
282 // Run a few AA driven optimizations here and now, to cleanup the code.
283 Passes.add(createGlobalsModRefPass()); // IP alias analysis
284
285 Passes.add(createLICMPass()); // Hoist loop invariants
286 Passes.add(createLoadValueNumberingPass()); // GVN for load instrs
287 Passes.add(createGCSEPass()); // Remove common subexprs
288 Passes.add(createDeadStoreEliminationPass()); // Nuke dead stores
289
290 // Cleanup and simplify the code after the scalar optimizations.
291 Passes.add(createInstructionCombiningPass());
292
293 // Delete basic blocks, which optimization passes may have killed...
294 Passes.add(createCFGSimplificationPass());
295
296 // Now that we have optimized the program, discard unreachable functions...
297 Passes.add(createGlobalDCEPass());
298
299 // Make sure everything is still good.
300 Passes.add(createVerifierPass());
301
Devang Patel998051a2006-09-07 21:41:11 +0000302 FunctionPassManager *CodeGenPasses =
303 new FunctionPassManager(new ExistingModuleProvider(M));
304
305 CodeGenPasses->add(new TargetData(*Target.getTargetData()));
306 Target.addPassesToEmitFile(*CodeGenPasses, Out, TargetMachine::AssemblyFile,
307 true);
Devang Patela89d47f2006-08-03 15:44:57 +0000308
309 // Run our queue of passes all at once now, efficiently.
310 Passes.run(*M);
311
Devang Patel998051a2006-09-07 21:41:11 +0000312 // Run the code generator, if present.
313 CodeGenPasses->doInitialization();
314 for (Module::iterator I = M->begin(), E = M->end(); I != E; ++I) {
315 if (!I->isExternal())
316 CodeGenPasses->run(*I);
317 }
318 CodeGenPasses->doFinalization();
319
Devang Patela89d47f2006-08-03 15:44:57 +0000320 return LTO_OPT_SUCCESS;
321}
322
323///Link all modules together and optimize them using IPO. Generate
324/// native object file using OutputFilename
325/// Return appropriate LTOStatus.
326enum LTOStatus
Devang Patelc7cfbc52006-09-21 17:22:55 +0000327LTO::optimizeModules(const std::string &OutputFilename,
Devang Patel26810232006-09-06 00:28:22 +0000328 std::vector<const char *> &exportList,
329 std::string &targetTriple)
Devang Patela89d47f2006-08-03 15:44:57 +0000330{
331 if (modules.empty())
332 return LTO_NO_WORK;
333
334 std::ios::openmode io_mode =
335 std::ios::out | std::ios::trunc | std::ios::binary;
336 std::string *errMsg = NULL;
337 Module *bigOne = modules[0];
338 Linker theLinker("LinkTimeOptimizer", bigOne, false);
339 for (unsigned i = 1, e = modules.size(); i != e; ++i)
340 if (theLinker.LinkModules(bigOne, modules[i], errMsg))
341 return LTO_MODULE_MERGE_FAILURE;
342
343#if 0
344 // Enable this when -save-temps is used
345 std::ofstream Out("big.bc", io_mode);
346 WriteBytecodeToFile(bigOne, Out, true);
347#endif
348
349 // Strip leading underscore because it was added to match names
Devang Patel94a0ac92006-08-03 17:25:36 +0000350 // seen by linker.
Devang Patela89d47f2006-08-03 15:44:57 +0000351 for (unsigned i = 0, e = exportList.size(); i != e; ++i) {
352 const char *name = exportList[i];
Devang Pateled872862006-09-06 00:45:52 +0000353 NameToSymbolMap::iterator itr = allSymbols.find(name);
354 if (itr != allSymbols.end())
355 exportList[i] = allSymbols[name]->getName();
Devang Patela89d47f2006-08-03 15:44:57 +0000356 }
357
Devang Patel3f0e5e22006-10-09 19:04:51 +0000358
Reid Spencere5c9cb52006-08-23 00:39:35 +0000359 std::string ErrMsg;
Devang Patel3f0e5e22006-10-09 19:04:51 +0000360 sys::Path TempDir = sys::Path::GetTemporaryDirectory(&ErrMsg);
Devang Patel9f5d48b2006-10-09 20:20:13 +0000361 if (TempDir.isEmpty()) {
362 std::cerr << "lto: " << ErrMsg << "\n";
363 return LTO_WRITE_FAILURE;
364 }
Devang Patel3f0e5e22006-10-09 19:04:51 +0000365 sys::Path tmpAsmFilePath(TempDir);
366 if (!tmpAsmFilePath.appendComponent("lto")) {
367 std::cerr << "lto: " << ErrMsg << "\n";
368 TempDir.eraseFromDisk(true);
369 return LTO_WRITE_FAILURE;
370 }
Reid Spencere5c9cb52006-08-23 00:39:35 +0000371 if (tmpAsmFilePath.createTemporaryFileOnDisk(&ErrMsg)) {
372 std::cerr << "lto: " << ErrMsg << "\n";
Devang Patel3f0e5e22006-10-09 19:04:51 +0000373 TempDir.eraseFromDisk(true);
Devang Patelca640122006-08-23 16:59:25 +0000374 return LTO_WRITE_FAILURE;
Reid Spencere5c9cb52006-08-23 00:39:35 +0000375 }
Devang Patela89d47f2006-08-03 15:44:57 +0000376 sys::RemoveFileOnSignal(tmpAsmFilePath);
377
378 std::ofstream asmFile(tmpAsmFilePath.c_str(), io_mode);
379 if (!asmFile.is_open() || asmFile.bad()) {
Devang Patel3f0e5e22006-10-09 19:04:51 +0000380 if (tmpAsmFilePath.exists()) {
Devang Patela89d47f2006-08-03 15:44:57 +0000381 tmpAsmFilePath.eraseFromDisk();
Devang Patel3f0e5e22006-10-09 19:04:51 +0000382 TempDir.eraseFromDisk(true);
383 }
Devang Patela89d47f2006-08-03 15:44:57 +0000384 return LTO_WRITE_FAILURE;
385 }
386
387 enum LTOStatus status = lto_optimize(bigOne, asmFile, exportList);
388 asmFile.close();
389 if (status != LTO_OPT_SUCCESS) {
390 tmpAsmFilePath.eraseFromDisk();
Devang Patel3f0e5e22006-10-09 19:04:51 +0000391 TempDir.eraseFromDisk(true);
Devang Patela89d47f2006-08-03 15:44:57 +0000392 return status;
393 }
394
Devang Patel26810232006-09-06 00:28:22 +0000395 targetTriple = bigOne->getTargetTriple();
396
Devang Patela89d47f2006-08-03 15:44:57 +0000397 // Run GCC to assemble and link the program into native code.
398 //
399 // Note:
400 // We can't just assemble and link the file with the system assembler
401 // and linker because we don't know where to put the _start symbol.
402 // GCC mysteriously knows how to do it.
Devang Pateldc4c3822006-10-09 21:16:05 +0000403 const sys::Path gcc = sys::Program::FindProgramByName("gcc");
Devang Patela89d47f2006-08-03 15:44:57 +0000404 if (gcc.isEmpty()) {
405 tmpAsmFilePath.eraseFromDisk();
Devang Patel3f0e5e22006-10-09 19:04:51 +0000406 TempDir.eraseFromDisk(true);
Devang Patela89d47f2006-08-03 15:44:57 +0000407 return LTO_ASM_FAILURE;
408 }
409
410 std::vector<const char*> args;
411 args.push_back(gcc.c_str());
412 args.push_back("-c");
413 args.push_back("-x");
414 args.push_back("assembler");
415 args.push_back("-o");
416 args.push_back(OutputFilename.c_str());
417 args.push_back(tmpAsmFilePath.c_str());
418 args.push_back(0);
419
Devang Patel9f5d48b2006-10-09 20:20:13 +0000420 if (sys::Program::ExecuteAndWait(gcc, &args[0], 0, 0, 1, &ErrMsg)) {
421 std::cerr << "lto: " << ErrMsg << "\n";
422 return LTO_ASM_FAILURE;
423 }
Devang Patela89d47f2006-08-03 15:44:57 +0000424
425 tmpAsmFilePath.eraseFromDisk();
Devang Patel3f0e5e22006-10-09 19:04:51 +0000426 TempDir.eraseFromDisk(true);
Devang Patela89d47f2006-08-03 15:44:57 +0000427
428 return LTO_OPT_SUCCESS;
429}