blob: cd3cc341c0b0fb5b95beab77037c9931e935722e [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 Patel0701a2f2006-09-06 18:50:26 +0000105/// InputFilename is a LLVM bytecode file. If Module with InputFilename is
106/// available then return it. Otherwise parseInputFilename.
107Module *
Devang Patelc7cfbc52006-09-21 17:22:55 +0000108LTO::getModule(const std::string &InputFilename)
Devang Patel0701a2f2006-09-06 18:50:26 +0000109{
110 Module *m = NULL;
111
112 NameToModuleMap::iterator pos = allModules.find(InputFilename.c_str());
113 if (pos != allModules.end())
114 m = allModules[InputFilename.c_str()];
115 else {
116 m = ParseBytecodeFile(InputFilename);
117 allModules[InputFilename.c_str()] = m;
118 }
119 return m;
120}
121
Devang Patela291a682006-09-06 20:16:28 +0000122/// InputFilename is a LLVM bytecode file. Reade this bytecode file and
123/// set corresponding target triplet string.
124void
Devang Patelc7cfbc52006-09-21 17:22:55 +0000125LTO::getTargetTriple(const std::string &InputFilename,
Devang Patela291a682006-09-06 20:16:28 +0000126 std::string &targetTriple)
127{
128 Module *m = getModule(InputFilename);
129 if (m)
130 targetTriple = m->getTargetTriple();
131}
132
Devang Patela89d47f2006-08-03 15:44:57 +0000133/// InputFilename is a LLVM bytecode file. Read it using bytecode reader.
134/// Collect global functions and symbol names in symbols vector.
135/// Collect external references in references vector.
136/// Return LTO_READ_SUCCESS if there is no error.
137enum LTOStatus
Devang Patelc7cfbc52006-09-21 17:22:55 +0000138LTO::readLLVMObjectFile(const std::string &InputFilename,
Devang Patel2198f9c2006-08-14 23:37:18 +0000139 NameToSymbolMap &symbols,
140 std::set<std::string> &references)
Devang Patela89d47f2006-08-03 15:44:57 +0000141{
Devang Patel0701a2f2006-09-06 18:50:26 +0000142 Module *m = getModule(InputFilename);
Devang Patela89d47f2006-08-03 15:44:57 +0000143 if (!m)
144 return LTO_READ_FAILURE;
Devang Patel30235da2006-08-14 22:36:16 +0000145
146 // Use mangler to add GlobalPrefix to names to match linker names.
147 // FIXME : Instead of hard coding "-" use GlobalPrefix.
148 Mangler mangler(*m, "_");
Devang Patela89d47f2006-08-03 15:44:57 +0000149
150 modules.push_back(m);
151
152 for (Module::iterator f = m->begin(), e = m->end(); f != e; ++f) {
153
154 LTOLinkageTypes lt = getLTOLinkageType(f);
155
156 if (!f->isExternal() && lt != LTOInternalLinkage
Devang Patel2198f9c2006-08-14 23:37:18 +0000157 && strncmp (f->getName().c_str(), "llvm.", 5)) {
Devang Patel30235da2006-08-14 22:36:16 +0000158 LLVMSymbol *newSymbol = new LLVMSymbol(lt, f, f->getName(),
Devang Patel2198f9c2006-08-14 23:37:18 +0000159 mangler.getValueName(f));
Devang Patel30235da2006-08-14 22:36:16 +0000160 symbols[newSymbol->getMangledName()] = newSymbol;
161 allSymbols[newSymbol->getMangledName()] = newSymbol;
Devang Patela89d47f2006-08-03 15:44:57 +0000162 }
Devang Patel30235da2006-08-14 22:36:16 +0000163
Devang Patela89d47f2006-08-03 15:44:57 +0000164 // Collect external symbols referenced by this function.
165 for (Function::iterator b = f->begin(), fe = f->end(); b != fe; ++b)
166 for (BasicBlock::iterator i = b->begin(), be = b->end();
Devang Patel2198f9c2006-08-14 23:37:18 +0000167 i != be; ++i)
168 for (unsigned count = 0, total = i->getNumOperands();
169 count != total; ++count)
170 findExternalRefs(i->getOperand(count), references, mangler);
Devang Patela89d47f2006-08-03 15:44:57 +0000171 }
172
173 for (Module::global_iterator v = m->global_begin(), e = m->global_end();
174 v != e; ++v) {
175 LTOLinkageTypes lt = getLTOLinkageType(v);
176 if (!v->isExternal() && lt != LTOInternalLinkage
Devang Patel2198f9c2006-08-14 23:37:18 +0000177 && strncmp (v->getName().c_str(), "llvm.", 5)) {
Devang Patel30235da2006-08-14 22:36:16 +0000178 LLVMSymbol *newSymbol = new LLVMSymbol(lt, v, v->getName(),
Devang Patel2198f9c2006-08-14 23:37:18 +0000179 mangler.getValueName(v));
Devang Patel30235da2006-08-14 22:36:16 +0000180 symbols[newSymbol->getMangledName()] = newSymbol;
Devang Pateled872862006-09-06 00:45:52 +0000181 allSymbols[newSymbol->getMangledName()] = newSymbol;
Devang Patel304d5f22006-08-04 19:10:26 +0000182
183 for (unsigned count = 0, total = v->getNumOperands();
Devang Patel2198f9c2006-08-14 23:37:18 +0000184 count != total; ++count)
185 findExternalRefs(v->getOperand(count), references, mangler);
Devang Patel304d5f22006-08-04 19:10:26 +0000186
Devang Patela89d47f2006-08-03 15:44:57 +0000187 }
188 }
189
190 return LTO_READ_SUCCESS;
191}
192
193/// Optimize module M using various IPO passes. Use exportList to
194/// internalize selected symbols. Target platform is selected
195/// based on information available to module M. No new target
196/// features are selected.
197static enum LTOStatus lto_optimize(Module *M, std::ostream &Out,
Devang Patel2198f9c2006-08-14 23:37:18 +0000198 std::vector<const char *> &exportList)
Devang Patela89d47f2006-08-03 15:44:57 +0000199{
200 // Instantiate the pass manager to organize the passes.
201 PassManager Passes;
202
203 // Collect Target info
204 std::string Err;
205 const TargetMachineRegistry::Entry* March =
206 TargetMachineRegistry::getClosestStaticTargetForModule(*M, Err);
207
208 if (March == 0)
209 return LTO_NO_TARGET;
210
211 // Create target
212 std::string Features;
213 std::auto_ptr<TargetMachine> target(March->CtorFn(*M, Features));
214 if (!target.get())
215 return LTO_NO_TARGET;
216
217 TargetMachine &Target = *target.get();
218
219 // Start off with a verification pass.
220 Passes.add(createVerifierPass());
221
222 // Add an appropriate TargetData instance for this module...
223 Passes.add(new TargetData(*Target.getTargetData()));
224
225 // Often if the programmer does not specify proper prototypes for the
226 // functions they are calling, they end up calling a vararg version of the
227 // function that does not get a body filled in (the real function has typed
228 // arguments). This pass merges the two functions.
229 Passes.add(createFunctionResolvingPass());
230
231 // Internalize symbols if export list is nonemty
232 if (!exportList.empty())
233 Passes.add(createInternalizePass(exportList));
234
235 // Now that we internalized some globals, see if we can hack on them!
236 Passes.add(createGlobalOptimizerPass());
237
238 // Linking modules together can lead to duplicated global constants, only
239 // keep one copy of each constant...
240 Passes.add(createConstantMergePass());
241
242 // If the -s command line option was specified, strip the symbols out of the
243 // resulting program to make it smaller. -s is a GLD option that we are
244 // supporting.
245 Passes.add(createStripSymbolsPass());
246
247 // Propagate constants at call sites into the functions they call.
248 Passes.add(createIPConstantPropagationPass());
249
250 // Remove unused arguments from functions...
251 Passes.add(createDeadArgEliminationPass());
252
253 Passes.add(createFunctionInliningPass()); // Inline small functions
254
255 Passes.add(createPruneEHPass()); // Remove dead EH info
256
257 Passes.add(createGlobalDCEPass()); // Remove dead functions
258
259 // If we didn't decide to inline a function, check to see if we can
260 // transform it to pass arguments by value instead of by reference.
261 Passes.add(createArgumentPromotionPass());
262
263 // The IPO passes may leave cruft around. Clean up after them.
264 Passes.add(createInstructionCombiningPass());
265
266 Passes.add(createScalarReplAggregatesPass()); // Break up allocas
267
268 // Run a few AA driven optimizations here and now, to cleanup the code.
269 Passes.add(createGlobalsModRefPass()); // IP alias analysis
270
271 Passes.add(createLICMPass()); // Hoist loop invariants
272 Passes.add(createLoadValueNumberingPass()); // GVN for load instrs
273 Passes.add(createGCSEPass()); // Remove common subexprs
274 Passes.add(createDeadStoreEliminationPass()); // Nuke dead stores
275
276 // Cleanup and simplify the code after the scalar optimizations.
277 Passes.add(createInstructionCombiningPass());
278
279 // Delete basic blocks, which optimization passes may have killed...
280 Passes.add(createCFGSimplificationPass());
281
282 // Now that we have optimized the program, discard unreachable functions...
283 Passes.add(createGlobalDCEPass());
284
285 // Make sure everything is still good.
286 Passes.add(createVerifierPass());
287
Devang Patel998051a2006-09-07 21:41:11 +0000288 FunctionPassManager *CodeGenPasses =
289 new FunctionPassManager(new ExistingModuleProvider(M));
290
291 CodeGenPasses->add(new TargetData(*Target.getTargetData()));
292 Target.addPassesToEmitFile(*CodeGenPasses, Out, TargetMachine::AssemblyFile,
293 true);
Devang Patela89d47f2006-08-03 15:44:57 +0000294
295 // Run our queue of passes all at once now, efficiently.
296 Passes.run(*M);
297
Devang Patel998051a2006-09-07 21:41:11 +0000298 // Run the code generator, if present.
299 CodeGenPasses->doInitialization();
300 for (Module::iterator I = M->begin(), E = M->end(); I != E; ++I) {
301 if (!I->isExternal())
302 CodeGenPasses->run(*I);
303 }
304 CodeGenPasses->doFinalization();
305
Devang Patela89d47f2006-08-03 15:44:57 +0000306 return LTO_OPT_SUCCESS;
307}
308
309///Link all modules together and optimize them using IPO. Generate
310/// native object file using OutputFilename
311/// Return appropriate LTOStatus.
312enum LTOStatus
Devang Patelc7cfbc52006-09-21 17:22:55 +0000313LTO::optimizeModules(const std::string &OutputFilename,
Devang Patel26810232006-09-06 00:28:22 +0000314 std::vector<const char *> &exportList,
315 std::string &targetTriple)
Devang Patela89d47f2006-08-03 15:44:57 +0000316{
317 if (modules.empty())
318 return LTO_NO_WORK;
319
320 std::ios::openmode io_mode =
321 std::ios::out | std::ios::trunc | std::ios::binary;
322 std::string *errMsg = NULL;
323 Module *bigOne = modules[0];
324 Linker theLinker("LinkTimeOptimizer", bigOne, false);
325 for (unsigned i = 1, e = modules.size(); i != e; ++i)
326 if (theLinker.LinkModules(bigOne, modules[i], errMsg))
327 return LTO_MODULE_MERGE_FAILURE;
328
329#if 0
330 // Enable this when -save-temps is used
331 std::ofstream Out("big.bc", io_mode);
332 WriteBytecodeToFile(bigOne, Out, true);
333#endif
334
335 // Strip leading underscore because it was added to match names
Devang Patel94a0ac92006-08-03 17:25:36 +0000336 // seen by linker.
Devang Patela89d47f2006-08-03 15:44:57 +0000337 for (unsigned i = 0, e = exportList.size(); i != e; ++i) {
338 const char *name = exportList[i];
Devang Pateled872862006-09-06 00:45:52 +0000339 NameToSymbolMap::iterator itr = allSymbols.find(name);
340 if (itr != allSymbols.end())
341 exportList[i] = allSymbols[name]->getName();
Devang Patela89d47f2006-08-03 15:44:57 +0000342 }
343
344 sys::Path tmpAsmFilePath("/tmp/");
Reid Spencere5c9cb52006-08-23 00:39:35 +0000345 std::string ErrMsg;
346 if (tmpAsmFilePath.createTemporaryFileOnDisk(&ErrMsg)) {
347 std::cerr << "lto: " << ErrMsg << "\n";
Devang Patelca640122006-08-23 16:59:25 +0000348 return LTO_WRITE_FAILURE;
Reid Spencere5c9cb52006-08-23 00:39:35 +0000349 }
Devang Patela89d47f2006-08-03 15:44:57 +0000350 sys::RemoveFileOnSignal(tmpAsmFilePath);
351
352 std::ofstream asmFile(tmpAsmFilePath.c_str(), io_mode);
353 if (!asmFile.is_open() || asmFile.bad()) {
354 if (tmpAsmFilePath.exists())
355 tmpAsmFilePath.eraseFromDisk();
356 return LTO_WRITE_FAILURE;
357 }
358
359 enum LTOStatus status = lto_optimize(bigOne, asmFile, exportList);
360 asmFile.close();
361 if (status != LTO_OPT_SUCCESS) {
362 tmpAsmFilePath.eraseFromDisk();
363 return status;
364 }
365
Devang Patel26810232006-09-06 00:28:22 +0000366 targetTriple = bigOne->getTargetTriple();
367
Devang Patela89d47f2006-08-03 15:44:57 +0000368 // Run GCC to assemble and link the program into native code.
369 //
370 // Note:
371 // We can't just assemble and link the file with the system assembler
372 // and linker because we don't know where to put the _start symbol.
373 // GCC mysteriously knows how to do it.
374 const sys::Path gcc = FindExecutable("gcc", "/");
375 if (gcc.isEmpty()) {
376 tmpAsmFilePath.eraseFromDisk();
377 return LTO_ASM_FAILURE;
378 }
379
380 std::vector<const char*> args;
381 args.push_back(gcc.c_str());
382 args.push_back("-c");
383 args.push_back("-x");
384 args.push_back("assembler");
385 args.push_back("-o");
386 args.push_back(OutputFilename.c_str());
387 args.push_back(tmpAsmFilePath.c_str());
388 args.push_back(0);
389
Reid Spencer023fcf92006-08-21 02:04:43 +0000390 sys::Program::ExecuteAndWait(gcc, &args[0], 0, 0, 1);
Devang Patela89d47f2006-08-03 15:44:57 +0000391
392 tmpAsmFilePath.eraseFromDisk();
393
394 return LTO_OPT_SUCCESS;
395}