StripSymbols.cpp revision 3a402b08cf903b02e1eabc6cef8a4186ddc87936 
1//===- StripSymbols.cpp - Strip symbols and debug info from a module ------===//
2//
3//                     The LLVM Compiler Infrastructure
4//
5// This file is distributed under the University of Illinois Open Source
6// License. See LICENSE.TXT for details.
7//
8//===----------------------------------------------------------------------===//
9//
10// The StripSymbols transformation implements code stripping. Specifically, it
11// can delete:
12//
13//   * names for virtual registers
14//   * symbols for internal globals and functions
15//   * debug information
16//
17// Note that this transformation makes code much less readable, so it should
18// only be used in situations where the 'strip' utility would be used, such as
19// reducing code size or making it harder to reverse engineer code.
20//
21//===----------------------------------------------------------------------===//
22
23#include "llvm/Transforms/IPO.h"
24#include "llvm/Constants.h"
25#include "llvm/DerivedTypes.h"
26#include "llvm/Instructions.h"
27#include "llvm/Module.h"
28#include "llvm/Pass.h"
29#include "llvm/ValueSymbolTable.h"
30#include "llvm/TypeSymbolTable.h"
31#include "llvm/Transforms/Utils/Local.h"
32#include "llvm/Support/Compiler.h"
33#include "llvm/ADT/SmallPtrSet.h"
34using namespace llvm;
35
36namespace {
37  class VISIBILITY_HIDDEN StripSymbols : public ModulePass {
38    bool OnlyDebugInfo;
39  public:
40    static char ID; // Pass identification, replacement for typeid
41    explicit StripSymbols(bool ODI = false)
42      : ModulePass(&ID), OnlyDebugInfo(ODI) {}
43
44    virtual bool runOnModule(Module &M);
45
46    virtual void getAnalysisUsage(AnalysisUsage &AU) const {
47      AU.setPreservesAll();
48    }
49  };
50
51  class VISIBILITY_HIDDEN StripNonDebugSymbols : public ModulePass {
52  public:
53    static char ID; // Pass identification, replacement for typeid
54    explicit StripNonDebugSymbols()
55      : ModulePass(&ID) {}
56
57    virtual bool runOnModule(Module &M);
58
59    virtual void getAnalysisUsage(AnalysisUsage &AU) const {
60      AU.setPreservesAll();
61    }
62  };
63
64  class VISIBILITY_HIDDEN StripDebugDeclare : public ModulePass {
65  public:
66    static char ID; // Pass identification, replacement for typeid
67    explicit StripDebugDeclare()
68      : ModulePass(&ID) {}
69
70    virtual bool runOnModule(Module &M);
71
72    virtual void getAnalysisUsage(AnalysisUsage &AU) const {
73      AU.setPreservesAll();
74    }
75  };
76}
77
78char StripSymbols::ID = 0;
79static RegisterPass<StripSymbols>
80X("strip", "Strip all symbols from a module");
81
82ModulePass *llvm::createStripSymbolsPass(bool OnlyDebugInfo) {
83  return new StripSymbols(OnlyDebugInfo);
84}
85
86char StripNonDebugSymbols::ID = 0;
87static RegisterPass<StripNonDebugSymbols>
88Y("strip-nondebug", "Strip all symbols, except dbg symbols, from a module");
89
90ModulePass *llvm::createStripNonDebugSymbolsPass() {
91  return new StripNonDebugSymbols();
92}
93
94char StripDebugDeclare::ID = 0;
95static RegisterPass<StripDebugDeclare>
96Z("strip-debug-declare", "Strip all llvm.dbg.declare intrinsics");
97
98ModulePass *llvm::createStripDebugDeclarePass() {
99  return new StripDebugDeclare();
100}
101
102/// OnlyUsedBy - Return true if V is only used by Usr.
103static bool OnlyUsedBy(Value *V, Value *Usr) {
104  for(Value::use_iterator I = V->use_begin(), E = V->use_end(); I != E; ++I) {
105    User *U = *I;
106    if (U != Usr)
107      return false;
108  }
109  return true;
110}
111
112static void RemoveDeadConstant(Constant *C) {
113  assert(C->use_empty() && "Constant is not dead!");
114  SmallPtrSet<Constant *, 4> Operands;
115  for (unsigned i = 0, e = C->getNumOperands(); i != e; ++i)
116    if (isa<DerivedType>(C->getOperand(i)->getType()) &&
117        OnlyUsedBy(C->getOperand(i), C))
118      Operands.insert(C->getOperand(i));
119  if (GlobalVariable *GV = dyn_cast<GlobalVariable>(C)) {
120    if (!GV->hasLocalLinkage()) return;   // Don't delete non static globals.
121    GV->eraseFromParent();
122  }
123  else if (!isa<Function>(C))
124    if (isa<CompositeType>(C->getType()))
125      C->destroyConstant();
126
127  // If the constant referenced anything, see if we can delete it as well.
128  for (SmallPtrSet<Constant *, 4>::iterator OI = Operands.begin(),
129         OE = Operands.end(); OI != OE; ++OI)
130    RemoveDeadConstant(*OI);
131}
132
133// Strip the symbol table of its names.
134//
135static void StripSymtab(ValueSymbolTable &ST, bool PreserveDbgInfo) {
136  for (ValueSymbolTable::iterator VI = ST.begin(), VE = ST.end(); VI != VE; ) {
137    Value *V = VI->getValue();
138    ++VI;
139    if (!isa<GlobalValue>(V) || cast<GlobalValue>(V)->hasLocalLinkage()) {
140      if (!PreserveDbgInfo || strncmp(V->getNameStart(), "llvm.dbg", 8))
141        // Set name to "", removing from symbol table!
142        V->setName("");
143    }
144  }
145}
146
147// Strip the symbol table of its names.
148static void StripTypeSymtab(TypeSymbolTable &ST, bool PreserveDbgInfo) {
149  for (TypeSymbolTable::iterator TI = ST.begin(), E = ST.end(); TI != E; ) {
150    if (PreserveDbgInfo && strncmp(TI->first.c_str(), "llvm.dbg", 8) == 0)
151      ++TI;
152    else
153      ST.remove(TI++);
154  }
155}
156
157/// Find values that are marked as llvm.used.
158void findUsedValues(Module &M,
159                    SmallPtrSet<const GlobalValue*, 8>& llvmUsedValues) {
160  if (GlobalVariable *LLVMUsed = M.getGlobalVariable("llvm.used")) {
161    llvmUsedValues.insert(LLVMUsed);
162    // Collect values that are preserved as per explicit request.
163    // llvm.used is used to list these values.
164    if (ConstantArray *Inits =
165        dyn_cast<ConstantArray>(LLVMUsed->getInitializer())) {
166      for (unsigned i = 0, e = Inits->getNumOperands(); i != e; ++i) {
167        if (GlobalValue *GV = dyn_cast<GlobalValue>(Inits->getOperand(i)))
168          llvmUsedValues.insert(GV);
169        else if (ConstantExpr *CE =
170                 dyn_cast<ConstantExpr>(Inits->getOperand(i)))
171          if (CE->getOpcode() == Instruction::BitCast)
172            if (GlobalValue *GV = dyn_cast<GlobalValue>(CE->getOperand(0)))
173              llvmUsedValues.insert(GV);
174      }
175    }
176  }
177}
178
179/// StripSymbolNames - Strip symbol names.
180bool StripSymbolNames(Module &M, bool PreserveDbgInfo) {
181
182  SmallPtrSet<const GlobalValue*, 8> llvmUsedValues;
183  findUsedValues(M, llvmUsedValues);
184
185  for (Module::global_iterator I = M.global_begin(), E = M.global_end();
186       I != E; ++I) {
187    if (I->hasLocalLinkage() && llvmUsedValues.count(I) == 0)
188      if (!PreserveDbgInfo || strncmp(I->getNameStart(), "llvm.dbg", 8))
189        I->setName("");     // Internal symbols can't participate in linkage
190  }
191
192  for (Module::iterator I = M.begin(), E = M.end(); I != E; ++I) {
193    if (I->hasLocalLinkage() && llvmUsedValues.count(I) == 0)
194      if (!PreserveDbgInfo || strncmp(I->getNameStart(), "llvm.dbg", 8))
195        I->setName("");     // Internal symbols can't participate in linkage
196    StripSymtab(I->getValueSymbolTable(), PreserveDbgInfo);
197  }
198
199  // Remove all names from types.
200  StripTypeSymtab(M.getTypeSymbolTable(), PreserveDbgInfo);
201
202  return true;
203}
204
205// StripDebugInfo - Strip debug info in the module if it exists.
206// To do this, we remove llvm.dbg.func.start, llvm.dbg.stoppoint, and
207// llvm.dbg.region.end calls, and any globals they point to if now dead.
208bool StripDebugInfo(Module &M) {
209
210  SmallPtrSet<const GlobalValue*, 8> llvmUsedValues;
211  findUsedValues(M, llvmUsedValues);
212
213  // Delete all dbg variables.
214  for (Module::global_iterator I = M.global_begin(), E = M.global_end();
215       I != E; ++I) {
216    GlobalVariable *GV = dyn_cast<GlobalVariable>(I);
217    if (!GV) continue;
218    if (!GV->use_empty() && llvmUsedValues.count(I) == 0) {
219      if (strncmp(GV->getNameStart(), "llvm.dbg", 8) == 0) {
220        GV->replaceAllUsesWith(UndefValue::get(GV->getType()));
221      }
222    }
223  }
224
225  Function *FuncStart = M.getFunction("llvm.dbg.func.start");
226  Function *StopPoint = M.getFunction("llvm.dbg.stoppoint");
227  Function *RegionStart = M.getFunction("llvm.dbg.region.start");
228  Function *RegionEnd = M.getFunction("llvm.dbg.region.end");
229  Function *Declare = M.getFunction("llvm.dbg.declare");
230
231  std::vector<Constant*> DeadConstants;
232
233  // Remove all of the calls to the debugger intrinsics, and remove them from
234  // the module.
235  if (FuncStart) {
236    while (!FuncStart->use_empty()) {
237      CallInst *CI = cast<CallInst>(FuncStart->use_back());
238      Value *Arg = CI->getOperand(1);
239      assert(CI->use_empty() && "llvm.dbg intrinsic should have void result");
240      CI->eraseFromParent();
241      if (Arg->use_empty())
242        if (Constant *C = dyn_cast<Constant>(Arg))
243          DeadConstants.push_back(C);
244    }
245    FuncStart->eraseFromParent();
246  }
247  if (StopPoint) {
248    while (!StopPoint->use_empty()) {
249      CallInst *CI = cast<CallInst>(StopPoint->use_back());
250      Value *Arg = CI->getOperand(3);
251      assert(CI->use_empty() && "llvm.dbg intrinsic should have void result");
252      CI->eraseFromParent();
253      if (Arg->use_empty())
254        if (Constant *C = dyn_cast<Constant>(Arg))
255          DeadConstants.push_back(C);
256    }
257    StopPoint->eraseFromParent();
258  }
259  if (RegionStart) {
260    while (!RegionStart->use_empty()) {
261      CallInst *CI = cast<CallInst>(RegionStart->use_back());
262      Value *Arg = CI->getOperand(1);
263      assert(CI->use_empty() && "llvm.dbg intrinsic should have void result");
264      CI->eraseFromParent();
265      if (Arg->use_empty())
266        if (Constant *C = dyn_cast<Constant>(Arg))
267          DeadConstants.push_back(C);
268    }
269    RegionStart->eraseFromParent();
270  }
271  if (RegionEnd) {
272    while (!RegionEnd->use_empty()) {
273      CallInst *CI = cast<CallInst>(RegionEnd->use_back());
274      Value *Arg = CI->getOperand(1);
275      assert(CI->use_empty() && "llvm.dbg intrinsic should have void result");
276      CI->eraseFromParent();
277      if (Arg->use_empty())
278        if (Constant *C = dyn_cast<Constant>(Arg))
279          DeadConstants.push_back(C);
280    }
281    RegionEnd->eraseFromParent();
282  }
283  if (Declare) {
284    while (!Declare->use_empty()) {
285      CallInst *CI = cast<CallInst>(Declare->use_back());
286      Value *Arg1 = CI->getOperand(1);
287      Value *Arg2 = CI->getOperand(2);
288      assert(CI->use_empty() && "llvm.dbg intrinsic should have void result");
289      CI->eraseFromParent();
290      if (Arg1->use_empty()) {
291        if (Constant *C = dyn_cast<Constant>(Arg1))
292          DeadConstants.push_back(C);
293        else
294          RecursivelyDeleteTriviallyDeadInstructions(Arg1, NULL);
295      }
296      if (Arg2->use_empty())
297        if (Constant *C = dyn_cast<Constant>(Arg2))
298          DeadConstants.push_back(C);
299    }
300    Declare->eraseFromParent();
301  }
302
303  // llvm.dbg.compile_units and llvm.dbg.subprograms are marked as linkonce
304  // but since we are removing all debug information, make them internal now.
305  // FIXME: Use private linkage maybe?
306  if (Constant *C = M.getNamedGlobal("llvm.dbg.compile_units"))
307    if (GlobalVariable *GV = dyn_cast<GlobalVariable>(C))
308      GV->setLinkage(GlobalValue::InternalLinkage);
309
310  if (Constant *C = M.getNamedGlobal("llvm.dbg.subprograms"))
311    if (GlobalVariable *GV = dyn_cast<GlobalVariable>(C))
312      GV->setLinkage(GlobalValue::InternalLinkage);
313
314  if (Constant *C = M.getNamedGlobal("llvm.dbg.global_variables"))
315    if (GlobalVariable *GV = dyn_cast<GlobalVariable>(C))
316      GV->setLinkage(GlobalValue::InternalLinkage);
317
318  // Delete all dbg variables.
319  for (Module::global_iterator I = M.global_begin(), E = M.global_end();
320       I != E; ++I) {
321    GlobalVariable *GV = dyn_cast<GlobalVariable>(I);
322    if (!GV) continue;
323    if (GV->use_empty() && llvmUsedValues.count(I) == 0
324        && (!GV->hasSection()
325            || strcmp(GV->getSection().c_str(), "llvm.metadata") == 0))
326      DeadConstants.push_back(GV);
327  }
328
329  if (DeadConstants.empty())
330    return false;
331
332  // Delete any internal globals that were only used by the debugger intrinsics.
333  while (!DeadConstants.empty()) {
334    Constant *C = DeadConstants.back();
335    DeadConstants.pop_back();
336    if (GlobalVariable *GV = dyn_cast<GlobalVariable>(C)) {
337      if (GV->hasLocalLinkage())
338        RemoveDeadConstant(GV);
339    }
340    else
341      RemoveDeadConstant(C);
342  }
343
344  // Remove all llvm.dbg types.
345  TypeSymbolTable &ST = M.getTypeSymbolTable();
346  for (TypeSymbolTable::iterator TI = ST.begin(), TE = ST.end(); TI != TE; ) {
347    if (!strncmp(TI->first.c_str(), "llvm.dbg.", 9))
348      ST.remove(TI++);
349    else
350      ++TI;
351  }
352
353  return true;
354}
355
356bool StripSymbols::runOnModule(Module &M) {
357  bool Changed = false;
358  Changed |= StripDebugInfo(M);
359  if (!OnlyDebugInfo)
360    Changed |= StripSymbolNames(M, false);
361  return Changed;
362}
363
364bool StripNonDebugSymbols::runOnModule(Module &M) {
365  return StripSymbolNames(M, true);
366}
367
368bool StripDebugDeclare::runOnModule(Module &M) {
369
370  Function *Declare = M.getFunction("llvm.dbg.declare");
371
372  if (!Declare)
373    return false;
374
375  std::vector<Constant*> DeadConstants;
376
377  while (!Declare->use_empty()) {
378    CallInst *CI = cast<CallInst>(Declare->use_back());
379    Value *Arg1 = CI->getOperand(1);
380    Value *Arg2 = CI->getOperand(2);
381    assert(CI->use_empty() && "llvm.dbg intrinsic should have void result");
382    CI->eraseFromParent();
383    if (Arg1->use_empty()) {
384      if (Constant *C = dyn_cast<Constant>(Arg1))
385        DeadConstants.push_back(C);
386      else
387        RecursivelyDeleteTriviallyDeadInstructions(Arg1, NULL);
388    }
389    if (Arg2->use_empty())
390      if (Constant *C = dyn_cast<Constant>(Arg2))
391        DeadConstants.push_back(C);
392  }
393  Declare->eraseFromParent();
394
395  // Delete all llvm.dbg.global_variables.
396  for (Module::global_iterator I = M.global_begin(), E = M.global_end();
397       I != E; ++I) {
398    GlobalVariable *GV = dyn_cast<GlobalVariable>(I);
399    if (!GV) continue;
400    if (GV->use_empty() && GV->hasName()
401        && strncmp(GV->getNameStart(), "llvm.dbg.global_variable", 24) == 0)
402      DeadConstants.push_back(GV);
403  }
404
405  while (!DeadConstants.empty()) {
406    Constant *C = DeadConstants.back();
407    DeadConstants.pop_back();
408    if (GlobalVariable *GV = dyn_cast<GlobalVariable>(C)) {
409      if (GV->hasLocalLinkage())
410        RemoveDeadConstant(GV);
411    }
412    else
413      RemoveDeadConstant(C);
414  }
415
416  return true;
417}
418