StripSymbols.cpp revision ba58d9ca78372162be346a67d5d413c9353fe9e2 
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/Support/Compiler.h"
32#include "llvm/ADT/SmallPtrSet.h"
33using namespace llvm;
34
35namespace {
36  class VISIBILITY_HIDDEN StripSymbols : public ModulePass {
37    bool OnlyDebugInfo;
38  public:
39    static char ID; // Pass identification, replacement for typeid
40    explicit StripSymbols(bool ODI = false)
41      : ModulePass(&ID), OnlyDebugInfo(ODI) {}
42
43    /// StripSymbolNames - Strip symbol names.
44    bool StripSymbolNames(Module &M);
45
46    // StripDebugInfo - Strip debug info in the module if it exists.
47    // To do this, we remove llvm.dbg.func.start, llvm.dbg.stoppoint, and
48    // llvm.dbg.region.end calls, and any globals they point to if now dead.
49    bool StripDebugInfo(Module &M);
50
51    virtual bool runOnModule(Module &M);
52
53    virtual void getAnalysisUsage(AnalysisUsage &AU) const {
54      AU.setPreservesAll();
55    }
56  };
57}
58
59char StripSymbols::ID = 0;
60static RegisterPass<StripSymbols>
61X("strip", "Strip all symbols from a module");
62
63ModulePass *llvm::createStripSymbolsPass(bool OnlyDebugInfo) {
64  return new StripSymbols(OnlyDebugInfo);
65}
66
67/// OnlyUsedBy - Return true if V is only used by Usr.
68static bool OnlyUsedBy(Value *V, Value *Usr) {
69  for(Value::use_iterator I = V->use_begin(), E = V->use_end(); I != E; ++I) {
70    User *U = *I;
71    if (U != Usr)
72      return false;
73  }
74  return true;
75}
76
77static void RemoveDeadConstant(Constant *C) {
78  assert(C->use_empty() && "Constant is not dead!");
79  SmallPtrSet<Constant *, 4> Operands;
80  for (unsigned i = 0, e = C->getNumOperands(); i != e; ++i)
81    if (isa<DerivedType>(C->getOperand(i)->getType()) &&
82        OnlyUsedBy(C->getOperand(i), C))
83      Operands.insert(C->getOperand(i));
84  if (GlobalVariable *GV = dyn_cast<GlobalVariable>(C)) {
85    if (!GV->hasInternalLinkage()) return;   // Don't delete non static globals.
86    GV->eraseFromParent();
87  }
88  else if (!isa<Function>(C))
89    C->destroyConstant();
90
91  // If the constant referenced anything, see if we can delete it as well.
92  for (SmallPtrSet<Constant *, 4>::iterator OI = Operands.begin(),
93         OE = Operands.end(); OI != OE; ++OI)
94    RemoveDeadConstant(*OI);
95}
96
97// Strip the symbol table of its names.
98//
99static void StripSymtab(ValueSymbolTable &ST) {
100  for (ValueSymbolTable::iterator VI = ST.begin(), VE = ST.end(); VI != VE; ) {
101    Value *V = VI->getValue();
102    ++VI;
103    if (!isa<GlobalValue>(V) || cast<GlobalValue>(V)->hasInternalLinkage()) {
104      // Set name to "", removing from symbol table!
105      V->setName("");
106    }
107  }
108}
109
110bool StripSymbols::runOnModule(Module &M) {
111  bool Changed = false;
112  Changed |= StripDebugInfo(M);
113  Changed |= StripSymbolNames(M);
114  return Changed;
115}
116
117
118// Strip the symbol table of its names.
119static void StripTypeSymtab(TypeSymbolTable &ST) {
120  for (TypeSymbolTable::iterator TI = ST.begin(), E = ST.end(); TI != E; )
121    ST.remove(TI++);
122}
123
124/// StripSymbolNames - Strip symbol names.
125bool StripSymbols::StripSymbolNames(Module &M) {
126
127  if (OnlyDebugInfo)
128    return false;
129
130  SmallPtrSet<const GlobalValue*, 8> llvmUsedValues;
131  if (GlobalVariable *LLVMUsed = M.getGlobalVariable("llvm.used")) {
132    llvmUsedValues.insert(LLVMUsed);
133    // Collect values that are preserved as per explicit request.
134    // llvm.used is used to list these values.
135    if (ConstantArray *Inits =
136        dyn_cast<ConstantArray>(LLVMUsed->getInitializer())) {
137      for (unsigned i = 0, e = Inits->getNumOperands(); i != e; ++i) {
138        if (GlobalValue *GV = dyn_cast<GlobalValue>(Inits->getOperand(i)))
139          llvmUsedValues.insert(GV);
140        else if (ConstantExpr *CE =
141                 dyn_cast<ConstantExpr>(Inits->getOperand(i)))
142          if (CE->getOpcode() == Instruction::BitCast)
143            if (GlobalValue *GV = dyn_cast<GlobalValue>(CE->getOperand(0)))
144              llvmUsedValues.insert(GV);
145      }
146    }
147  }
148
149  for (Module::global_iterator I = M.global_begin(), E = M.global_end();
150       I != E; ++I) {
151    if (I->hasInternalLinkage() && llvmUsedValues.count(I) == 0)
152      I->setName("");     // Internal symbols can't participate in linkage
153  }
154
155  for (Module::iterator I = M.begin(), E = M.end(); I != E; ++I) {
156    if (I->hasInternalLinkage() && llvmUsedValues.count(I) == 0)
157      I->setName("");     // Internal symbols can't participate in linkage
158    StripSymtab(I->getValueSymbolTable());
159  }
160
161  // Remove all names from types.
162  StripTypeSymtab(M.getTypeSymbolTable());
163
164  return true;
165}
166
167// StripDebugInfo - Strip debug info in the module if it exists.
168// To do this, we remove llvm.dbg.func.start, llvm.dbg.stoppoint, and
169// llvm.dbg.region.end calls, and any globals they point to if now dead.
170bool StripSymbols::StripDebugInfo(Module &M) {
171
172  Function *FuncStart = M.getFunction("llvm.dbg.func.start");
173  Function *StopPoint = M.getFunction("llvm.dbg.stoppoint");
174  Function *RegionStart = M.getFunction("llvm.dbg.region.start");
175  Function *RegionEnd = M.getFunction("llvm.dbg.region.end");
176  Function *Declare = M.getFunction("llvm.dbg.declare");
177
178  std::vector<GlobalVariable*> DeadGlobals;
179
180  // Remove all of the calls to the debugger intrinsics, and remove them from
181  // the module.
182  if (FuncStart) {
183    while (!FuncStart->use_empty()) {
184      CallInst *CI = cast<CallInst>(FuncStart->use_back());
185      Value *Arg = CI->getOperand(1);
186      assert(CI->use_empty() && "llvm.dbg intrinsic should have void result");
187      CI->eraseFromParent();
188      if (Arg->use_empty())
189        if (GlobalVariable *GV = dyn_cast<GlobalVariable>(Arg))
190          DeadGlobals.push_back(GV);
191    }
192    FuncStart->eraseFromParent();
193  }
194  if (StopPoint) {
195    while (!StopPoint->use_empty()) {
196      CallInst *CI = cast<CallInst>(StopPoint->use_back());
197      Value *Arg = CI->getOperand(3);
198      assert(CI->use_empty() && "llvm.dbg intrinsic should have void result");
199      CI->eraseFromParent();
200      if (Arg->use_empty())
201        if (GlobalVariable *GV = dyn_cast<GlobalVariable>(Arg))
202          DeadGlobals.push_back(GV);
203    }
204    StopPoint->eraseFromParent();
205  }
206  if (RegionStart) {
207    while (!RegionStart->use_empty()) {
208      CallInst *CI = cast<CallInst>(RegionStart->use_back());
209      Value *Arg = CI->getOperand(1);
210      assert(CI->use_empty() && "llvm.dbg intrinsic should have void result");
211      CI->eraseFromParent();
212      if (Arg->use_empty())
213        if (GlobalVariable *GV = dyn_cast<GlobalVariable>(Arg))
214          DeadGlobals.push_back(GV);
215    }
216    RegionStart->eraseFromParent();
217  }
218  if (RegionEnd) {
219    while (!RegionEnd->use_empty()) {
220      CallInst *CI = cast<CallInst>(RegionEnd->use_back());
221      Value *Arg = CI->getOperand(1);
222      assert(CI->use_empty() && "llvm.dbg intrinsic should have void result");
223      CI->eraseFromParent();
224      if (Arg->use_empty())
225        if (GlobalVariable *GV = dyn_cast<GlobalVariable>(Arg))
226          DeadGlobals.push_back(GV);
227    }
228    RegionEnd->eraseFromParent();
229  }
230  if (Declare) {
231    while (!Declare->use_empty()) {
232      CallInst *CI = cast<CallInst>(Declare->use_back());
233      Value *Arg = CI->getOperand(2);
234      assert(CI->use_empty() && "llvm.dbg intrinsic should have void result");
235      CI->eraseFromParent();
236      if (Arg->use_empty())
237        if (GlobalVariable *GV = dyn_cast<GlobalVariable>(Arg))
238          DeadGlobals.push_back(GV);
239    }
240    Declare->eraseFromParent();
241  }
242
243  // llvm.dbg.compile_units and llvm.dbg.subprograms are marked as linkonce
244  // but since we are removing all debug information, make them internal now.
245  if (Constant *C = M.getNamedGlobal("llvm.dbg.compile_units"))
246    if (GlobalVariable *GV = dyn_cast<GlobalVariable>(C))
247      GV->setLinkage(GlobalValue::InternalLinkage);
248
249  if (Constant *C = M.getNamedGlobal("llvm.dbg.subprograms"))
250    if (GlobalVariable *GV = dyn_cast<GlobalVariable>(C))
251      GV->setLinkage(GlobalValue::InternalLinkage);
252
253  // Delete all dbg variables.
254  const Type *DbgVTy = M.getTypeByName("llvm.dbg.variable.type");
255  const Type *DbgGVTy = M.getTypeByName("llvm.dbg.global_variable.type");
256  if (DbgVTy || DbgGVTy)
257    for (Module::global_iterator I = M.global_begin(), E = M.global_end();
258         I != E; ++I)
259      if (GlobalVariable *GV = dyn_cast<GlobalVariable>(I))
260        if (GV->hasName() && GV->use_empty()
261            && !strncmp(GV->getNameStart(), "llvm.dbg", 8)
262            && (GV->getType()->getElementType() == DbgVTy
263                || GV->getType()->getElementType() == DbgGVTy))
264          DeadGlobals.push_back(GV);
265
266  if (DeadGlobals.empty())
267    return false;
268
269  // Delete any internal globals that were only used by the debugger intrinsics.
270  while (!DeadGlobals.empty()) {
271    GlobalVariable *GV = DeadGlobals.back();
272    DeadGlobals.pop_back();
273    if (GV->hasInternalLinkage())
274      RemoveDeadConstant(GV);
275  }
276
277  // Remove all llvm.dbg types.
278  TypeSymbolTable &ST = M.getTypeSymbolTable();
279  for (TypeSymbolTable::iterator TI = ST.begin(), TE = ST.end(); TI != TE; ) {
280    if (!strncmp(TI->first.c_str(), "llvm.dbg.", 9))
281      ST.remove(TI++);
282    else
283      ++TI;
284  }
285
286  return true;
287}
288